INATIONAL CENTER Series 2For HEALTH STATISTICS Number 19

VITAL and HEALTH STATISTICS

DATA EVALUATION AND METHODS RESEARCH

Participation of Hospitalsin the Pilot Study of theHospital Discharge Survey

Evaluation of the wi Ilingness of short-stay hospitals to partici-pate in a continuing patient-oriented survey.

DHEW Publication No. I(HRA) 74-1285

U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE

Public Health Service

Health Services and Mental Health Administration

National Center for Health Statistics

Rockville, Maryland

..

Vital and Health Statistics-Series 2-No. 19First issued in the Public Health k-vice Publication Series No. 1000- october 1966

NATIONAL CENTER FOR HEALTH STATISTICS

EDWARD B. PERRIN, Ph. D., Acting Director

PHILIP S. LAWRENCE, SC.D., Deputy Director

GAIL F. FISHER, Assistant Directorfor Health Statistics Deve~opment

WALT R. SIMMONS, M. A., Assistant Director for Research and Scientific Development

JOHN J. HANLON, M. D., Medical Advisor

JAMES E. KELLY, D. D. S., Dental Advisor

EDWARD E. MINTY, Execu tiue Officer

ALICE HAYWOOD, Information Officer

DHEW Publication No. (HRA) 74-1 285-Series 2-No. 19

Library of Congress Card Number 66-62090

PREFACE

The National Health Survey collects data on the healthof the population in three different ways: (1) by househoIdinterview techniques; (2) by direct examination; and (3) byabstracting information from existing records.

In connection with the latter method, a contract wasmade with the Graduate School of Public Health, Universityof Pittsburgh, to explore the feasibility of conducting anational patient-oriented survey in short-stay hospitals. Infiscal year 1965 the Hospital Discharge Survey was in-augurated with a large-scale pilot study by the NationalCenter for Health Statistics in cooperation with the Bureauof the Census and the University of Pittsburgh GraduateSchool of Public Health. As part of a contract with theCenter, the University of Pittsburgh has evaluated thewillingness of hospitals to participate in a continuingsurvey.

Milton C. Rossoff, Chief, Hospital Discharge SurveyBranch, Division of Health Records Statistics, edited thereport for publication.

CONTENTSPage

Preface --------------------------------------------------------------

Introduction ----------------------------------------------------------

The FeasibiliW Smdy --------------------------------------------------Introduction --------------------------------------------------------Hospitals Visited ---------------------------------------------------Procedure ---------------------------------------------------------Findtigs -----------------------------------------------------------Summary of Recommendations ---------------------------------------

Pilot Study Methodolo~------------------------------------------------Groundwork --------------------------------------------------------Initial Contact With the Hospital ---------------------------------------Personal Interviews ------------------------------------------------

Coopration ----------------------------------------------------------Acceptabilityof Data-Collection Procedures ---------------------------Payment ---------- -------- -------- --------------- ------- ------- ----

Interview With the Administrator ------------------- ------------- -----Interview Wititie,Metical Record Libraian --------------------------

Evaluation Program ---------------------------------------------------Procedure ---------------------------------------------------------Cooperation Obtained --------------------------- ----------- ---------

Recommendations --------------------------------------------------

Summary -------------------------------------------------------------

References -----------------------------------------------------------

Appendix I. FeasibilitySmdy -------------------------------------------Exhibit A. Letter to Hospital Administrator ----------------------------Exhibit B. Abstract Sheet --------------------------------------------

Appendix H. Pilot Study ---------------------------- --------- -------------Exhibit C. Conventional AbsmactForm --------------------------------Exhibit D. IBMAbstractForm ----------------------------------------Exhibit E.NCHS ''Fact Sheet''-------=--------------------------------Exhibit F.AHA Lefierto HospitaI Atiinistrator -----------------------ExhibitG. NCHS Letter to Hospital Administrator ----------------------Exhibit H. Statement on theSurvey ------------------------------------Exhibit I. Hospital Interview Form-- -------- -------- --------------- -.Exhibit J. MemorandumofAgreement ------------------- ---------------

i

1

113347

79

1516

1718202226

27272728

30

32

333334

383839414243444655

IN THIS REPOR T— which relates to a national continuing suwey ofpatients in sho~t-stay hospitals—findings, conclusions, and recommen-dations aye made concerning two studies.

A study conducted in fiscal yeay 1964 investigated the feasibility ofconducting a Hospital Discha?’ge Su?wey and concluded that such a suY-vey was practicable and would be welL ~eceived by the hospitals fallinginto the sample.

The Hospital DischaYge SuYvey commenced in fiscal yeav 1965 as apilot study with the intent of testing vayiows pvoceduyes in 84 hospitals,involving pevsonnel, fovms, and suitability of quality checks. Only 2 ofthe 84 hospitals which weye asked to pa~ticipate in the pilot stwdyfailed to do so. UnYeseYved coope~ation was obtained fvom 74 of thehospitals, and qualijted cooperation was Yeceived fyom the vemaining8 participating hospitals.

The appendixes contain many of the foyms used in the feasibility studyand in the pilot study.

SYMBOLS

I Data not available ------------------------ ---

I Category not applicable ------------------ . . .

Quantity zero ---------------------------- -

Quantity more than O but less than 0.05 ----- 0.0

Figure does not meet standards ofreliability or precision ------------------ *

PARTICIPATION OF HOSPITALS IN THE PILOT

STUDY OF THE HOSPITAL DISCHARGE SURVEY

The report was prepored by Ann M. Brown, A. B.; lsidore Altman, Ph. D., Professor of Medical Care Statistics; ond Dono -

van J. Thompson, Ph. D., Professor of Biometry, Department of Biostatistics, Graduate Schaol of Public Health, University

of Pittsburgh. The findings, conclusions, and recommendations are those of the authars.

INTRODUCTION

In the summer of 1964 the National Center forHealth Statistics of the U.S. Public Health Serviceinaugurated the Hospital Discharge Survey, acontinuing study designed to provide comprehen-sive general-purpose statistics on morbidity inpatients discharged from the Nation’s short-termhospitals. The authority for the survey is derivedfrom the National Health Survey Act of 1956(Public Law 652–84th Congress), which author-izes the Public Health Service to conduct contin-uing surveys on the health of the Nation’s peopleand to develop and test new methods for gatheringsuch information. 132The survey has the endorse-ment of the American Hospital Association.

The principal source of information for thesurvey is the medical record in the hospital,or more precisely, for the time being, the faceor summary sheet of the medical record. Thedata are to be obtained from probability samplesof medical records abstracted in a sample of theNation’s general hospitals. They will provide use-ful information on length of stay, diagnosis, andoperative procedures for the general populationclassified according to such demographic charac-

teristics as age, sex, color, and marital status.The Hospital Discharge Survey is one part

of a comprehensive National Health Survey 3 com-posed of three basic programs: the Health Inter-view Survey 4 the Health Examination Survey, 5>and the Health Records Survey. Ike Health Rec-ords Survey, which includes the Hospital Dis.charge Survey, is actually a family of surveysdesigned to gather and publish statistics fromfacilities which provide medical care; it isunder the direction of the Division of HealthRecords Statistics.

One of the early projects of this divisionwas the compilation of the most complete listingyex available of the Nation’s hospitals and resi-dent institutions providing medical, nursing, per-sonal, or custodial care. This Master FacilityInventory, Gas it is called, functions dually byserving as a sampling frame for the varioushealth records surveys andtistics on the numbers andthese establishments, Everykeep this inventory up to date.

THE FEASIBILITY STUDY

by providing sta-characteristics ofeffort is made to

INTRODUCTION vey. More than a year before the survey itselfwas initiated, and as part of this planning, the

A great deal of exploratory work and care- Graduate School of Public HeaIth of the Universityful planning must necessarily precede a national of Pittsburgh undertook a study, by contractundertaking such as the Hospital Discharge Sur- with the National Center for Health Statistics

1

of the Public Health Service, to examine thefeasibility of sampling hospital medical recordsin order to collect statistics on morbidity inhospitalized patients. A report on this study,

“Study of Feasibility of Sampling Hospital Dis-charge Records, ” has been made to the Division

of Health Records Statistics. This section is asummary of that report.

Forty-five purposively selected hospitalsthroughout the United States were visited to

obtain information about the following: the cir-cur’nstances under which hospitals would be will-ing to cooperate in a national discharge !survey;the questions they would ask; whose permissionwould be needed; the problems that would be en-countered in translating records of discharges into

Table 1. Distribution of hospitals visited in the 1963 feasibility study, by geographicregion, type of control, and bed capacity:l United States

Type of control and bed capacityg

All hospitals

All bed capacities ---------------------

Under 100 beds-------------------------------100-199 beds---------------------------------200-299 beds---------------------------------300 beds or more-----------------------------

Government

All bed capacities ---------------------

Under 100 beds -------------------------------100-199 beds---------------------------------200-299 beds---------------------------------300 beds or more-----------------------------

Nonprofit

All bed capacities---------------------

Under 100 beds-------------’ -----------------100-199 beds---------------------------------200-299 beds---------------------------------300 beds or more-----------------------------

Proprietary

All bed capacities---------------------

Under 100 beds-------------------------------100-199 beds---------------------------------200-299 beds---------------------------------300 beds or more-----------------------------

Allregions

45

1712

1:

9

52

Geographic region

North-east

NorthCentral

2

7

33

i

South

- 12—

8

:1

3

“1

11

5

32

4

4

West

10

1432

1

;22

3

12

lData relate to the year ending September 30, 1962.

2Exclusive of bassinets for newborn infants.

a sampling frame; the maximumwhich wouid be uniformly available

informationin all hos -

pitals; and the methods most appropriate forextracting the desired information from themedical records. Emphasis was on the problemsthat might arise in connection with securingcooperation.

The study of these 45 hospitals in 1963-64 will be referred to as the feasibility study;the activities to be described for the succeedingyear, 1964-65, will be referred to as the pilotstudy.

The feasibility study procedure involved, asa rule, separate interviews with the hospitaladministrators and their medical record librar-ians, the selection of a small sample of the med-ical records of discharged patients, and theabstracting of certain information from theserecords. The interview with the administratordealt primarily with elements of cooperation,whereas the medical record librarian was inter-viewed mainly about the record keeping systemin relation to establishing a procedure withinthe hospital for the regular, periodic collectionof information from a sample of the medicalrecords.

HOSPITALS VISITED

All but 2 of the 45 hospitals visited in the1963 feasibility study were general hospitals.The two exceptions were a small eye, ear, nose,and throat hospital and a short-term psychiatrichospital. we characteristics of the 45 hospitalsare presented in table 1. As may be seen fromthe table, nearly two-thirds of the hospitalswere nonprofit, and the remainder were nearlyevenly divided between government and propri-etary hospitals. Of the nine government hospitals,one was federally operated and the others wererun by State or local governments. Approxi-mately one-third of the nonprofit hospitals werechurch operated.

Most of the hospitals ranged in size from 30to 700 beds, but a few had well over 1,000 beds.In general, the government hospitals were thelargest and the proprietary hospitals were thesmallest. The feasibility study hospitals were

larger on the average than U.S. short-term hos-pitals, but the distribution of the study’s hospitalsby type of control was much like that of the Na-tion’s hospitals. A disproportionately large num-ber of Northeastern hospitals were included in thestudy. Because this particular study was not con-cerned with obtaining actuaI data but was a be-ginning study to learn about hospital procedures,the cooperation that might be anticipated, and thelike, a national probability sample was not feltto be necessary.

PROCEDURE

The administrators of the 45 hospitals to bevisited were first notified of the study by aletter from the Graduate School of Public Healthexplaining its purpose and procedure and re-questing cooperation in the study (exhibit A).About 1 week later the letter was followed by atelephone call in which the university repre-sentative requested an appointment to visit thehospital. Two administrators among the original45 who were contacted failed to cooperate evento this extent, and their hospitals were replacedby others in the same cities. Both “alternate”hospitals readily agreed to participate.

At nearly all of the hospitals visited, theadministrators and their medical record librar-ians were most cooperative and helpful. Theygave considered replies to the questions asked,and the medical record librarians often wentto considerable effort to locate records which theinterviewer requested. Unpleasant experienceswere surprisingly few, in view of the fact thatmany of the medical record departments gaveevidence of being overcrowded, understaffed, andburdened with work.

At only four hospitals was the universityrepresentative refused access to a sample ofmedical records once the interviews had takenplace. At least one of the refusals was basedon a question of legality; the administrator statedthat he had been advised by the local hospitalcouncil not to participate because of the possi-bility of Iaw suits. Another refusal probablystemmed more from embarrassment about thedisorganized state of the hospital’s medical rec-

orals than from concern about confidentiality.At a third hospital the refusal came from a

generally uncooperative medical record librar-ian who was interviewed before the administra-tor. Although the issue was not reopened, theinterviewer felt certain that the administratorwould have let him abstract records had theadministrator been talked to first. A small pro-prietary hospital was unwilling to let its recordsbe seen but said it would participate in the eventof a national survey.

FINDINGS

This section summarizes the 1963-64 feasi-bility study findings and presents the recommen-dations made at the conclusion of the study. Noattempt has been made to modify the recommen-dations, though modifications might have beenindicated by the 1964-65 pilot study experience,which is described in subsequent sections.

The feasibility study showed that for theelements it covered, a hospital discharge surveyof national scope is clearly practicable. Coop-eration of hospitals falling into the survey samplecan be anticipated, with very few exceptions,

provided the survey is tailored to the hospital’sproblems of personnel and space and to thehospital’s reasonable wishes about the mode ofabstracting information from the medical records.The majority of the administrators interviewedindicated that there would be no question abouttheir willingness to participate in a nationalhospital discharge survey. Some replied thatparticipation would depend on such practicalconsiderations as time, expense, and personnellimitations.

A few administrators expressed concernabout confidentiality and legality, but their fearsseemed for the most part to be allayed uponreassurance about the maintenance of confiden-tiality. The administrators were in general agree-ment that legal objections would be unlikelyif patients’ names were not abstracted. However,two administrators suggested that individual pa-tient authorization might be required before rec -orals could be looked at for statistical purposes,even though names were not to be copied. Thename, they said, would still be there for the

abstracter to see. On the other hand, the adminis -

trator of one large hospital explained that al-though the consent of the patient may be techni-cally required by law, it can usually be clmittedwhen a responsible agency is conducting theresearch. This would appear to be the generalfeeling among administrators.

Ready willingness to participate in a dis-charge survey seemed to be positively associatedwith hospital size; that is, the administrators ofthe larger hospitals seemed more receptive tothe survey than those of the smaller hospitals.One likely reason is that larger hospitals receivemore requests of this nature and are accustomedto research endeavors. The small hospitals,receiving fewer or no requests, would have littleexperience on which to base an immediate deci-sion. By type of control, the government hospitals,which were the largest visited, seemed to be themost willing to participate, and the proprietaryhospitals, which were the smallest visited, seemedto be the least willing.

The initial request for participation’ in thesurvey should be directed to the administrator,who may approve it himself or refer it to, some-

one else. At half the hospitals visited, the admin-istrator said that his approval would be suffi-cient, but at the other hospitals the administratorsaid he would have to seek the approval of atleast one other person or group of persons suchas the medical staff or the board of trustees.Some administrators said they would” clear therequest with a legal advisor even though no legalobjections were anticipated. Authority to approverequests of this nature was more likely to restsolely with the administrator in the larger hos-pitals, whereas consultation with a second partywas more likely to be required at the smallerhospitals.

The time required for obtaining cooperationwill depend partly on who is to make the decision.

If the authority is the administrator’s alone, aspeedier decision should result than if severalpeople have to be consulted. It seems likely thatwhen the board of trustees must be contacted,

more time will be required, particularly if therequest must be held until a regularly scheduledboard meeting.

Before any hospital is approached about thesurvey, the endorsement of the American HOS-

pital Association and similar organizations such

4

as the State hospital association should be obtainedand publicized, The local hospital council shouldalso be informed of the survey, perhaps by theAmerican Hospital Association, and encouragedto lend its support. As some administratorshinted, such support could mean the differencebetween cooperation and refusal.

It is also recommended that a strong state-ment on the value of the survey be sent to eachhospital selected and that a representative of theNational Center for Health Statistics visit eachhospital to explain the survey personally andsecure cooperation. Although there are no indi-cations that legal problems will present seriousdifficulty for the survey, it might be advisable toreview the various State laws and regulations(where such exist) governing the release ofmedical records for statistical purposes. Shoulda question be raised, the position of the personwho visits the hospital to explain the survey willbe stronger if he is informed about the relevantlaws and thei~ interpretation.

While some hospitals indicated they mightbe willing to collaborate in the survey withoutreimbursement, it was clear that, most wouldexpect some compensation for their contribution,especially where it was felt that overtime or theemployment of additional personnel might berequired. The tentative form shown to the ad-ministrator (exhibit B) was quite detailed andimplied a fairly exhaustive review of the medicalrecord. It contained questions on the character-istics of the patient and his hospitalization, in-cluding final diagnoses, operations, complica-tions, Iaborator y tests, therapies, and the like.

There were some differences about whetherpayment should be made to the hospital or to thepersonnel doing the work, but most administratorsfavored payment to the hospital, It is recom-mended that a uniform policy be adopted for thecompensation of hospitals and that fair payment,based on further examination of the true costto the hospitals, be made. Establishing a work-ing relation between the Federal governmentand the individual hospital is not a simple proc-ess, especially where transfer of funds is in-volved. Contracts will have to be made (at whichsome hospitals might conceivably balk), thestatus of the persons doing the abstracting de-fined. and so forth,

Administrative opinions as to who should dothe abstracting ranged from a strong preferencefor utilizing hospital persomel to an equallystrong preference for having someone sent into do the work. Any reasonable procedure forabstracting records that the National Center forHealth Statistics decided to adopt could probablybe made to work.

Generally the administrators indicated theywould be willing to have an employee of a surveyorganization come to the hospital periodically toselect a sample of discharges and abstract in-formation from records, provided the work wascarried out in an ethical manner by properlytrained personnel. In favor of having the samplingand abstracting done by representatives of theNational Center for Health Statistics rather thanby the medical record librarians is the fact thatrepresentatives’ training and performance couldbe closely controlled by the Center. Also, theabstracter’s first responsibility would be to thesurvey, rather than to the hospital. The surveywould naturally be of secondary importance tothe medical record librarian, who might notalways be able to direct his or her best effortstoward it. (However, greater training and ex-perience might more than compensate for anylack of enthusiasm or time.)

Contraindicating the use of abstracters withno medical record training would be their lackof familiarity with medical terminology; suchfamiliarity is a great aid in making out thehandwriting of the physicians. If the work is tobe done by persons without experience in themedical field, special instruction in medicalterminology might be included in the trainingprogram. Such instructions might also proveadvantageous even if hospital personnel do theabstracting, because not all hospitals have well-trained persons handling their records. It wasnot unusual to find a clerk responsible for therecords at smaller hospitals.

The majority of administrators said thathospital personnel could be made available forabstracting records; yet, the majority of themedical record librarians indicated they couldnot handle an added assignment without overtimeor additional help. Certainly many did not appearto welcome the idea of another assignment.

5

Aside from the obvious advantages of havingthe abstracting done by persons familiar withthe records and the doctors’ handwriting, theutilization of medical record personnel couldresult in improved medical records as the surveyprogresses, Because of training and experiencewith the particular hospital’s records, the medi-cal record librarian is in a far better positionthan an outsider would be to handle abstractingproblems when they arise and to encouragephysicians to complete records. If medical rec-ord personnel are to do the sampling and ab-stracting, a representative of the National Centerfor Health Statistics could instruct them initially

and then call periodically to review their work,which would be edited for completeness on a

continuing basis.A question as to whether records would be

abstracted on a daily basis, before the recordswere filed away, or at monthly intervals was

raised by a number of administrators and medi-cal record librarians. Those who mentionedthis point clearly favored the former alternative.

The monthly procedure, they said, meant thatcharts would have to be pulled and refiled andthe record room routine disrupted.

The importance of cordial relationships withthe administrator and, especially, with the medi-cal record librarian cannot be overstressed. The

success of the entire venture depends on thelatter’s cooperativeness, willingness to devotetime and effort, and motivation for completenessand accuracy. Good will is a sine qua non. To themaximum extent possible, the procedures for

sampling the records and abstracting data fromthem should be in line with the customs and prac-tices already being observed in the hospital andwith the wishes of the administrator and the

medical record librarian. Unless the decision ismade to follow a single procedure in all hos-pitals, it is recommended that the choice as towhether the abstracting be done by hospitalpersonnel or by an outsider, presumably well-trained, should be the hospital’s.

Some lag occurs between the patient’s date ofdischarge and the date the medical record is com-pleted. The amount of time an outside abstractershould wait before seeking to abstract records

from a particular month’s discharges was notdetermined, but obviously the time which is

optimal for one hospital may not be so foranother. Nevertheless, the experience indicatedthat 3 months would probably be an adequateperiod in most hospitals. It is recommendedthat the sample for month X (JanL12Wy, forexample) be selected in month X + 2 (March)and abstracted in month X + 3 (April). This ,procedure should allow the medical record librar-

ian sufficient time to locate and review the samplecharts for completeness. No matter how long aninterval is established or what kind of procedureis employed there will inevitably be some chartswhich cannot be located or made available forabstracting at the prescribed time. Under these

circumstances the medical record librarian mightbe asked to complete and mail abstracts for themissing records as soon as they become available,to avoid unnecessary y delay in processing themonthly data. An alternative would be to have thecharts put aside to be abstracted with the nextmonth’s records. Whatever policy is deemedadvisable, an instruction covering the situationshould be prepared.

The discharge rosters observed in the hos-

pitals visited consisted usually of simple straight-

forward lists of the discharges on a day-by-daybasis. The order in which discharged patients werelisted varied from hospital to hospital,, The fourprincipal ways of listing were by record number,alphabet,room number, and time of day. If a

simple systematic sample of every kth dischargeis to be drawn, this can be accomplished readilyin most hospitals. On the other hand, if a-ny within-hospital control on age, hospital service, or other

variable is contemplated, many hospitals may

have to be asked to develop a discharge rostertailored specifically to the needs of the survey.To insure the maintenance of the desired within-hospital sampling rate and overall control, a dis-charge roster specifically designed for the pur-poses of the Hospital Discharge Survey mayhave to be developed for at least some of anysample of hospitals. Small hospitals may not

have a discharge list, but a suitable form caneasily be designed for them.

The several items of information called foron the abstract form which was tested (exhibit B)were extracted from the hospitals’ medical rec-ords with varying degrees of success. Becauseof the importance of establishing mutually satis-

6

factory relationships with each hospital from thevery start, data collection should at first besim-ple. It is recommended that data collection belimited, at least initially, to information generallyavailable on the medical record face sheets—admission and discharge dates, date of birth(or age, if birth date is not available), color,sex, marital status, final diagnoses, and oper-ations. Expansion to items within the record,which present greater difficulties, can comeafter collection of data from the face sheet hasbecome systematized. Variations in record formsand recording practices for newborn infantssuggest that the newborn had best be excludedfrom the study, at least at first.

The medical record face sheets were foundto. be eminently suitable for fairly speedy ab-stracting, especially where the diagnoses andoperations were typed rather tian written in long-hand. Thought should be given to possibilitiesof “automating” the abstracting procedures to theextent practicable. Marked-sense or other de-vices might be tried in hospitals which arewilling to experiment in this fashion.

SUMMARY OF RECOMMENDATIONS

To summarize, the more important recom-mendations and suggestions coming out of thisfeasibility study were as folIows:

1. Endorsement by the American HospitalAssociation and State and local hospitalgroups would be influential in obtaininga hospital’s cooperation. Personal visitsby representatives of the National Centerfor Health Statistics to describe the Hos-pital Discharge Survey would be helpful,as would a statement about the uses ofthe survey.

PILOT STUDYTo some extent on the encouragement gained

from the feasibility study findings, which have justbeen described, the Division of Health RecordsStatistics of the National Center for Health Sta-tistics undertook a pilot study during the 1964-65 fiscal year to pretest procedures for theHospital Discharge Survey. In designing and

2.

3.

4.

5.

6.

7.

8.

Payment to the hospitals should be on auniform basis calculated ultimately ontrue cost to the hospitals.Contractual arrangements would seem tobe necessary to formalize procedures; thematter appeared to be generally accept-able. ”

It was found that a lag in time oftenoccurred between discharge of the pa-tient and completion of the medical rec-ord. This is an impm?ant factor to Wconsidered in establishing procedures forthe abstracting of medical records.

The procedures for coHection of datashould be simple and, at least in the earlystages of the Hospital Discharge Survey,limited to the summary or face sheet ofthe medical record. Newborn bifants mightat first be excluded.

Because of the fairly high degree ofuniformity noted in the style of the sum-mary sheet employed in hospitals and be-cause of their relative simplicity, tech-niques of mechanical processing shouldbe tried out.

The choice as to whether the abstractingis to be done by hospital or by govern-ment personnel should be the hospital’s,(Note: In the subsequent pilot study, whichis described next, the hospitals wereassigned one procedure or the other.Therefore, further trial of this recom-mendation could not be made.) “

FinaIly and most important, the feasi-bility study showed that a hospital dis-charge survey of national scope waspracticable and would be well receivedby the hospitals falling into the sample.

k

METHODOLOGYconducting the study, the National Center forHealth Statistics received assistance from boththe U.S. Bureau of the Census and the Departmentof Biostatistics, Graduate School of Public Health,University of Pittsburgh. The details of this studywill be reported elsewhere. The emphasis in thepresent report is on the cooperation and the man-

7

I ‘ea’~!!5-E---------.-----t2!!rY

i

Selection of sample and distributionof hospitals among participating

agencies

Letters from AHAand NCHS to sample

hospitals andhospital councils

+

Interview with Interview with- ~

1

,1First sampleadministrator MRL of records

and agreement reached selected

i

I Interview materialssent to NCHS I

+

Sampling —and

abstracting

Hospital contacted

I for evaluationprogram

Y 1 I

Evaluation of Evaluation Reproduction ofsampling frame of sampling face sheet

J

1 1.

Data processing* and 4

analysis

MFI - Master Facility Inventory NCHS - National Center for HealthAHA - American Hospital Association Statistics

MRL - Medical record librarian

Figure 1. Pilot study of the Hospital Discharge Survey.

8

ner of participation of hospitals in the pilot studyof the Hospital Discharge Survey.

The survey activities which were carriedout prior to and during the pilot study are sum-marized in figure 1, a flow chart depicting theinitiation of the Hospital Discharge Survey.

GROUNDWORK

Definition Qf Hospital

For the purpose of the study a hospital wasdefined in the following way:

1.

2.

3.

4.

5.

maintains six beds or more for inpa-tient care;

is licensed by the State in which it islocated (if the State has a licensure law);

provides inpatient care under the super-vision of a duly licensed doctor of medi-cine or doctor of osteopathy;

provides nursing service 24 hours a dayunder the supervision of a registerednurse;

maintains medical records for each pa-tient admitted.

The study was limited to “short-term” hos-pitals meeting all the above requirements, that is,to hospitals in which the average length of staywas less than 30 days.

Selection. of Sample

From the Master Facility Inventory a prob-ability sample of 95 hospitals was selected forthe 1964-65 pilot study. Of these hospitals, twowere judged to be out of scope on the basis ofthe ahove criteria—one because it providedservices for outpatients only, the other becauseit was a long-term hospital—and were conse-quently dropped frcm the study. ‘l%ese two hos-pitals, as well as the nine Veterms Adminis-tration Hospitals which fell into the sample,are not included in this report. The VeteransAdministration Hospitals are excluded becausethe details of their participation were not settledduring the pilot study year.

The distribution of the remaining 84 hos-pitals by geographic region, type of control, andbed capacity may be seen in tables 2 and 3.Although the sample was drawn from the MasterFacility Inventory, the data shown in the tablesare based on information provided by the hos-pitals on first contact. Since the Master FacilityInventory was by this time over 2 years old,some changes occurred in bed capacity and typeof control, and the figures here may differslightly from figures published elsewhere.

Nearly one-half the hospitals visited hadmore than 300 beds, and 20 percent had over1,000 beds. The reason for this high proportionof large hospitals is that the sample was madeup of two panels: (1) a certainty panel of all hos-pitals having 1,000 beds or more and (2) a prob-ability panel of hospitals with fewer than 1,000beds. As might be expected, the government-owned hospitals—State and local-were the larg-est, with nearly one-half of these falling intothe category of 1,000 beds or more, and theproprietary hospitals were the smallest, with noneof the four having more than 400 beds. The govern-ment hospitals comprised 33 percent of the pilotstudy hospitals. Church-owned hospitals account-ed for 23 percent, and other nonprofit hospitalsaccounted for 39 percent. The proprietary hos-pitals amounted to only 5 percent of the total.By geographic region, nearly one-third of the ●

sample hospitals were located in NortheasternStates; one-seventh were in the West. The re-maining 45 hospitals were evenly divided betweenthe North Central and Southern States. All but 7of the 84 hospitals were members of the AmericanHospital Association. The nonmembers were, forthe most part, small hospitals with fewer than100 beds.

Distribution Among Agencies

The responsibility for contacting hospitalsand bringing them into the Hospital DischargeSurvey fell largely to the Bureau of the Census,whose personnel setup was particularly wellsuited for such a study. llleir field personnel,located throughout the Nation, were alreadyexperienced in sampIe surveys in the healthfield. It is they who conduct the field work forthe Health Interview Survey.

9

Table 2. Distribution of pilot study hospitals in each geographic region, by selectedcharacteristics: Hospital Discharge Survey, 1964-65

Characteristic

Total----------------------------------

Bed capacityl

Under 50 beds--------------------------------50-99 beds-----------------------------------100-199 beds---------------------------------200-299 beds---------------------------------300-499 beds---------------------------------500-999 beds---------------------------------1,000 beds or more ---------------------------

Control

State and local government-------------------Church---------------------------------------Other nonprofit ------------------------------Proprietary ----------------------------------

AHA membership

Member---------------------------------------Nonmember ------------------------------------

Allregions

84

8

1;

;:1017

2819334

777

Geographic region

North-east

27

NorthCentral

22

23

:523

:8

202

South

23

West

12

213

21

5232

111

lExclusive of bassinets for newborn infants.

NOTE: AHA-American Hospital Association.

Table 3. Distribution of pilot study.hospitals, by type of control and bed caplacity:Hospital Discharge Survey, 1964-65

Type of control

Bed capacitylState Other

Total Pro-and local Church non-government profit prietary

All bed capacities--------------- 84 28 19 33 4

Under 50 beds-------------------------- 8 350-99 beds----------------------------- 2 i ; i100-199 beds--------------------------- 1? 5 1200-299 beds--------------------------- 11 : 3 ;300-499 beds--------------------------- 14 6 i500-999 beds--------------------------- ; :1,000 beds or more--------------------- :; 13 :

‘Exclusive of bassinets for newborn infants.

10

The 84 hospitals which comprised the pilotstudy sample were distributed among the threeparticipating agencies in the following way:

Hospi-Agency ta.k

Bureau of the Census --------------- 72Gradpate School of Public Health ----- 8National Center for Health Statistics-- 4

Methods of Data Collection

Several different data-collection procedures,or combinations of procedures, were developedfor the pilot study year and assigned to the hos-pitals by a quasi-random process. These are de-fined below. Figure 2 shows for each procedure(which will be designated A, B, C, D, and E) theagency responsible for each phase of the datacollection. The distribution of the sample hos-pitals by type of data-collection procedure offeredand selected characteristics is shown in table 4.In two instances, the procedure offered to thehospital was not the one originally set by theNational Center for Health Statistics. For thisreason, the figures in this report for “offered”procedure may differ slightly from figures pub-lished elsewhere on “assigned” procedure.

In procedures A and E, the sampling andabstracting were done by nonhospital personnel.In procedure A, representatives of the Bureauof the Census conducted the initial interview,sampled and abstracted the medical records,and edited the abstracts; in procedure E, theseoperations were handled by representatives of theNational Center for Health Statistics. As may beseen from figure 2, 43 hospitals were offeredprocedure A and 4 were offered procedure E.

In the protocol for procedure A, the datacollection was set up in such a way that theabstracter would visit the hospital once every3 months, select a ,sample of records to be ab-stracted on the next visit, and then abstractthe records which had fallen into the sample onthe preceding visit. The purpose of the intervalbetween sampling and abstracting was to allowtime for the records to be completed and madeavailable for abstracting. Procedure E was thesame as procedure A, except that the hospitals

were visited every other month instead of every3 months.

In procedure B, which is essentially a com-bination of procedures A and C (below), a rep-resentative of the Bureau of the Census selecteda sample of records each month but preparedonly the demographic portion of the abstractform. An employee of the hospital—the medicalrecord librarian or someone under her super-vision—completed the medical portion of the form,that is, the diagnostic and surgical “information.The Census representative conducted the initialinterviews, and the abstracts were sent to theBureau of the Census each month for editing.

In procedures C and D, the sampling and ab-stracting were done by hospital personnel. BothC and D called for monthly sample selection andcompletion of abstracts. The abstracts werethen sent to the Bureau of the Census for editingif a Census representative had made the initialcontact with the hospital (procedure C), orto the National Center for Health Statistics ifthe contact had been made by a representativeof the Graduate School of Public Health (pro-cedure D). Twenty-one hospitals were offeredprocedure C, and eight were offered procedure D.

Because procedure E, for our purposeshere, is essentially the same as procedure A,and procedure D the same as procedure C, weshall refer simply to procedures A, B, and C—as they affect and are affected by the matterof cooperation. Table 4 shows the distributionof the sample hospitals by both offered and finaIprocedure and by selected characteristics of thehospitals.

Abstract Forms

Two different abstract forms were developedfor the pilot study: (1) a conventional form inwhich the entries were written in (exhibit C), and(2) a form in which appropriate spaces wereblocked in for scanning by an IBM “reader”(exhibit D). The first form was assigned to 41hospitals and the second to 43. The two formswere fairly evenly distributed among the hos-pitals by size, type of control, geographic region,and assigned data-collection procedures. Theassignment of the abstract form to a particularhospital was random within paired groupings ofhospitals.

11

Survey operation

Interviewing ----------------------------

Sampling --------------------------------

Abstracting -----------------------------

Editing ---------------------------------

Number of hospitals ---------------------

Data-collection procedure

A B c D E

Agency

Censusl Census 1 Census 1 GSPH NCHS

Census Census Hospital .Hospital NCHS

CensusCensus and Hospital Hospital NCHS

hospital

Census Census Census NCHS NCHS

43 8 21 8 4

lAt a small number of hospitals the Census Bureau supervisor observed while a repre-sentative of the National Center for Health Statistics or the Graduate School of PublicHealth conducted the interviews.

NOTE:tistics.

Figure 2.

Sampling

GSPH-Graduate School of Public Health; NCHS—National Center for Health Sta-

Alternative procedures that were offered, defined in terms of survey operations and aqenciesinvolved.

Rate

The proportion of records sampled withineach hospital was designed to vary inverselywith the size interval of the hospital as given

in the Master Facility Inventory. Table 5 pre-

sents the sampling rates established for eachhospital-size group. The rates ranged from 1percent at the largest hospitals with 1,000 bedsor more to 20-40 percent in hospitals with fewerthan 50 beds. Bed capacity as reported by theadministrator would in a few instances have

placed the hospital in another size interval interms of the sampling rate. However, no adjust-ment in the original rate was made because each

hospital had fallen into the sample on the basisof its size as reported in the Master FacilityInventory.

Sampling Procedure

The methodof sampling the medical recordswas determined in each hospital by the format

of its discharge listing. If the listing containedthe patients’ medical record numbers, terminal-digit sampling could be used; otherwise, imervalsampling was required.

In terminal-digit sampling, the sample foraparticular month would consist of all patients

discharged that month whose medical record

12

Table 4. Distribution ofcollection procedures, by

pilot study hospitals with various offered and final data-selected characteristics: Hospital Discharge Survey, 1964-65

Characteristic

Total ------------------------------------

Bed capacity2

Under 50 beds ----------------------------------50-99 beds -------------------------------------100-199 beds -----------------------------------200-299 beds -----------------------------------300-499 beds -------- -------- --.----- -.------ ---500-999 beds..----------------------------------1,000 beds or more-----------------------------

Control

State and local government---------------------Church-----------------------------------------Other nonprofit--------------------------------Proprietary------------------------------------

Region

Northeast--------------------------------------North Central----------------------------------South----w-------------------------------------West--------------------------------------------

Offered procedure

Total

84

—

AorE

47

2777

1:

17

2;2

16101:

—

corD

29

Final procedural

Total

81

—

AorE

48

corD

30

lFigures for “final procedure” exclude 2 hospitals which failed to participate inthe pilot study and 1 hospital where the final procedure could not be classified as A,B, C, D, or E.

ProceduresA and E—Bureau of the Census or National Center for Health Statisticsselects the sample and abstracts all the required data.

Procedure B—Bureau of the Census selects the sample and abstracts the demographicdata; hospital abstracts the diagnostic and surgical information.

Procedures C and D—Hospital selects the sample anddata.

abstracts all the required

2Exclusive of bassinets for newborn infants.

numbers ended in a specified digit or pair ofdigits. For example, at one hospital where thesampling rate was four percent, an abstractwould be prepared for each discharged patientwhose medical record number ended, say, in24, 44, 74, or 94, thusyielding(inthelongrun)a sample of 4 in every 100 discharges.A sep-aratesetof’’samplekeynumbers”was assigned

to each hospital-size group in each of the fourgeographic regions.

Interval sampling ,which was employed whenthe patients’ medical record numbers were notshown in the discharge list, can perhaps bestbe explainedby an illustration.In a hospitalwhere the sampIing rate was 4 percent, the firstmonth’s sample might consist of the 24th entry

13

Table 5. Rates of sampling records with-in pilot study hospitals:l HospitalDischarge Survey, 1964-65

Bed capacity2 Sam- Number(Master Facility pling of

Inventory) rate hospitals

Per-cent

Under 50 beds:Northeast -----------North Central------- :: ;South--------------- 40 3West ---------------- 30 2

50-99 beds ------------ 14100-199 beds ---------- !;200-299 beds ---------- : 13300-499 beds ---------- 3 12500-999 beds ---------- 21,000 beds or more---- 1 1:

lRates were assigned on the basis ofbed capacity reported inthe Master Facil-ity Inventory and do not necessarily re-late to current bed capacity.

2Exclusive of bassinets for newborninfants.

in the discharge listing for that month (a randomstart) plus every 25th entrythereafter. Theappli-cation of the sampling interval 25 was contin-uous, so that the starting point (24) would beused only once, at the beginning of the survey.Thus, in interval sampling, a starting point wasassigned to each hospital-size group in eachof

the four geographic regions. That, incombinationwith the assigned interval, yielded the desiredrate of sampling for the hospital. Interval sam-pling is generally more tedious than terminal-

digit sampling and has an additional disadvantage

that the medical record numbers have to beobtained from the alphabetical patient index be-fore the sample records can be located in thenumerical file. (Interval sampling had to he

employed at only two hospitals.)

Other things being equal, a discharge listis preferable to an admission list in samplingbecause the sample selected has to relate tothose patients discharged during a specifiedperiod. The use of an admission list for sampling(to be employed only when no discharge list is

available) introduces some complications. TOobtain from an admission list a sample ofpatients discharged during a certain period itis necessary to begin the sample selectionsome time prior to the month under consider-ation, in order to be reasonably certain ofscreening all those eligible for discharge dur-ing the survey month. A sample selected in thisway, by either the terminal-digit or interval

method, will obviously contain patients dis-

charged before and after the month under con-

sideration. When admission lists are used, itis therefore necessary to obtain the dates ofdischarge for all the “sample” patients andex-clude or hold those not discharged during thesurvey month.

Not every patient who fell into the samplewas necessarily included .in the survey. Certainspecial classes of patients were considered tobe out of scope even though the hospitals mighthave regularly recorded one or more of theseclasses on the discharge or admission lists.Such out-of-scope cases included thefollowing:well-newborn infants, persons dead on arrival

at the hospital, stillbirths, and a few other kinds

of cases less frequently encountered. Abstractswere not completed for such out-of-scope cases.

Reimbursement Offer

Each hospital participating inthe surveywas

offered some payment, the amount or rate de-pending on who was to do the sampling andab-stracting of the medical records, Where hospital

personnel wereto handle all the work (procedureC), 30 cents per completed abstract was offered:Half that amount per abstract was offered to theB-procedure hospitals, where the hospital staffwere to complete the medical portion of theab-

stract’but were not to select the sample. Whereboth the sampling and abstracting were tobedoneby a representative of the National Center forHealth Statistics or the Bureau of the Census(procedure A), the hospital was offered a flatannual amount related to the expected annual

number of abstracts. These amounts varied from$10 per year for fewer than 100 abstracts

to $50 per year for 750 abstracts or more(table 6). Forallparticipating hospitals, paymentwas to be made at the end of each fiscal year.

14

Table 6. Schedule of payment for data-collection Procedures in Pilot study~; i:;ls:

LHospital Discha;ge Strrveyj

Procedure andestimated annual number Payment

of abstracts

I Per year

A and E1-99 abstracts ----------100-299 abstracts -------300-499 abstracts ------- 30500-749 abstracts ------- 40750 abstracts or more--- 50

I Per abstract

------- -.----- ------- ---- lsf$~ and D------------------- 30C

lProcedures A and E—Bureau of theCensus or National Center for Health Sta-tistics selects the sample and abstractsall the required data.

Procedure B—Bureau of the Censusselects the sample and abstracts the demo-graphic data;hospital abstracts the diag-nostic and surgical-information.

Procedures C and D—Hospital selectsthe sample and abstracta all the requireddata.

NCHS ‘Fact Sheetw

The National Center for Health Statisticsprepared an information sheet on each sa,mplehospital’, giving background information aboutthe hospita~—name, address, telephone number,bed capacity, type of control, and annual numberof discharges—as well as details pertaining tothat hospital’s participation in the survey, suchas the sampling rate and sample key digits,expected number of abstracts per year, andmethod of collecting the data (exhibit E). Thesewere valuable aids in preparing for the personalcontacts with administrators and medical recordlibrarians.

Special Pretest

Once the methodology for the pilot studyhad been developed, a special pretest was con-

ducted in 10 of the 84 sample hospitals to testthe procedures for inducting hospitals into thesurvey. A l-day conference for those involvedin the special pretest was held at the GraduateSchool of Public Health on September 22, 1964.Then letters were sent from the American Hos-pital Association andadministrators were calledon the telephone for appointments. The adminis-trators and their medical record librarianswere subsequently interviewedandinitialsamplesof medical records were selected. This method-ology will be described more fully in the sectionwhich follows. Here it is sufficient to saythat theproblems which were encountered in the specialpretest were remedied and the survey materialswere revised accordingly.

At the conclusion of the special pretest atraining program was held in Chicago on Novem-ber 5 and 6, 1964, for the Bureau ofthe Censusfield supervisory personnel who would be work-ing on the survey. The pilot study then con-tinued in the remaining 74 hospitals. Becausethe 10 hospitals which were visited during thespecial pretest were an integral part of the pilotstudy, no distinction will be made between themand thereport.

Letters

other 74 hospitals in the balance of this

INITIAL CONTACT

WITH THE HOSPITAL

The administrators of the 84 hospitals whichhad fallen into this first Hospital Discharge Surveysample were notified of the survey during Sep-tember, October, and November of 1964 by aletter from the Director of the American Hos-pital Association (exhibit F). This letter stressedthe importance of the study andrequested thehos-pital’s participation. It was followed, in about 1week, by a letter from the Director or the DeputyDirector of the National Center for Health Sta-tistics (exhibit G), in which the survey wasdescribed in more detail and cooperation wasagain requested. The letter from the NationalCenter for Health Statistics contained a two-page statement on the survey entitled “TheHospital Discharge Survey” (exhibit H). Thismay be seen in Appendix II, following the letterswhich were sent to the administrators.

15

Most of the hospitals which fell into thesurvey sample were sent both of the letters de-scribed above. However, in the special pretest,the letter from the National Center for HealthStatistics” was omitted because of time consider-ations. The statement on the survey which wouldhave been enclosed was given to the administratorat the start of the interview. The approach wasalso modified somewhat for the two sample hos-pitals which were not listed in the directory ofhospitals published by the American HospitalAssociation. These hospitals received only a letterfrom the National Center for Health Statistics.

The local hospital associations were informedof the survey and asked to lend their support. Theywere told the names of the hospitals in theirarea which had fallen into the survey sample,

Telephone Call

The letter from the National Center forHealth Statistics informed the administratorsthat a survey representative would telephonewithin a few days to make an appointment to

discuss the hospital’s participation in the survey.

Approximately half of the administrators werecalled within 2 weeks of the letter, and all exceptone were called within 4 weeks. With one ex-ception, the telephone contacts were completed

by the middle of December.

PERSONAL INTERVIEWS

Interview With the Administrator

AH of the administrators but two were inter-viewed by the end of December, generally within8 weeks of the letter from the American Hos-pital Association. Seventy-five percent of theadministrators were interviewed within 6 weeks

of this letter.The primary objective of this interview was

to establish a good relationship with the hospitaland to secure participation in the Hospitai Dis-

charge Survey under mutually satisfactory con-ditions. Another important purpose was to obtaina certain amount of information about the hos -pital, for example,

discharges. These

16

the current annual number of

objectives were served by

Section B of the interview form (exhibit I).When an agreement had been reached on

the manner of abstracting and the amount ofpayment, the details of the agreement wererecorded on a form designed for this purpose(exhibit J). This document was needed in orderto prepare a type of purchase order. (Althoughthe form carried the heading’ ‘Contract Proposal, ”the interviewers were informed that this couldbe modified to read “Memorandum of Agree-ment” or altered by merely striking out theword “Contract,” to read “Proposal.”)

Interview With the Medical Record

Librarian

By and large, the medical record librarianswere interviewed on the same day as the administ-

rator. At approximately half the hospitals, themedical record librarian was present for theinterview with the administrator, and she and theadministrator were interviewed together. At mostof the other hospitala the interview with themedical record librarian took place immedi-

ately following the interview with the adminis-trator. At only two hospitals was the surveyrepresentative not able to see the medical recordlibrarian, or her delegate, that day because ofsome question about the hospital’s participation

in the survey.The purpose of the interview with the medical

record librarian was to develop a procedure with-in the hospital for sampling and abstracting themedical records of discharged patients and toactually inaugurate that procedure. To this endthe medical record librarian was asked the seriesof questions in Sections C, D, E, and F of theinterview form (exhibit I), covering the com-position of the discharge list to be used in selecting

the sample and the procedures for sampling andabstracting.

Once a sample scheme had been devised theinterviewer selected the initial sample of dis-

charges and set a date for another meeting withthe medical record librarian. The agenda for this

second meeting depended largely upon ‘the ab-stracting procedure which had been establishedduring the first interview. Where the Bu[reau ofthe Census was to handle the mechanics- of the

survey (procedure A), the interviewer returnedto introduce an assistant who would at regularintervals visit the hospital to sample and ab-stract records. Where the work was to bedone byhospital personnel (procedure C), the interviewerreturned to assist the medical record librarianwith any problems she might have in connectionwith the survey. Most of the hospitals in thestudy had been visited a second time by the.end

of January. By the end of the second visit, thepattern for sampling and abstracting the medicalrecords was set.

In the spring of 1965 theparticipating hos-pitals were contacted in connection with the sur-vey’s evaluation program, which was essentiallya review of the sampling and abstracting whichhad been done. This program is described laterin this report.

COOPERATIONSome of the main purposes of the pilot

study were to obtain a further check on thedegree of cooperation that might be anticipatedby the Hospital Discharge Survey (HDS), to dis-cover the questions that might be raised, and todevelop techniques for dealing with them. Thissection describes these important aspects ofthe pilot study. It is based largely upon the re-actions of the interviewers recorded on the inter-view form (exhibit I) and submitted in responseto a subsequent request for more detail.

In a general way, it can safely be said thatthe results of the pilot study in terms of cooper-ation received from the sample hospitals werehighly satisfactory. As an indication, only 2 of the84 hospitals which were asked to participate in thepilot study failed to do so, and one of these latercame into the regular survey. In the opinion ofthe HDS representatives who conducted the initialinterviews, unreserved cooperation was obtainedfrom 74 of the hospitals which were asked toparticipate in the survey and qualified cooperationwas received from the remaining 8 participatinghospitals.

Obtaining Interviews

The interview with the administrator as anearly step in the pilot study procedure was de-scribed in the preceding chapter. With one ex-ception, the interviews with the administrators ofthe 84 hospitals were quite easily arranged bytelephone, Even the two administrators who with-held participation during the pilot study readilyagreed to an appointment to discuss the survey.Occasionally an administrator asked questionsabout the survey over the telephone, but generallythe telephone contacts were brief and limited to

the detail of establishing a mutually satisfactoryappointment time. Some of the administrators in-dicated on the telephone that their hospitals wouldparticipate in the survey, and at least two ad-ministrators notified the National Center forHealth Statistics of their willingness to cooperateeven before the telephone call had been made.

Approval of Request

At 73 of the 84 hospitals visited, the requestfor participation in the Hospital Discharge Surveywas approved by the administrator during theinitial interview. At 7 of the remaining 11 hos-pitals there was no question about cooperation.A signed memorandum of agreement was ulti-mately obtained from these hospitals as well asfrom two of the four hospitals where participationhad been questionable.

The reason usually given for the delay insigning the proposaI form was that the approvalor advice of someone other than the hospitaladministrator had to be sought. Where this wasthe case, the question was generally referredto a legal advisor, the medical director, theexecutive committee of the medical staff, theboard of trustees, or to the record room com-mittee. Usually this was a mere formality.Occasionally more than one such agent was con-sulted.

Nonparticipation

The two hospitals which failed to participateduring the pilot study year gave different reasonsfor holding back. One of the hospitals had re-cently had a time study conducted throughout thehospital, aimed at improving efficiency and re-

ducing expenses. As a result of the study, themedical record staff had been reduced for a trialperiod of 1 year. Because this action createdan additional workload for the remaining staff,

the administrator was unwilling to add to theirresponsibilities. In his opinion, 4 hours eachmonth would have been required to pull andrefile the medical records needed by the HDSabstractor,s, and he said he doubted that theultimate results would justify this time.

At the other hospital the reason given fornonparticipation was primarily a legal one. Thequestion of legality, raised by the medical direc-tor, was referred to the county attorney for anopinion in connection with the State statute onprivileged communication. The opinion, whichwas received too late to allow for participationin the pilot study, did not preclude participation,and the hospital is now taking part in the regularsurvey. Although a few other administrators mayhave raised this question with legal advisors,none was apparently advised against participation.It seems to be the general policy to approve asurvey of this nature when carried out by re-sponsible government agencies, particularly whencare is taken to insure the confidentiality of theinformation.

Two hospitals almost refused to cooperatein the study because of the attitude of some mem-bers of the medical staff. A few of the staffwere inclined to be somewhat suspicious of any

project sponsored by the Federal government.In addition, one hospital was currently partici-pating in a local study which some physiciansfelt to be a nuisance since it required them tocomplete their medical records sooner than theymight otherwise have been expected to. However,the hospital did agree to participate after discus-sions with its medical leaders.

ACCEPTABILITY OFDATA-COLLECTION PROCEDURES

The data-collection procedure offered bythe Hospital Discharge Survey was acceptedby 75 of the 82 hospitals which agreed to par-ticipate in the study. The changes in proceduremade at the other 7 hospitals are reflected intable 7, which shows the distribution of all 84

hospitals by both the offered procedure and thefinal procedure agreed upon.

Table 7. Distri_buti.on of pilot study hos-pitals, by offered and final data-collection procedures:1 Hospi,tal. Dis-charge Survey, 1964-65

I Offered procedure

Fins 1 procedureagreed upon

mF

All procedures--

HwA or E-------------- 48 44 3 1B--------.-- -------C or D------------- 3; i ? 2;Other procedures--- 1 1 .- -Nonparticipation--- 2 1 IL -

lprocedures A and E--Bureau of the Cen-sus or National Center for Health Sl:ati_a -tics selects the sample and absmacl:s allthe required data.

Procedure B—Bureau ‘of the Censusselects the sample and abstracts the demo-graphic data; hospital abstracts the di-agnostic and surgical information.

Procedures C and D—Hospital selectsthe sample and abstracts all the requireddata.

Procedure A

In procedure A the responsibility for datacollection rests largely with nonhospital per-sonnel. Where the medical record departmentis already overburdened with work this cannothelp but be a consideration. Several adminis-trators and medical record librarians wereplainly relieved to learn that the mechanics ofsampling and abstracting were to be ‘handledby persons connected with the survey. Of the47 hospitals which were offered procedure A,44 cooperated on that basis. Of the remainingthree hospitals, one failed to participate in thepilot study altogether, another was permittedtodo its own sampling and abstracting (procedureC), and the third hospital, after accepting pro-cedure A, suggested an alternate method of datacollection which was eventually employed. Thatmethod is described later in this chapter.

18

The hospital which switched to procedure Chad two objections to procedure A. The first con-cerned the confidential nature of the medical rec-ords. Although the administrator was willing toparticipate in the study to the extent of providingabstracts of the face sheet which did not containthe patient’s name, he was unwilling to have themedical records seen by nonhospital personnel.Even though the administrator agreed that divul-gence of information by an HDS abstracter wasextremely unlikely, he preferred to have theabstracts prepared by hospital employees. Thisadministrator believed further that the workcould be handled more efficiently by hospitalpersonnel, for the medical record departmentwould be able to absorb the abstracting in thedaily routine. He felt that the abstracts of thesample records should be prepared as soon asthe records were completed by the physiciansand before they were filed away. In this waythe need to retrieve records for survey repre-sentatives or to keep the records out of file forappreciable periods of time would be eliminated.

It should be noted here that these objectionsto procedure Awere also voiced by administratorswho nevertheless accepted the procedure. Someof those who agreed to procedure A would havepreferred to handle the survey themselves. l%eactual number who felt this way could not beascertained because administrators were notasked to state a preference and, in fact, wereencouraged to accept the procedure the inter-viewer offered.

An interesting modification in procedure Awas made at one of the hospitals to which itwas assigned. The administrator there preferredto have hospital personnel prepare the abstractsfor certain patients who were well-known to thegeneral public and for patients with illnesses whichcarry a stigma in the public mind. The adminis-trator himself wanted to review the listing ofsample cases in order to select those fewwhich should be abstracted by the hospital. Withthis modification he was quite willing to partici-pate in accordance with procedure A.

Procedure C

Only 1 of the 29 hospitals which were offeredprocedure C, whereby hospital personnel would be

responsible for the sampling and abstracting, wasnot willing to cooperate on this basis. When pro-cedure A was offered as an alternative, the hos-pital agreed to participate. Even though the other28 hospitals were willing to cooperate in accord-ance with this procedure, some administratorsand medical record librarians did voice objectionto the procedure because it placed the responsi-bility for sampling and abstracting on medicalrecord persomel who were already quite busy.On the other hand, to the extent that the adminis-trators had confidence in the integrity of theemployees who handled the medical records, theycould be assured that confidentiality would bemaintained.

In addition, under procedure C the samplingand abstracting could be incorporated in the dailyroutine, if the hospital so desired, and sample rec-ords would not have to be pulled for. survey rep-resentatives. During the interviews, at least fiveof the hospitals which agreed to do the work them-selves indicated that they would like to abstractthe sample records daily as such records werecompleted and before they were filed away. Prob-ably many of the remaining C-procedure hospitalsare also handling the abstracting in this way.

Procedure B

From table 7 it may be seen that procedureB, in which the abstracting was to be shared byhospital employees and survey personnel, was theleast favorably received by the sample hospitals.Only three of the eight hospitals which were offeredprocedure B were willing to cceperate underthese circumstances. The procedure was gener-ally regarded as inefficient by administrators whocould see no reason to have two persons ab-stracting from the same records, particularlyat different times. lhree of the four hospitalswhich rejected procedure B preferred to haveall the work done by Hospital Discharge Surveypersonnel, while the fourth hospital preferredto have its own employees handle the survey, sothat completed records would not have to remainunfiled or be pulled for outside abstracters. Oneof the hospitals which were offered procedure Bdid not participate in the pilot study—for otherreasons which have been noted.

19

Since procedure B was so poorly received bythe pilot study hospitals, it is recommended thatit not be among the procedures settled upon inlater stages of the survey. While procedure Bshould not be proposed ‘to any hospital, thissuggestion is not meant to preclude all arrange-ments whereby hospital employees and surveypersonnel would share the abstracting. For ex-ample, if the demographic data for the sampledischarges can be obtained directly from a com-puter tape and only the medical informationneeds to be taken from the medical record, adivision of labor might prove to be efficient.

Other Data-Collection Procedures

At one of the hospitals which had beenoffered procedure A, the medical record librar-

ian was able to suggest an alternate method ofdata collection which was accepted and followedat her hospital. When a record which fell intothe sample was completed, a Xerox copy of the

face sheet was made and sent to the NationalCenter for Health Statistics for abstracting. Thismethod relieved the hospital of the need to makesample records available to abstracters on pre-scribed dates and also avoided tying up the rec-ords within the hospital. Since this was a largeteaching hospital where the medical recordswere quite active, these were important con-siderations. For its participation on this basisthe hospital was paid 20 cents for each Xeroxcopy submitted. At the end of the pilot studyyear this procedure should be carefully evaluatedand a determination made as to its generalapplication. It is possible that ‘its use might be

indicated at other hospitals as well.At another A-procedure hospital where the

medical records were very active and not alwaysavailable at prescribed times, the medical record

librarian suggested that she make copies of thesample medical record face sheets and have themavailable for the HDS abstracter at the time of

his visit, Without further study it is not possibleto say whether the additional expense of re-producing the face sheets would be warrantedin this situation, but the suggestion deservesconsideration.

Conclusion

The experience in the pilot study indicatesthat either one of the basic data-collectionprocedures, A or C, would work at most hos-pitals. To summarize briefly, only 4 of the 76hospitals which were offered procedure A or Cdid not participate in accordance with the pro-posed procedure. (One such hospital did not par-

ticipate at all. ) On the other hand, four of theeight hospitals which were offered procedure Brejected it and an additional one failed to par-ticipate.

It seems clear from this experience that

hospitals will almost always accept the procedureoutlined to them by the Hospital Discharge Survey.Offering the hospital a choice of procedures maycreate good will, but if such a policy is deemedimpracticable, the large majority of hospitals

can be persuaded to accept a preassigned method.Any variations in the basic method suggested bythe hospital should be given consideration.

PAYMENT

As was previously stated, hospitals in whichnonhospital personnel did the sampling and ab-stracting (procedure A) were to be paid amountsranging from $10 to $50, depending on the numberof discharges; where hospital personnel per-formed this job (procedure C) the payment wasto be 30 cents per abstract. (In the few hospitalswhere the procedure was mixed, 15 cents was tobe paid.)

The amount of payment proposed by the Hos-pital Discharge Survey, in accordance with thedata-collection procedure, was acceptecl by 86

percent of the participating hospitals. Ten per-cent of the hospitals declined any payment, while4 percent sought an additional amount (table8). While the administrators who refused paymentwere not asked to give their reasons, somevoluntarily stated they felt they were performinga public service by participating in the /survey;others’ indicated that the amount involved did not ‘warrant the cost of setting up and maintaining aspecial account.

The 30 hospitals which were to do their ownsampling and abstracting seemed, for the most

20

Table 8. Distribution of pilot study hospitals, by acceptance of proposed payment andselected characteristics:l Hospital Discharge Survey, 1964-65

Characteristic

Total-----------------------------------------

Final procedure

Census or NCHS samples andabstracts all the data-----------------------------

Census samples and abstractsdemographic data; hospitalcompletes abstract---------------------------------

Hospital samples and abstractsall the data---------------------------------------

Bed capacitys

Under 50 beds---------------------------------------50-99 beds------------------------------------------100-199 beds--------------------------4-------------200-299 beds----------------------------------------300-499 beds----------------------------------------500-999 beds----------------------------------------1,000 beds or more----------------------------------

Control

State and local government --------------------------Church----------------------------------------------Other nonprofit -------------------------------------Proprietary-----------------------------------------

Region

Northeast ..-----------------------.------------------.Nyu[:C;tral --------------..-----------.------------

-----------------------------------------------West---J --------------------------------------------

Total

81

48

3

30

89

H

H15

27

;;4

26212212

Payment

Ac-cepted

70

39

3

28

;151010

1;

22

;;4

22

H9

f%ed

8

In- ~creased

3

lFigures exclude 2 hospitals which failed to participate in the pilot study and 1hospital where the final procedure could not be classified as A, B, C, D, or E.

21ncludes only those hospitals which requested an increase in payment. (Does not in-clude hospitals where payment was increased on the basis of a revised estimated numberof annual discharges.)

3Exclusive of bassinets for newborn infants.

part, satisfied with a payment of30 cents per to maintain cost records during the pilot studyabstract. One administrator in this grouprefused year. Where survey representatives were topayment; however another was reported asobvi- handle the sampling and abstracting (procedureously pleased that the hospital would be paid. A), the monetary offer, never more than $50One hospitaI felt that the amount given would per year, was received humorously by somenot cover thesurvey-related expensesandplanned administrators, especially where the annual pay-

21

ment was to be $10 or $20. The A-procedurehospitals accounted for seven of the eight hos-pitals which refused payment and for two of thethree hospitals where an increase was requested.

To evaluate the payment scale is beyond thescope of this report. The National Center forHealth Statistics will undoubtedly want to appraiseit carefully.

INTERVIEW WITH THE

ADMINISTRATOR

Although a structured interview was de-veloped for the pilot study (exhibit I), inter-viewers found it necessary or expedient to de-viate from a formal approach. The consensusamong the interviewers seemed to be that theinterview should begin with an informal discussionof the Hospital Discharge Survey—its purpose,general program, and details of participation—and then proceed to the matter of collectingstatistics about the hospital, which now is placedfirst. The start of the interview is generally not

a good time td solicit data. Once the adminis-trator has obtained a clear understanding of thesurvey and its requirements and it appears thatan agreement will be reached, the statistical datamay be obtained. The memorandum of agreementcan then be prepared at the conclusion of theinterview in accordance with the kind of agree-

ment ultimately reached.

It would be helpful to the interviewer, whenobtaining information about the hospital itself,to have the pertinent information from the “factsheet” also available in the interview form. Thiswould facilitate comparing the respondent’s infor-

mation with the corresponding information sup-plied by the National Center for Health Statistics.If a serious difference occurred, the interviewercould investigate further to ascertain whichsource of information was correct. Comparisonsof the two sets of data might also be facilitatedby presenting the “fact sheet” data in outlineform with the numbers of the outline correspond-ing to the item numbers in the questionnaire.

Administrators’ Questions and comments

The questions asked by the administratorsduring the interviews were of two general types-

straightforward questions seeking informationabout the survey, which usually required onlybrief replies, and questions which implied someobjection to the survey. The same questionmight be asked in different ways by differentadministrators, depending on their attitudes to-

ward the survey. For example, numerous aciminis-trators inquired about the purpose of the survey.Some were satisfied by a brief explanation statedin terms of the printed statement on the surveywhich had been sent to all the hospitals. Others

questioned the v,alue of the survey and wanted toknow more about its specific objectives.

The administrator of a large teaching insti-tution asked how he, as a hospital administrator,would be able to use the published data from thesurvey. In his opinion, combined utilization figuresfor a number of hospitals were meaningless,even if the hospitals were located in the same

geographical area, and were of no value to anadministrator in making policy decisions for one

hospital. Another administrator wanted to” knowwhy the objectives of the survey could not beaccomplished by using the statistics publishedby the American Hospital Association. Othersinquired about duplication of effort; quite a few

hospitals were already participating in somelocal project for which medical record abstractshad to be prepared.

Quite a few administrators asked how (or why)their hospitals had been selected for the survey.

Other administrators showed concern that theneeds of the survey would not be met by theirhospital’s participation; because their hospitalswere not “typical” short-term institutions, a

point a number of them made, they feared theoverall statistics might not be sound. A briefexplanation of the sampling theory underlyingthe selection of the hospitals, including the

reasons why one hospital cannot readily besubstituted for another, should be prepared forthe interviewers, who may need to react to suchstatements.

Because questions of this kind were en-countered frequently and were often difficult toanswer, it is suggested that the survey materials,particularly the letters and the enclosed state-

ment, be made more specific concerning ob-jectives of the survey and practical uses of thedata. The interviewers should be better pre-

22

pared to answer such questions than they werein the pilot study.

Occasionally an administrator asked whatother hospitals in the area were participatingin the survey. Since it has been the generalpolicy of the Bureau of the Census not to releasethe names of participants in sample surveys, theCensus interviewers generally tried to evadethis question. However, the local hospital coun-cils had been notified of the survey by the NationalCenter for Health Statistics and given the namesof the area hospitals selected for the survey.Some of the councils had in turn passed thisinformation on to the hospitals themselves. Theposition of the interviewer under these circum-stances was uncomfortable. The policy of thesurvey with respect to releasing participants’names should be well defined, and the inter-viewers should be informed of that policy.

“When is the survey to begin?” and “Howlong will it last?” were other questions fre-quently encountered. The answers were not pro-vided in the written statement on the survey.The answer to the latter question had, in fact,not yet been determined for individual hos-pitals at the time the interviews were held.Since the administrators deserve an answer tothis question, it is hoped that such details willbe available for the next set of interviews.

At Ieast one administrator gained the im-pression from the correspondence that his hos-pital’s participation was being requested for1 year only. He may have been led to believethis by a statement in the letter from the Ameri-can Hospital Association to the effect that thehospital was being asked to participate in the pre-testing phase of the study. The printed statementwhich was enclosed in the followup letter fromthe National Center for Health Statistics couldhave been interpreted to imply that the pilotstudy would end by July of 1965. It is unfortunatethat such misunderstandings occurred. Providingpertinent details in the letters and any otheradvance materials may help to prevent suchdifficulties in the future.

Quite a few administrators inquired about theamount of work the survey entailed. Some merelywanted to have the knowledge. But to others thequestion was a matter of real concern since theirmedical record personnel were already workingto near capacity.

Some administrators had not realized thatonly a sample of the medical records was to beabstracted. The letter from the American Hos-pital Association did not mention this point, andin the letter from the National Center for HealthStatistics the reference to the record sample wasrather general. (“The survey will collect infor-mation for a sample of discharges in a nationalsample of short-term hospitals.”) In the state-ment on the survey, “sample discharges” wasused twice, but this term was not defined.It was not until the second page of the statementthat one learned that there would be ‘‘. . . asample of a relatively small number of dischargeswithin each sample hospital. ” This is one of thebig “selling” points of the survey, and it cannotbe overemphasized. The administrator shouId beaware of this before the interview is even held.Consideration should be given to the possibilityof informing the administrator in one of theadvance letters of the actual sampling ratio forhis hospital.

The reference in the survey statement topossible future expansion of the survey dis-turbed some administrators. (“Several yearshence it is likely that additional informationabout the sample discharges available from thehospital records would be collected.”) One ad-ministrator inquired about the kinds of datawhich might later be collected the interviewerwas unable to provide the answer. Anotheradministrator stated that he was willing tocooperate with the study as it then stocd butmight not want to participate if it became moreinvolved. Still, a few administrators were alittle critical about the seeming paucity of infor-mation being collected initially.

Some administrators wanted to know if thesigned agreement was legally binding on thehospital, or if the hospital could drop out of thesurvey at a later date if it became too much ofa burden. The interviewers were told that theymight reply in the negative to the first partand hence affirmatively to the second part ofthe question. An instruction about the extentof obligation should be prepared for the manual;it should explain at the same time why thememorandum of agreement is necessary.

“How are the records to be sampled?” wasfrequently asked by both administrators and medi-cal record librarians. A simple explanation in-

23

volving the sample key digits usually sufficed.It is both desirable and necessary to quote thesample key digits when explaining the sampling

procedure. Since the digits are not now recordedin the interview form, space should be providedfor them at the beginning of the interviews withthe administrator and the medical record librar-ian and perhaps on the first page of the form, aswell, in the box where the hospital’s name andaddress are recorded (exhibit I). It is incon-venient and awkward for the interviewer to haveto refer to some other sheet when he is readyto explain the sampling process. If the samplingratio is stated in one of the letters to the admin-istrator, the sample key digits might be quotedas well.

A few administrators inquired how the amountof payment for participation had been determined.It seems to be quite important that the inter-viewers have some knowledge in this respect.Of course, the questions of confidentiality andlegality were also raised. These topics havebeen discussed previously.

Presence of Medical Record Librarian

The Hospital Discharge Survey had recom-mended that if possible the medical record librar-

ian be present at the interview with the ad-ministrator, but that suggestion had to be madetactfully inasmuch as it was entirely the ad-ministrator’s prerogative as to which membersof his staff should be present at the interview.About one-half the administrators chose to havetheir medical record librarians present from thefirst (table 9). It is likely that many felt that sheshould be there for the entire interview sincethe survey directly concerned her department.

Indeed, at one hospital the administrator askedthe medical record librarian to handle the inter-view and left to return only to sign the memoran-dum of agreement.

In 70 percent of those instances where themedical record librarian had been present duringthe interview with the administrator, the inter-viewer said that her presence had been helpful,and a number of interviewers recommended that

this become standard procedure. When the ad-ministrator and the medical record librarianare interviewed together, a second explanation

of the survey becomes unnecessary. Also, havingboth present facilitates reaching an agreementwhich is acceptable to all concerned. If the ad-

ministrator and the medical record librarian arenot in accord about some of the details, theirdifferences can be reconciled before an agree-

ment is signed. Because of her special Icnc)wledgeand insight about the medical record system in theparticular hospital, the medical record librarianmay be able to recommend an alternate data-collection procedure which is acceptable. Theultimate procedure is more likely to fit the hos-pital when the administrator and the medical rec-ord librarian are interviewed at the same time.

Another advantage to interviewing the ad-ministrator and the medical record librarian to-gether is that the medical record librarian isoften in a better position than the administratorto furnish the needed statistical information ondischarges, births, and the like. It is frequentlythe medical record department which compilesthese data. If the medical record librarian is

present, the questions can be directed to both, andwhoever has the information can supply iL

Although the interviewers at 10 hospitalsreplied that the presence of the medical recordlibrarian was not particularly helpful, for only 2of the 10 was it reported that her presence hadhad a detrimental effect. At one of these hos-pitals the medical record librarian objected tothe survey, and cooperation was obtained only byagreeing to furnish Census personnel to do thework. At the other hospital the medical record

librarian reportedly complicated the interviewwith the administrator by asking questionsirrelevant to her role in the survey.

The. potential advantages of interviewing theadministrator and the medical record librariansimultaneously probably far outweigh the poten-tial disadvantages, and it is recommended that

this become standard practice. It is also suggestedthat, as was done in the pilot study, after the sur-vey has been explained and an agreement reached,the interview with the medical record librarian

be continued in the medical record department.In that way the administrator’s time need not betaken up with the mechanics of sampling and ab-stracting which do not directly concern him. Thepropitious time for adjourning to the medical rec-ord department might be the point at which, in

24

Table 9. Distribution of pilot study hospitals, by presence and helpfulness of medicalrecord librarian (MRL) during interview with administrator and by selected character-istics:] Hospital Discharge Survey, 1964-65

Characteristic

Total---------------------------------

Final procedure2

A or %-------------------------------------B--------------------.----,------------------C or D--------------------------------------Other procedure-----------------------------Nonparticipation ----------------------------

Bed capacitys

Under 50 beds-------------------------------50-99 beds----------------------------------100-199 beds--------------------------------200-299 beds-----------,---------------------300-499 beds--------------------------------500-999 beds------------------------>-------

1,000 beds or more--------------------------

Control

State and local goverment ------------------church--------------------------------------Other nonprofit -----------------------------Proprietary ---------------------------------

Total

80

MRL present

Help-ful

Nothelp-ful

10

Undeter-mined

2

1

1

1

i

2

Munot

present

39

lFigures exclude 3 hospitals where the administrator was alsolibrarian and 1 hospital where the administrator

the medical recordasked the medical record librarian

to handle the induction interview.

‘Procedures A and E—Bureau of the Census or National Center for Health Statisticsselects the sample and abstracts all the required data.

Procedure B—Bureau of the Census selects the sample and abstracts the demographicdata; hospital abstracts the diagnostic and surgical information.

Procedures C and D—Hospital selects the sample and abstracts all the requireddata.

3Exclusive of bassinets for newborn infants.

the pilot-studyprocedure,theinterviewwiththe do soimpIies thatthehospitalwillparticipatein

medical record librarianis supposed to start. the survey,apresumption thattieadministratorThe medical record librarianshould not be may resent.

interviewedin advance of the administrator. To

25

INTERVIEW WITH THE MEDICAL

RECORD LIBRARIAN

Most of the medical record librarians werequite receptive to the survey, although some,naturally, had reservations about the amount ofwork involved. One medical record librarian whowas asked to prepare abstracts from the face

sheets of approximately 550 medical records ayear felt this amount would place too much of aburden on her limited staff. Other medical recordlibrarians were concerned about the amount oftime that would be required to locate and pull

the sample medical records. At large teaching

hospitals where the records are very active, thetime taken in tracking records down can be con-siderable. If the hospital requires an “out-card”to be prepared for each record removed fromfile, even more time is consumed in pulling eachchart.

The survey’s policy of requesting abstractsfrom participating hospitals as of July 1, 1964,regardless of the date an agreement was reached,drew complaints from some. of the medical rec-ord librarians who were to do their own abstract-ing. This request meant that if the medical recordlibrarian wished to incorporate the sampling andabstracting in her daily routine, the work wouldbe some months behind before it even began.

Such circumstances do not create good will withthe hospitals. To avoid this situati?n in the futureit is suggested that sampling, particularly for thehospitals that are to do the work themselves, be-gin with the discharges of the month following theone in which an agreement is ;eached. At mosthospitals this would be the month following the

month of initial interview. In that way the medi-cal record librarians could do the abstractingmonthly or daily, whichever they preferred.

The cooperation and interest of the medicalrecord librarians are reflected in their discerningquestions, their helpful suggestions, and theirwillingness to modify office procedures. For ex-ample, several medical record librarians who

normally did not prepare lists of dischargesagreed to maintain daily records for the HospitalDischarge Survey. Others who had lists of dis-charges without the medical record numbersoffered to begin adding the numbers as of the day

of interview. Where more difficult sampling prob-lems were presented, the medical record librar-ians went to even greater lengths to assist thesurvey representatives. It is true that in some

instances the medical record librarian tended tobe resistant and was difficult to interview, butthis is to be expected in any study which involvespersonal contact with a sizable number of indi-viduals.

The vast majority of interviewers seemed to

be satisfied with the questionnaire used in theinterview with the medical record librarian,, Theirobjections tended to be minor in nature and toconcern specific items rather than the generalapproach. Perhaps the most frequent difficultiesarose in connection with the definitions of thespecial classes of patients, referred to in ques-tions 3a and 4a (page 7 of exhibit I), whiclh weredesigned to establish the composition of the dis-charge lists to be used in sampling. The definition

of “well-newborn infants” was especiall y trouble-some. This term was not precisely defined for thepilot study. Consequently, when medical :recordlibrarians inquired about its meaning—ancl manydid inquire— the interviewer was often unable togive a satisfactory reply. Medical record librar-ians also asked if abstracts should be preparedfor the following classes of newborn infants: (a)premature births which were otherwise nqrmal,(b) well-newborn infants who were discharged be-

fore the mother, and (c) well-newborn infants whowere to be adopted and therefore remained afterthe mother was discharged. Although answers toeach’ of these specific questions were circulatedearl y in the pilot study, it would seem well to addthis information to the Hospital Discharge Surveyreference manual. Even with specific writteninstructions, situations will occasionally arise inwhich the medical record librarian is in doubtabout whether to prepare an abstract. .Perhapsthere should be a general rule for abstracting if

in doubt.The wording of questions 3a anti 4a could

stand improvement. It was the impression of thewriters that these questions did not fully serve

their intended purpose of defining the compositionof the daily discharge sheets.

Another question which should be rewrittenis number lb, which asks if the newborn infantreceives the same medical-record number as the

26

mother. Since the infant is far more likely to beassigned a number of its own, the question shouldbe worded in this way. Also, it should provide forthe reply that the infant receives the mother’snumber but with a letter attached. The latterpractice seems to be as common as that ofassigning the mother’s number to the infant.

The person interviewed about the medicalrecords should be the person who will be responsi-

EVAN.JATIONin the spring of 1965 an evaluation program

was carried out inmost of the pilot study hospitalsin order to check the quality of the sampling andabstracting. me purpose of this program was notto assess blame for errors but rather to gaugethe magnitude of these errors and the effect theymight have on the data which are collected andtabulated.

PROCEDURE

The check on the quality of the medical recordsamples consisted of two parts—first, an evalua-tion of the completeness of the frame from whichthe sample was selected, and second, a check onthe accuracy of the sample selection. The com-pleteness of the sampling frame was verified bycompiling the following figures for a 3-monthperiod: (1) the number of patients in the hospitalat the beginning of the period, (2) the number ofadmissions during the period, obtained by countingthe patients on the appropriate admission lists,(3) the number of discharges during the period,obtained by counting the number of patients on theappropriate discharge lists (i.e., the samplingframe), and (4) the number of patients in thehospitzd at the end of the period. The samplingframe was considered to be complete if the sumof the first two figures minus the third figurewas equal to the fourth figure. Small discrepancieswere acceptable, but differences larger than thesampling interval for that hospital (for example,25, where the sampling rate was 4 percent) hadto be reconciled.

Once the sampling frame had been evaluatedin terms of total numbers, the sampling itself waschecked for a l-month period. The procedurewhich was followed depended on the types of

ble for future work on the survey. This isespecially important at hospitals which will bedoing their own sampling and abstracting. If theinterviewer learns that. someone other than themedical record librarian will handle the survey,he should request that person’s presence for theinterview.

PROGRAMrecords which the individual hospitals had availa-ble. The usual procedure was to check the sampledrawn from the discharge lists against what thesample would have been had it been taken fromthe admission lists.

The evaluation of the data obtained from themedical records also consisted of two parts.First, the medical-record face sheet form usedby each hospital was scrutinized by Hospital Dis-charge Survey staff to determine its adequacy interms of information needed for the survey. Thena sample of the medical records which had al-ready been abstracted was reabstracted by theNational Center for Health Statistics, and the twosets of data were compared to see if the instruc-tions for abstracting were being uniformly fol-lowed by all abstracters.

In order for the duplicate abstracts to be com-pleted at the National Center for Health Statistics,facsimiles of the sample medical-record facesheets had to be prepared at the participatinghospitals. The methtis by which these face sheetswere reproduced will be discussed in the followingsection.

COOPERATION OBTAINED

Seventy-two of the 82 hospitals which partici-pated in the piIot gtudy ccqerated in reproducingthe, sample face sheets. Of the remaining 10 hos-pitals, 1 was already submitting facsimiles aspart of the regular data-collection program; itis not included in table 10. Five hospitals whichhad hesitated about participating in the surveyinitially were not asked to submit facsimiles be-cause of the risk that an additional request mightresult in their withdrawing from the study entirely,and four hospitals which were asked refused therequest.

27

Table 10. Distribution of pilot study hospitals, by procedure for sampling and ab-stracting records and degree of cooperation in supplying face sheet facsimilea:l Hos-pital Discharge Survey, 1964-65

Final procedure2

Degree of cooperation All A cpro- or B or

cedures E D

Total -- -------------------------------------- - -- -- --- -- --- 81 48 3 30

Cooperation obtained -------------------------------------------- 72 44 3 25Cooperation refused--------------------------------------------- 4 3Not asked to cooperate ------------------------------------------ 5 ; : 2

lFigures exclude 2 hospitals which failed to participatehospital which supplied facsimiles

in the pilot study and 1as part of the regular data-collection procedure.

2Procedures A and E— Bureau of the Census or National Center for Health Statisticsselects the sample and abstracts all the required data.

Procedure B—Bureau of the Census selects the sample and abstracts the demographicdata; hospital abstracts the diagnostic and surgical information.

Procedures C and D—Hospital selectsdata.

Ninety-two percent of the hospitals whichwere following procedureA and83percentofthose

which were following procedure C were willingto submit facsimiles (table 10). Although thehos-pitals were instructed by the Hospital DischargeSurvey to block out the patient’s name when pre-paring the facsimile, a number of hospitals didnot feel it necessary to do so. About halfof thehospitals which agreed to cooperate stipulatedthat the facsimiles should be made at the sametime the initial abstracts were prepared, inorderto avoid pulling the sample records solely fortheevaluation program.

Most of the hospitals which agreedtoprovide

face sheet facsimiles weregiventheir choiceastohow the reproductions would be made. (Afewofthe hospitals which didtheabstracting themselveswere asked to make copies of the face sheets atthe same time the abstracts were preparedandto

send them to the National Center for Health Sta-tistics alongwiththe regular monthlytransmittal. )The facsimiles were preparedbyhospital person-nel at slightly more than one-halfofthecooperat-ing hospitals and by nonhospital personnel at theremaining hospitals (table 11).

the sample and abstracts all the required

Eighty-eight percent of the hospitals whichdid their own abstracting prepared their own

facsimiles, whereas 61 percent of the hospitals

at which the abstracting was done by nonhospitalpersonnel preferred to have the facsimiles madeby nonhospital personnel. Where thehospitalshadno equipment suitableforreproducingface sheets,the Bureau of the Census provided a special

camera which could be used for this purpose. In29 hospitals the facsimiles were made by survey

representatives using the Census camera.

RECOMMENDATIONS

Considerable tact may be required of theindividual who introduces the evaluation programin a hospital where the sampling and abstractingare done by hospital personnel, to make sure themedical record librarian does not feel that thequality of her work is being questioned. Anim-pression of this kind can create resentment. TheHospital Discharge Survey should continue thepractice of informing themedicalrecord librarianof the evaluation program at the time of initialinterview, explaining at the same time the reason

28

Table 11. Distribution of pilot study hospitals cooperating in the evaluation program,by procedure for sampling and abstracting records and method of obtaining. face sheetfacsimiles: Hospital Discharge Survey, 1964-65

I Final procedural

Method of obtaining facsimiles

All methods -----------------------------------------------

NCHS or Census representative with Census camera----------------NCHS or Census representative with hospital equipment -----------Hospital. personnel with hospital equipment ----------------------

~ All A cpro- or B or

cedures E D

72 44 3 25

29 25 22 1 ;

3: 17 - 22

1Procedures A and E—Bureau of the Census or National Center for Health Statistics

selects the sample and abstracts all the required data.

Procedure B—Bureau of the Census selects the sample and abstracts the demographicdata; hospital abstracts the diagnostic and surgical information.

Procedures C and D—Hospital selects the sample and abstracts all the requireddata.

.

for the evaluation .An official statement preparedby the National Center for Health Statistics ontheneed for evaluation might prove to be helpful.

The arrangements fortheevaluation programwere made with the medical record librarian,occasionally by telephone. When the appointmentwas made bytelephone, theinterviewertentativelydetermined which of several evaluation methodshewouldfol,low at the hospitalby asking aseriesof prescribed questions. Because it is difficultfor the medical record librarian to understand

from the questions precisely what materials theinterviewer will require, itis suggested that theinterviewer write to the medical recordlibrarianto confirm the appointment and to requestthat thenecessary materials be made ava:lable. The ma-terials should be listed preciselytoavoid possiblemisunderstanding. The list should include ma-terials needed for the verification ofthesamplingframe as wellas those neededto verify thesam-pling itself. Such advance notice might help toavoid confusion on the day of the visit.

29

In the summer of 1963 the Graduate Schoolof Public Health, University of Pittsburgh, under-took a study for the National Center for HealthStatistics of the U.S. Public Health Service todevelop the mechanics of securing cooperationof hospitals in the continuing collection of mor-bidity statistics on hospitalized patients and toexplore methods of obtaining such data. The PublicHealth Service had already inaugurated a House-hold Interview Survey and a Health ExaminationSurvey and planned to complement its programwith a Hospital Discharge Survey of “nationalScope .

In order to determine the circumstancesunder which hospitals would be willing to co-operate in a national survey, the problems thatwould be encountered in translating records ofdischarges into a sampling frame, and the amountof information that would be uniformly availablein all hospitals, the Graduate School of PublicHealth interviewed hospital administrators andmedical record librarians and scrutinized medicalrecords in 45 purposively selected hospitalsthroughout the United States. A detailed report onthe study was presented to the National Center forHealth Statistics in 1964. An earlier section of thepresent report has summarized the feasibilitystudy findings, with emphasis on those which re-late to securing cooperation in a hospital dischargesurvey.

While the feasibility study was still in prog-ress, methodology was developed by the Nationa 1Center for Health Statistics, with the assistanceof the U.S. Bureau of the Census and the GraduateSchool of Public Health, for a national hospitaldischarge survey. The new procedures and formswere then tested by these three agencies in aprobability sample of 84 of the Nation’s short-term hospitals during the fiscal year 1965. Thisreport has, in earlier sections, described themethodology and presented some of the resultsof this pilot study. Again, aspects of the studywhich relate to the matter of cooperation havebeen emphasized. The more important findingsand recommendations are restated in this sum-mary.

‘l’he procedure for getting in touch with theadministrators of the 84 hospitals which fell intothe pilot study sample consisted of the followingsteps: (1) a letter from the Director of theAmerican Hospital Association, in which the sur-vey was briefly described and the hospital’sparticipation was requested; (2) a letter from anofficial of the National Center for Health Statis-tics, in which the survey was described in moredetail and participation was again requested and(3) a telephone call from a representative of theNational Center for Health Statistics, the Bureauof the Census, or the Graduate School of PublicHealth, depending on who was to conduct theparticular interview, requesting a personal inter-view with the administrator and his medicalrecord librarian. All of the administrators butone readily granted interviews, and an interviewwas eventually arranged with the remainingadministrator.

The primary objective of the interview withthe administrator was to establish rapport withthe hospital and to secure cooperation undermutually satisfactory circumstances; anotherimportant objective was to obtain certain statis-tical information about the hospital itself. Nearlyone-half of the administrators chose to have theirmedical record librarians present from thebeginning of the meeting, and where this was thecase, the interviewers generally believed thather presence had been helpful. When the adminis-trator and the medical record librarian are inter-viewed together, a second explanation of the surveyis not required. Also, having both present facili-tates reaching an agreement which is acceptableto all concerned. If the administrator and themedical record librarian are not in accord aboutthe most acceptable way of proceeding, their dif-ferences can probably be reconciled without delay:Another advantage to interviewing the two to-gether is that the medical record librarian is of-ten the one who can furnish the needed statisticalinformation on discharges, births, and the like.After the survey has been explained and an agree-ment reached, that part of the interview whichrelates specifically to sampling and abstracting

30

medical records can be continued with the medicalrecord librarian in her own office.

Only 2 of the 84 hospitals which were askedto participate in the pilot study failed to do so,and 1 of these is now participating in the regularsurvey. The memorandum of agreement, whichserved as a record of understanding with thehospital, was signed during the interview with theadministrator at 73 hospitals. At the remaining9 participating hospitals, the approval or adviceof someone other than the administrator @d tobe sought, which resulted in a delay in obtaininga signed agreement.

During the interview with the medical recordlibrarian, the initial sample of medical recordswas selected. The proportion of records sampledwithin each hospital varied inversely with the sizeof the hospitals and ranged from 20 to 40 percentof the records at the smallest hospitals (fewerthan 50 beds) to 1 percent at the largest hospitals(1,000 beds or more). The actual method ofselecting the sample was .determined by thenature of the patient-listing maintained by thehospital. The preferred type of list was a dailyor monthly listing of discharged patients showingtheir medical record numbers.

Several different methods of data collectionwere tried out during the pilot study year. Thesewere of three basic types. The sampling andabstracting of the medical records were doneeither by nonhospital personnel (procedure A),by hospital personnel (procedure C), or by bothhospit.d and nonhospital personnel (procedure B).The pilot study experience indicated that eitherof the first two data-collection procedures (A orC) would work at most hospitals. Forty-four ofthe 47 hospitals which were offered procedurkA and 28 of the 29 hospitals which were offeredprocedure C agreed to cooperate on that basis.Only three out of eight hospitals accepted pro-cedure B; it was generally regarded as inefficientby administrators who could see no reason to havetwo persons working on the same abstracts,particularly at different times.

The hospitals in which survey personnel didthe sampling and abstracting were offered pay-ment ranging from $10 to $50 per year, the amountdepending on the expected annual number of ab-stracts. Hospitals which did the work themselveswere offered 30 cents per abstract, and those

which shared the work with survey personnel wereoffered 15 cents per abstract. Altogether, eighthospitals declined any payment, and three soughtan additional amount. The A-procedure hospitalsaccounted for seven of the hospitals which refused

Pyment and for two of those which requested anincrease. Some administrators who did not acceptPWment let it be ~0~ that they considered theirparticipation a public servic~ others indicatedthat the amount offered did not warrant the cost ofsetting up and maintaining a special account.

Two different abstract forms were usedin thePflot study (1) a conventional form in which theentries were written in (exhibit C), and (2) aform in which appropriate spaces were blockedin for scanning by an IBM “reader” (exhibit D).The first form was used in 40 hospitals and thesecond in 42. Both forms found ready acceptance,although a few hospitals had a little difficultygetting accustomed to the IBM form.

In the spring of 1965 an evaluation programwas carried out in most of the pilot study hos-pitals. The purpose of this program was not toassess blame for errors but to measure the errorsand gauge the effect they might have on the datawhich are collected and tabulated. The samplingwas evaluated by studying the sampling fkame forpossible deficiencies and by checking the accuracyof the sample selection. T%equality of the data wasevaluated by studying the medicaI record facesheet forms to determine their adequacy in termsof information needed for the survey and bychecking a sample of the abstracts which hadalready been prepared.

In order to accomplish this last objective itwas necessary to make reprmluctions of themedical record face sheets from which the ab-stracts had. been prepared. Seventy-two of the82 hospitals complied with this request, but 33of these stipulated that the facsimiles should bemade at the same time the initial abstracts wereprepared so that the records would not have to bepulled for the evaluation program.

This report has tried to make plain that thespirit of cooperation among hospital adminis-trators and medical record librarians is strong.If any serious problems are encountered in ex-panding the Hospital Discharge Survey sampleto several hundred hospitals, lack of cooperationwill probably not be one of them.

31

Once again, in order to summarize, among

the main findings and recommendations relatedto cooperation in the hospital were the following:

1.

2.

‘3. .

The hospitals by and large accepted theprocedure for abstracting requested ofthem by the Hospital Discharge Survey.The exception was the procedure in whichboth hospital and nonhospital personnelwould share in the abstracting (procedureB). This procedure should be dropped.

Some hospitals requested modificationsin the basic procedure because of per-sonnel, space, and similar problems.These changes should be accepted wherethe y do not affect the data gathering.

At times, the interviewers seeking to ob-tain the hospital’s cooperation were,somewhat to their embarrassment, un-able to answer questions about the surveyput to them by the administrators. It is

obviously impossible to anticipate everysuch question, but some of the need formore precise informing of the interview-ers can be met. Statements about the

purpose and uses of the survey should bestrong and specific.

The interview with the administratorshould not start with the solicitation ofnumerical data. Rather, it should beginwith an informal discussion of the survey.

4.

5.

6.

7.

Early in the pilot study, the recommen-dation was made to the interviewers by theHospital Discharge Survey that the admin-istrator and medical record librarian beinterviewed at the same time. This ispreferable to separate interviews. In anyevent, the medical record librarian shouldnot be interviewed in advance of theadministrator.

There are some minor matters which can

further the cause of cooperation but whichmay not perhaps have received sufficientstress during the interviews. Among theseare (1) the relatively small proportionof records actually to be abstracted,(2) the maintenance of confidentiality,

and (3) the voluntary. character of the hos-pital’s participation.

Some definitions involved in the interviewwith the medical record librarian seem toneed sharpening. While this matter is alittle removed from the question of co-operation, it is a possible source ofirritation and should be attended to.

The evaluation program, which was de-

scribed in the foregoing section, was ap-

proached with wariness so that hospitalpersonnel would not feel their abilitieswere being questioned. With proper pre-cautions, which have been mentioned, itappears to have turned out successfully.

REFERENCES

Woolsey, T. D.: Hospital Statistics Fmm the U.S. Na-tional Health Survey. Prepared for presentation at the annual

Summer Conference for Executives of Allied Hospital Asso-

ciations, Mackinac Island, Mich., July 26-27, 1963.

2Sirken, M. G.: National hospital patient statistics. HOS-

pitazs, Journal of the American Hospital Association, in press.

3N~tiona] center for Health Statistics: Origin, Progr~)

and operation of the U.S. National Health Survey. Vitrd and

Health Statistics. PHS Pub. Nc. 1000-Series l-No. 1. Public.

Health Service. Washington. U.S. Government Printing Office,

Aug. 1963.

%ational Center for Health Statistics: Health Survey pro-

cedure, concepts, questionnaire development, and definitions

in the Health Interview Survey. Vital and HeaW Statistic.PHS Pub. No. 1000-Series l-No. 2. Public Health Service.

Washington. U.S. Government Printing Office, May 1964.

5National cen~r for Health Statistics: plan and initial

program of the Health Examination Survey. Vita2 and f#eah%Statistics. PHS Pub. No. 1000-Series l-No. 4. Pmblic Health

Service. Washington. U.S. Government Printing (Office, July1965.

6National Center for Health Statistics: Development and

maintenance of a national inventory of hospitals and insti-tutions. Vita? and ffeatth Statistic. PH~ Pub. No. 1000-

Series 1-No. 3. Public Health Service. Washington. U.S.

Government Printing Office, Feb. 196,5.

—.ooo —

32

APPENDIX I

FEASIBILITY STUDY

Exhibit A. Letter to Hospital Administrator

UNIVERSITY OF PITTSBURGH

GRADUATE SCHOOL OF PUBLIC HEALTH

PITTSBURGH, PENNSYLVANIA, 15213

DEPARTMENT OF B1OSTATIST1CS July Z, 1963

Dear (Hospital Administrator):

The University of Pittsburgh is conducting a study on behalf ofthe National Center for Health Statistics of the U.S. Public HealthService to determine what problems there may be in collecting statis -tical information from hospital records._ The purpose of this letter isto request your cooperation in the study.

The main objectives are to determine what information is availa -ble in hospital records from which comprehensive national statistics canbe generated, the limitations the data might have for statistical purposes ,and what problems may be encountered in abstracting the information.To get some answers to these questions we are visiting about 50 short-stay hospitals in various parts of the nation. The study procedure in-volves a brief interview with the administrator and medical recordlibrarian. k addition, it calls for selecting a small random sample ofdischarges and abstracting information from medical records. The namesof patients would be of no interest to us except as a means of locatingthe records.

We will contact you by telephone within the next few days to de -termine the most convenient time for visiting your hospital.

We will greatly appreciate your cooperation.

Sincerely yours,

Is idore Altman, Ph.D.Professor of Medical Care Statistics

33

Exhibit B. Abstract Sheet

i“orm HRS 3(d)-3 Budget Bureau No. 68-6341; Approval expires Dec. 31, 1963

University of Pittsburgh - Graduate School of Public HealthStudy of Feasibility of Sampling Hospital Discharge Records

DISCHARGE ABSTRACT SHEET

-—Name of Hospital

I

Address.—

I

Abstracter Date

1 I

1. HOSPITAL CODE NUMBERl-Govt. -Nonfed. 3-VO1. -Other

2. CONTROL 2-VO1. -Church 4-Proprietary 5-Govt. -Fed.NUMBER 1-25-49 3-100-299

3. OF BEDS 2-50-99 4-300+l-Northeast 3-south

4. GEOGRAPHIC REGION 2-North Central 4- West

I i I I l-l

1. AD IMLSS ON— 99T

- looi-White 3‘-Not -a

OY-N:ed

wborn (Go to 12— ‘EM 29)

11. COLOR 2- Nonwhite 4-Not answered [ P1- .Male

12. SEX 2-Femalel-Never married 3- Widowed

13. MARITAL STATUS 2-Married 4-Divorced 5-Separated1

OCCUPATION 1- Patient 3-Husband14. ITEM: 2-Father 4-Other 5-Not asked—-.

OCCUPATION (specify) 000-Not asked15. OOY-Not a !!1

. .l-No

16. EMERGENCY? 2- Yes I—ACCOMMODATION l-On admission 3-Preferred 5-Most of sta~ I

17. ITEM: 2-At discharge 4-C ornbination 6-Not asked1-Private 3-Ward 5- Other

18. ACCOMMODATION 2-Semi- 4-Nurser y 6-CombinationAlive 1-With approval Died

DISC IMRGE — 2-Against advice ~Autops ySTATUS 3-Transferred 6-No autopsy

19. 4-Other II—,.—. -

CONFIDENTIAL . This information is colkcted un~cf ~u!hority Of ~ubl;c Law ~$2 O( the 84th COnRfess (70 %at. 48%

42 IJ.s.(:. ?I05). All information which would permit identification of an individual Or of an establishment will be held

strictly confidcnci ai, will bc used only by persons cngmgcd in and for che purposes of :h@ survey and will not b.? dis-

closed or released co ocher persons or used for any othcv purpose (22 FR 1687).

34

Form HR.8 3(d)-3

(Hospital) (City, State) (Medical Record No. )

20. ADMITTING DIAGNOSES OR COMPLAINT (Number 1, 2, 3, etc. )

I

21. FINAL DIAGNOSES OR COMPLAINT (Number 1, 2, 3, etc. )

II 1

IIOPERATIONS (Specify)

COMPLICATIONS (Specify)1-No

9

23. 2-Yes

Physicians)

22. 00- None

24. CONSULTATIONS (No. of01 - Not asked II

25. ADMITTING HEIGHT (inches 1 02 - Not answered888 - Not asked

26. ADMITTING WEIGHT (lbs) 999 - Not answered1-99.0 or less 5-102 .0-102.9

ADMIT TING 2-99.1 -99.9 6-103 .0-103.9 9-Not askedTEMPERATURE 3-100 .0-100.9 7-104.0-104, 9 0- Not answered

27. 4-101 .0-101.9 8-105.0+

Form HRS 3(d)-3

(Hospital) (City, State) (Medical Record No. )

28. LABORATORY TESTS AND THERAPIES

URINAL YSIS

HEMATOLOGY

SEROLOGY

BLOOD TYPE

x-RAY

CHEMISTRY

BACTERIOLOGY

GASTRIC STOOL

BODY FLUIDS

HISTOLOGY

FUNCTION (EKG, EEG, BMR)

THERAPIES

OTHER

36

Form HRS 3(d)-3

(Hospital) (City, State) (Medical Record No. )

NEWBORN

29. COLOR 2- Nonwhite 4- Not answered !1- White 3 - Nbt asked I

1 - Male30. SEX 2 - Female I

01 - Not asked31. AGE OF MOTHER 02 - Not answered

1- llLive birth order”

32. BIRTH ORDER ITEM 2- llBirth orderll 3 - Neither asked IBJFITUfi.lh’13S2QfQ34. PR

m .4 n wJs,J.7aJx

32! OY . Not answered 1l-NQ

:EMATURIT Y? 2 - Yes

2- Yes3 - Not asked }

2- Yes3- Not asked

38.

2 - Yes3 - Not asked }

DISCHARGE 2 - Against advice = AutopsySTATUS 3 - Transferred 6 - No autopsy }

40. 4- Other-.

— I

35. BYRTH WEIGHT (Gin.) I

36. BIRTH LENGTH (Cm.BIRTH INJuRIES (Specify) l-No

37. IMALFORMATIONS (Specify) l-No

OTHER ABNORMALITIES (Specify) 1-NO

39. IAiive 1 - With approval Died (inc. Stillborn)

37

Exhibit

APPENDIX II

PILOT STUDY

c. Conventional Abstract Form

E..— .———

[ {) NFIDENTIAf-.Thi. jnfomatiofi is collected uncle, .uthotj~ .f pub]jcLaw(jszof !h~ad!htin~css (70 %*c 4SR 42 IJ.S,C. 242 c). All in

).)

form .tion wI,ich would permit idemi[ictwion of an imdivid.d OCof am est.blishmetu will be held strictly confidential, will be .sed anly by personsc. a cd ,.- .md for the pumoses of & survey aqd will not be dklosed m reltased to + ptrsons or used for my other PUP sc (22 FR 16s7

ABSTRACT OF -PATIENT RECORD-HOSPITAL DISCHARGE SURVEY-——.—— . ._-———

1. Hospital Number 1

~._ &atient Control Number 13. Medical Record Number

lm. dmy w

4a. Date of Birth HIHHI1Complete 4b and c onJJ if date of birth is not given

I i 1

4r. AVI= is mta+md in: 1 . Year= 2 - Mtmthm !4 . mvs I II

5. Sex 1- Male 2- Female I 11- White 4- “Nrmwhite”

6. Color: 2- Negro 5- Not stated rl3- Other nonwhite I J1- Married 4- Divorced I

7. Marital Stature 2- Single 5- Separated3- Wicbwed 6- Not stated

m. day w.

8. Date of Admission 1~111 II 1.Uay

9. Date of Discharger—1

10. Discharge Stature 1,- Alive 2- Dead 1 I11. Final Diagmoeem

Ma. was an Op eration performed? 1- ,Yee 2-No12b. Ope-raticmm

1111

Completed by Abstracter Date

PHs-4734-1 DF>ART?4ENT OF HEALTH, EDUCA2’SON, md WELFARE Fom “kppmvek 68 -R62Q. RL28-64 Public Hedtb Semite

fkionaf Center dx Sfeakb Sutiwtics

38

D. IBM AbstractExhibit

CONFIDENTIAL- This information is collected under authority of Public aw~e:::f the 84* Congrem (70 Stat. 489; 42 U.8.C. 242 c,). All inform- -ation which would permit ident~lcation of an individual or an eatablin 11be held strictly confidential, wifl be used only by permrrs en- ——gaged in and for the purposes of the survey and will nut be disclosed or released to other persons or used for any other purpose (22 FR 1087). z

DEPARTMENT OF F.,m App,a”, d

HEALTH, EOUCATION. AND WELFARE 8.6*. I M... M.. 68.R620.R2 ,-PHS.4734.2

PUEILIC HSALTH SERVICE8.64

NATIONAL CENTER FOR HEALTH STATISTICS [1. HOSPITAL .NUMBER

..... ..... ,=,:, ~::,, ::::, -... ..... ..... ..... ----- ------ ----- ..... ..... ---- ..... ..... _----- ----- ..... ..... ..... ----- ----- ::==:..... ..... ----- ----- ..... ..... ..... ::=: ::::: _

0. 2?.4 Ss----- ..!_ . . . . . . . . . . . . . .

7 ●..-. ---- ::=: -.:... . . . . . . . . . —. ..-. . . . . . . . . . . . . . . . . . . . . ---- . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . -. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

----- :=::, ~,=, ----- -----. . . . . . . . . . ---- ---- ---- . . . . . . . . . . ------ -... ---- ---- -... . . . .

ABSTRACT OF PATlENT RECORO-Hospital Okcharge Suwey

2. PATlENT NUM8ER

..... ,=:, :=,, ----- --------- ----- .-.. :=,, :==,----- ---- ....- ..... -.-.. .....01 2 34 5 s 7 B 9=

----- ----. ---- . . . . . . . . . . ----- :::::----- . . . . . . . . . . . . . . . . . . . . ..- . . . . . . . . . .----- . . . . . . . . . . . . . . . _

3. MEDICAL RECORD NUMBER ----- ..... ..... ..... ..... ----- ....- ..... ..... ---- -..... ..... ..... ..... ----- ---- ..... ----- ---- .......... ..... ----- ..... ~:::: ..... ..... ---- ----- .....----- ----- ----- ..... ..... ..... ..... ..... ..... --... ----- :::,, -----...-. .....----- ----- ..... ..... ----- ---- .........- -... ..... ..... ..... ----

J,,==

MONTHM

. . ..-. . . . .

0. . . . .-----

O&v0

.-. .-----

-l-—..-.

-!..-.T-1

---------

I.—..-----

. . . . .. . . . .

1-.... . . . .

4a. DATE OF BIRTHc==::,::

s=..........

..... -.....s=..........—-

M1----- ---- J.---- ----- ..---

Aso. . ..- . . . . . . . . . .. . . . . . . . . . . . . . .

5 s 7..-. . . . . . ----. . . . . . . . . . ----

::::: mm

. . . . . . . . . . . . . . .. . . . . . . . . . . . . .

M. . . . .-----

8.--------

. . . . .. . . . .

8. . . . ..-..

2 a. . . . . . . . .----- ----- mm

z 3----- . . . . . --:-- Ulms----- . . ..- . . ..-

:::=: 1800

----- . . . . . . . . . .----- . . . . . ---- TV(S

Complete4b and 4C if

date of birth is not given.

w-a . . . . .. . . . .

0----. . . . .

r. 34---- -... . . . . .-... . . . . . ----- W7S

5 6 7. . . . . . . . . . . . .. . ..- ---- . . . .

4.b. AGE ....- ..... ..... ..... ..... Tch’s .-_. . . . . . . . . . . . . . . . . . . . .----- . . . . . ---- . . . . . . . . . . -. . . . . . . . . . . . . . . . . . . . . . . . .

0 s e7 *..-. . . . . . ..z_. . . . . . ..?-. ~,* ----- ----- :::= -4--

0=. . . . . . . . . .---- ----- . . . . ----- . . . . . ----- . . . . .----- ●

:::= YrARs =::= mums ::== MM

::::: MAlf :Zz: fmblc

4.c. AGE IS STATED IN

5. SEx

:::=: W14111 ::::: ,,”~~~ ,,

::=: lam ::=: aorslmo

::ZZ OTHCR NOliWllTS

6. COLOR

:::,, MAmmco ::::= Wouw

::::: SIIKIE :== acmmlm

::=: Mrwn’m ::=: NOT SIAICO

JF----- -----. ---- . ----

MQmnA 0=

. . . . . -!.. ~:,, $ 0-L .-!_ ..... .. .. ..... :::::----- ----- ....- ..... ----- ..... ----- ..........01 2 s

. . . . . ---- . . . . . . . . .. . . . . ---- ---- . . . . . lEIN

DAY01 2 a 4 s s 7 s *=

..-. . . . . . . . . . . . . . . . . . . . . WITS . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . ---- ----- ----- . . . . . . . . . . . . . . . . . . . . . . . ...

01 z 3* 6 e 7 #WAR

s=. . . . . ---- ---- ---- . . . . .---- ---- . ---- . . ..- . . . . . . . . . . . . . . . . . . . ::::=-... . . . . . ----- :::::

7. MARITAL STATUS

8. DATE OF AOMISSION

,,..... . ...-----mm ‘~--

As----- ..*L .-”.- ..... ..... 0 N 0=..-. ::::: ~:=: ----- -----....- .... . ..... ..... ..... ..... ..... . ....

01 2 s. . . . . ---- ---- . . . . ...-. . . . . . . . . . ----- TensDAY

o I z s s..:.. WITS

o 7 # s=----- ----- ----- . . . . . . . . . . . ---- -... . . . . . . . . . . . . . . .. ---- . ..- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .— -----

inn 0 1.2 s 4 s 6 7 @s----- . . . . . ----- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . ----- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9. OATE OF OISCHARGE

10. DISCHARGE STATUS ::::: &uvc ,:::: (mm IBM HS12S4

39

——I —

IL FINAL DIAGNOSES:

PATlENT NUMBER

. . . . . . . . . . ,,, ,, . ..-. . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . —. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I ~,,,,: ,,,,::,::, ,,,::,,:,, ..... ..... ..... ..... ..... —....- ----- ..... ..... ..... _I ——

—_

—_

——_

—— —

——

—_12a, WAS AN OPERATION

YES NO—

PERFORMEO ?. . . . . . . . . .. . . . . . . . . .

12b, OPERATIONS: ——

—_

——_—

—_——

————

——

COMPLETED BY ABSTRACTOR

—

DATE

F

o

R

N

c

H

s

u

s

E

o

N

L

Y

DIAGNOSIS CODES

1

I..... ,,,,: .... ..... ..... .......... ..... ..... ..... .......... ..... ..... ..... .....I,:,,, Y ::::: ,:::: ::,,, ::::, . . . . .. . . . . ,:, ,, -----. . . . . ,,, :: -----. . . . .

l::=: ,::::::::::::::::,:, . ... ,,,,,- -.--, ,,:,. ----..... .....

.. . ...... . ,:,,: ..... .......... . . . .... ..... ..... .. ..F’H6~4734.2 (BACK)

40

—— ——

OPERATION CODES—

,:,,: ..... ..... ..... ..... —.......... ,::= ,,,,, ,,,:...... ..... ..... ..... .......... _—

————

..... ..... ..... ..... ..... ..... ,,:,, ...-. —..... ..... ..... ..... ..... ,:,,, ,,,,,..... .....———

,,:,: ..... ,:::, ..... .......... ,,:,, ..... ..... ,,,,, ..... —..... ..... ..... ..... .....——————

———

IBM H912S5

Exhibit E. NCHS ‘rFact Sheet”

Hospital Number 6703Procedure A-1Stratum Number 54

Name of Hospital Sampling Digits 12, 42, 72AddressName of administratorTelephone number

This is a church-owned general hospital. It maintains 371 beds and employs 688persons. There are operating room and obstetrical delivery room facilities. TheMFI reports a chronic disease ward with 30 beds. This hospital does not maintaina nursing home unit. The average length of stay is 10 days.

The hospital is accredited and is a member of the American Hospital Association.It is reported in the 1964 AH4 Directory as having 10,494 admissions for the oneyear period ending September 1963. Using a sampling ratio of 3 out of 100, thi~office would expect about 315 patient record abstracts a year from this facility.An employee of the Bureau of the Census will abstract the data from this hospitaland record it on conventional forms (PHs-4734-1).

Hospital Discharge Survey BranchHealth Records Statistics Division

NCHSOctober 26, 1964

41

Exhibit F. AHA Letter to Hospital Administrator

Dear (Hospital Administrator):

The National Center for Heaith Statistics of the U. S. Public Health

Service is inaugurating a national system of collecting statistical.information from hospital medical records as part of the National

Health Survey. The importance of such data in contributing to ourknowledge about the Nationfs health and in providing a basis for

sound administrative decision -making cannot be overstated. The

American Hospital Association has been cooperating with the

National Center for Health Statistics in the conduct of the survey

and through its Board of Trustees has endorsed the survey,

Your hospital has fallen into the probability sample developed bythe National Center for Health Statistics. The purpose of thisletter is to request your participation in the pretesting phase ofthe survey. The Public Health Service will make every effort to

minimize the burden of reporting on the part of your hospital. An

equitable basis for compensation will be arranged by the National

Center for Health Statistics. All information collected will be givenconfidential treatment and will be used for statistical purposes only.

Any published summary will be presented in such a way that no

individual hospital or patient can be identified.

Dr. Forrest Linder, director of the National Center for Health

Statistics, has informed me that within the next few days he will

send you a more detailed statement concerning the survey. He will

greatly appreciate your cooperation.

Without the cooperation of your hospital and others like yours, this

important research effort cannot succeed. 1, therefore, urge you to

make every effort to cooperate in this survey.

Sincerely

Edwin L. Crosby, M. D.Director

42

Exhibit G. NCHS Letter to Hospital Administrator

Dear (HospitalAdministrator)

A few days ago, Dr. Edwin L. Crosby,Hospital Association, requested your

Executive Director, Americancooperatfl.onin the Hospital

Discharge Survey to be conducted by the National Center for HealthStatistics. He indicated that I would follow up his letter bysending you a more detailed statement concerning the plsns andobjectives of the Survey. This statement is enclosed.

!!TheHospital Discharge Survey will produce administrative andmorbidity hospital statistics. The Survey will collect informationfor a sample of discharges in a national sample of short-term hospitals.In the remaining months of this fiscal year, a pilot study will beundertaken in order to try out alternative procedures for conductingthe Survey.

Since your hospital was one of those randomly selected for the Surveywe would like very much to discuss matters relating to participationof your hospital. Therefore, within the next several days, a repre-sentative of the Bureau of the Census, acting as an agent of theNational Center for Health Statistics, will telephone you to arrangefor an appointment. This meeting should not take more than an hourof your time.

The National Health Survey program, of which the Hospital DischargeSurvey is a part, has been approved by the House of Delegates of theAmerican Medical Association, and the Hospital Discharge Survey itselfhas been endorsed by the Board of Trustees of the American HospitalAssociation.

Your cooperation in this Survey will be very much appreciated.

Sincerely yours,

Forrest E. Linder, Ph.D.Director

Enclosure

43

Exhibit H. statement on the Survey

September

DEPARTMENTOF HEALTH,EDUCATION,AND WELFAREPublicHealthService

NationalCenterfor HealthStatistics

THE HOSPITALDISCHARGESURVEY

18, 1964

The Hospital.DischargeSurveywill involvethe continuingcollectionand analysisof nationaldischargestatisticsfor short-stayhospitals. This surveyis a partof the NationalHealthSurveyprogramof the PublicHealthService. The NationalHealthSurveyis authorizedby PublicLaw 652, also known as the NationalHealthSurveyAct,whichwas passedin 1956by the 84th Congress. Earlier,thisyear,the Houseof Delegatesof the AmericanMedicalAssociation,impressedwith thescientificobjectivityof the NationalHealthSurvey,adopteda resolutionapprovingthe Surveyprogram.

The short-termhospitalsof the UnitedStatesrepresentan enterpriseof enormousproportions.Not onlydo they constitutea largeindustryemployingaboutamillionand a halfpersonsand providingservicesvaluedat over $10 billionannually,but theyare increasinglyoccupyinga centralpositionin the entiresystemof medicalcarein this country. At the sametime,thishospitalenter.priseis undergoingsignificantchanges. For example,the typicalhospitalisincreasingboth in sizeand complexity.

For thesereasons,the utilizationof hospitals,is beingstudiedmore and moreintensivelyby hospitalsthemselvesin orderto make betterdecisionsregardingoptimumuse of the facilities.Furthermore,the medicalrecordsof the hospitalscontaininvaluableinformationregardingthe relativefrequencyof occurrenceofmany diseasesthat are largelytreatedin hospitals.

To servesuchbroadneedsas these,the NationalCenterfor HealthStatistics(NCHS),the major statisticalagencyin,the FederalGovernmentresponsibleforthe collectionof healthstatistics,is undertakingthe HospitalDischargeSurvey. The Surveyhas been endorsedby the Boardof Trusteesof the AmericanHospitalAssociation.

Initially,the informationobtainedfor the sampledischargesin the Hosp55talDischargeSurveywill be limitedto’’’core”datawhichin nearlyallhospitalswouldbe availablefrom the face sheetof the patients’sfolder. The coredatawill includea few personalcharacteristicsaboutthepatient,suchas sex andage,and it will also includeadmissionand dischargedatesas.wellas finaldiagnosesand operationsperformed. Severalyearshenceit is likelythatadditionalinformationaboutthe sampledischargesavailablefromthe hospitalrecordswouldbe collected.Informationfromhomital recordswill be ~ostedon abstractformsprovidedbymentmadewith the individualeitherby the hospitalstaff,

44

the NCHS and,hospital,theagentsof the

depen~ingupon the type of &range.abstractingwouldbe performedCenter,or possiblybyboth.

-2

The HospitalDischargeSurveywill most likelycontaina nationalsampleofseveralhundredshort-stayhospitalswidelydistributedover the majorregionsof the country. The NCHSwill make arrangementswith eachof the samplehospitalsto participatein the survey. Severaltimesduringthe year,theprescribedinformationwill be abstractedfor a scientificallyselectedsampleof a relativelysmallnumberof dischargeswithineachsamplehospital. Theparticipatinghospitalswouldremainin the surveyfor severalyearsbeforebeingreplacedby otherhospitals.

The abstractedinformationfor the sampledischargeswill be transmittedtothe NCHSwhereit wouldbe processedpriorto the preparationof statisticaltabulations.Statisticalanalysesof the data on a nationaland geographicalbasis,but not for individualhospitals,wouldbe preparedfor publicationbythe Centerand made availableto hospitalsand otherinterestedparties. Allinformationcollectedin the surveywhichwouldpermitthe identificationofindividualsor hospitalswill be held strictlyconfidentialand will not bepublishednor be disclosedor releasedto othersfor any purpose.

Exploratorywork on developingthe plansfor the Hospital.DischargeSurveywasstartedlastyear. A more formalpilotstudyof someof the operational.problemsis beingundertakenin a nationalsampleof 100hospitalsat this thein orderto try out alternativeproceduresfor conductingthe full Survey,which,accordingto presentplans,will be startedwith thesehospitalsaboutJuly 1965.

45

Exhibit 1. Hospitai Interview Form

>RM NHS-HDS-2xD.2a-e4]

U.S. DEPARTMENT OF COMMERCEBuREAu OF THE CENSUS

ACTING AS COLLECTING AGENT FOR THE

U.S. PUBLIC HEALTH SERVICE

HOSPITAL INTERVIEW

HOSPITAL DISCHARGE SURVEY

~- R620. R:

:DN FIDE NTIAL - This information is collected for the U.S. Public Health -411 information which would permit identification of the individual will be l,eId;rrictlv con~ldenrial. will be used only by persons em?aged in and for k purposes—.>f the survey, and w’ill not be disclosed or-r.&ased to &b_crsfor any other p&ses.

tome and address (Verify and correct as necessary)

lProcedure I Abstract form —

. Initial

mail

contacts

sctian A - SUMMARY OF TELEPHONE APPOINTMENT AND FIRST PERSONAL VISIT

}Date sent Received —

Is. American Hospital Association letter D Yes n NrI

. Tclephene

contoct

. Summary

of persona I

interviow

with

Adminis-trator

i. inter-

viewer

b. National Center for Health Statistics letter I,Date

o. Date of telephone contact

b. Appointment made for personal visitDate Time a.m.

p:m.

Name lTitleI

Room and buildingI

Namec. Medical Record Librarian~

d. Comments and important items discussed with Administrator on the telephone

a. Hospital cooperating - n I,Jnquslified cooperation m Qualified cooperation n ~&c;d$-(Deacribe)

b. Contract Proposal - m Signed and attached I-J NOW#&rl

c. Final procedure agteed upon - DA UB UC UD UE m Other(lhacribe)

If additional space is needed for comments use the last pcge of questionnaire.

Name Regional Office Date foort completed

46

section B - INTERVIEW WITH ADMINISTRATOR

You hove probably tiad a chance to review the statement Name

Dr. Lindar sent you concerning the Hospital DischargeSurvey. If you have any questions obaut the survey, I Location (if different from that shown on address label)

will try to answer them for you.

Before discussing the details of the survey, I must Type of service (e.g., tuberculosis uait, psycbhttic

verify some infontrotion about - -.unit, general hospitel, etc.)

1. Is it correct that -- is a Number of discharges in 1963Patient-days DaiIy censu

awned hospitol? Total patient:days or

a Yesavetege daily census in 1963

a NO (Indicate correction)

2.0. How many hospital beds are maintained forAversge length of stay

inpatient use, ●xcluding bossiaets? Does this unit maintain sepomte discharge records?m Yes m No - &;:: ~~ij~n;r;:~,s fmm

(Nuder af beds)

b. Haw many bassinets are maintained?

(Number of bassinets)

c. If less than Z5 beds in question 2a, ask:

Does --Location (if different from that shown on address label)

provide araund-the-clock nursingservice by a registered nurse?

Type of service ( e.6., tuberculosis unit, psycbkttricunit, general hospital, cCC.)

3a. ~o~imately how many discharges were there-- in 1963, ●xcluding well-newborn? Number of beds for inpatients

Number of discherrres in 1963

Toml patient-days or Patientd8ys Daily censw(Number of discharges) average daily census in 1963

b. How many births were there in -- in 1963? Average leagth of stay

(Number of births)Dees fhis unit maintain sepamte discharge records?

= Yes4. What is the average length of stay for

n No - How can the discharges fromthis unit be identified?

patients in -- ?

(Number of days)

5a. Is - - a hospital complex, port of a hospital ,complex, or neither?

Location (if different from that shown on eddress label)

I-J complex = l%: ;~xa D NeitherP Type of service ( e+, tuberculosis -unit,psychiatric

b. If “Part of a complex,” ask: umt, genersl bospttala, etc.)

What is the name and address af theparent organization?

Name

Addmss

Ii6. If “Complex” or “Part of a complex, ” enter the

in fortnstion requested at right for EACH sepsrmteunit.

Number of beds for inpatients

Number of discharges in 1963

Torsl petient-days or Pstient-days Daily census

average daily census io 1963

Avekage Ieagth of stay I

Does this unit maintain sepomte discharge records?

D Yes D No - How can the discharges fromthis unit be identified?

IFORM NHS-HDS-Z X (t O-2 S-64J

47

Section B - Continued

.ls -- associated with any other estobl ishment, 8. Does -- maintain a nursing home or nursingsuch as a resident institution or a school, hame unit?excluding medical and nursing schools?

a Yes a No (Skip to question 9)

U. Yes n No (Skip to question 8) If “yes,” ask:

f“yes,” ask: a. How many nursing home beds are maintained?

a. What is the name, address, and nature of theestablishment with which - - is associated? (Number of beds)

Name b. Were these beds excluded from the countaf hospital beds?

Address

m Yes = No

9. Does -- maintain a convalescent unit?

Nature

= Yes D No (Skip to question 10)

b. What type of service does the establishmentnamed above provide. If “Yes, ” ask:

a. How many convalescent unit beds are maintained?

c. Does - - limit its services primarily to thepopulation af a residerlt institution?

(Number of beds)

~ yes (Describe population 8erved = No b. Were these beds excluded from the count

and services provided> af hospital beds?

= Yes = No

c. Daes the date of discharge an the hospitol recordrefer to the dat6 of discharge from the convalescentunit or ta the convalescent unit?

m From = Toconvalescent unit convalescent unit

Lemarks

.——

— ——

USC OMM-DC

48

Section B.-Continued

10. I will outline briefly the steos in the srrrvew.Fora sample ofpa~ents, a iimited antount;finfomration will be abstmcted to this farm(Show abstmct form.) As you can see, thisirrfamration will genemlly be available on themedical record face sheet.

Forahospital ofthissize, the sampling ratio

will be in , or about

records in the course of a yeor -

approximately records per day.

(Read Procedure at right.)

‘INTERVIEWER:

Wbich procedure was finally agreed upoP?

(Procedure)

Comments

Il. Before lgotathe next item, doyouhave. any(other) questions on what has been covered?

Dyes (Enter que8tion8 m Noand yorrrmaponaerr)

:ORM NHS.HDS.2X (10.2 s.64)

PROCEDURE A - An agent of the Public HealthService will visit your hospital about once every 3months, on a day and at a time convenient to yourMedical Record Librarian. Atihattime we will selecta sample of patients who hove been discharged.This list will be left with the Merfical Record Libmr.ian so that the face sheet of the medical records forthese sample patients can be ready for our next visit.On that second visit, we will abstmct information fromthe face sheet of the medical records and select a new

sample of discharges. On all succeeding visits, wewill repeat the process - each visit lasting only a fewhours. Once or twice a year, arepresentative of thePublic Health Service wiIl be into review the work ofthe abstracter.

PROCEDURE B - An agent of the Public HealthService will visit your hospital about once every 3months, on a day and at a time convenient to yourMedical Record Librariai. At that time we willselect a sample of patients who have been discharged.This list of cases will be left with the Medical Rec-ord Librariar so that the face shed of the medicalrecords for these satnple patients can be r~ady faraur next visit. We plan to copy all data except thatrelating to the diagnosis ad operations. Before thetime of our next visit, some member of your staff whois familiar with the handwriting of the physicianswould be assigned by you to enter these medicofitems from the face sheet to the abstract farm. If youprefer, this operation can be performed at the time ofour visit. We will abstract the remaining data andselect a new sample. On all succeeding visits, wewill repeat the process - each visit lasting only afew haurs.

In order to be sure that the procedure is working thesame at all hospital ~ we plan to have someone visiteach hospital once or twice a year and review thesample selection and look at a few of the records thathave fallen in the sample far previous months.

PROCEDURE C - Someone assigned by you or yourMedical Record Librarian will sefect a sample ofdischarged patients accordin to instructions which

fwe will provide. Based oa al the discharges for onemonth, she will select a sample of patients and thenabstract data ftom the face sheet to the abstractfarm. This can be rbne at a time convenient to you.The completed abstmcts would be mailed to our regionaloffice each month. We will provide the packet formailing.

In order to be sure that the procedure is working thesame at al I hospitals, we plan to have someone visiteach hospital once or twice a year and review thesample selection and look at a few of the records thathave fallen in the sample far previous months.

INTERVIEWER: ShouId the Administrator objectto the procedure assigned to this hospitsl, an alter-native may be suggested, namely the priority 2 pro-cedure as shown in the “Procedure” box on the frontof the questionnaire.

49

Section B - Continued

2. The Notionol Heolth Survey offi ciols hove a uniformreimbursement fee in accordance with the manner ofabstracting data, taking into consideration the numberof records that could be abstracted per hour. For thishospital, the amount is set ati

(If item 1-A is proposed) per year.

(If item 1-B or 1-C is proposed)

per completed abstract or about

per year.

This amount will be subiect fo renegotiation if itproves to be inequitablee.

In order for reimbursement to be made by the PublicHealth Service, they require a contract, which willbe mailed to you shortly far your signature. In themeantime, I must, provide them with information anwhich to base the contract. This is in the brn ofa contract preposal.

(Complete contract proposal, Form PHS-47344, basedon the agreement reached in the discussion of theprocedures. Fill all applicable sections of theproposal, sign, and present it to the administrator forhis signature. Any problems in obtaining the signa-ture or any questions aaked about the proposal shouldbe noted below. )

= Signed m Not signed

Thank you very much --- Naw I would like, if I may,to meet with your Medical Record Libmrian.

.emarks

USCOMM-DC

50

Section C - INTERVIEW WITH MEDICAL RECORD LIBRARIAN

Title and/or form numberMTERVIEWER:

Introduce self and explain survey .as fo!fows --IS the list a completa count ofeach cloy’s discharges from - -? . . .’. . . . . n Ye. m No

Iricfly, the survey involvok the collection of a limited If ‘(No,” describe exceptions and how they are handled.

mount of infornurtion kmr the retards of a SAMPLE ofatients discharged from short-term hospitals. All, or●arly all, of the information will be available from theme sheet of the patient’s chart. This is the abstra~ma which will be used. (Show form PHS-4734-7 or 4734-U IS tho data of discharge

showrrmr the list? . . . . . . . . . . . . . . . . . . n Yes n No

NTERVIEWER:Is disch~rged patient’s medical record

If Medical Record Librarian was not present duringnumber shown on the list? . . . . . . . . . . . . n Yes a No

~aily listing, can the Ilstingthe explanation of the procedure, explak the pro- far o colendar month be mode availnble? . . . D Yes m No

cedrrre agreed upon by the administmtor.About how long IS the Iist”mtoined by the hospital?

m 1 year or more OR (No. of months)

M I said, we wi II abstract the records of only a sampleHow is the list compiled?

f patients discharged from ifre hospital. For this hospitalt will be about cases a mo~th.

.a. What medical record numbering system is used in --? Title and/or form nunrber

a Serial a Unit n Serial-unit

m Other (Describe)IS the list a complate count ofeach day’s discharges from - -? . . . . .. . . . n Yes m No

If “No,” des tribe exceptions and how they are handled.

Is the date of dischargoshown on the list? . . . . . . . . . . . . . . . . . a Yes m No

Is discharged patient’s medical racordnumber shown on the list? . . . . . . . . . . . . D Yes n No

l~this is a daily listing, can the Iittingfrs o calendar monthbe mmb availoblo? . . . n Yes a No

Aboti how long is the Ilst ratcined by the hospital?

D 1 year or more OR (No. of months)

b. Does a newborn infant rnceive the same number How is the list compiled?

as the mother?

m Yes D No (Skip to question 2)

INTERVIEWER:

1. If - - maintains more than one discharge listing,

c. Is an infant assigned a new number if he remainsin your opinion, which one should be used toselect the sampIe?

in the hospital after the mother is discharged?

n Yca D No

!. [n order to select the sample, we will need to wodrwith a list showing the medical record number of 2. Describe any supplemental listings which shouldpatients who were discharged from -- during a given be used.month. We know that these listings vary from hospitalto hospital; for example, some hospitals maintain a“daily analysis of hospital service,” while othershave a “daily admission and discharge list”. Whatkinds of discharge lists am maintained in - -?

May I 90 over them with you? 3. Attach two blank copies of all “discharge lists”(Review discharge fists and enter information reque=tedto the r;ght for each list. )

maintained by ---

FORM NHs-HDS-2 X (10-2 S-64)

51

Section C - Continued

1.a.

1.

2.

3.

4.

5.

hr addition to the usual inpatient discharges, we wantto be sure that we INCLUDE certain other classes ofpotients in the survey. Are the following included intbe list of discharges (i. e., the list to be used forsampling as described in question 2) --

Inpatients admitted,assigned a bed, anddischarged later thatday, either with orwithaut approval? . . . . . . . . Dyes i=INo

Newborn infants dis-charged after havingbeen admitted to oDediatric or special&are ward af ~he hospital -those not discharged withthe mother? . . . . . . . . . . . . OY,S ONO DNA

Inpatients whd died inthe hospital?. . . . . . . . . . . nYes ONO

A patient di schorged toanother hospital? (Some-times spoken of as a“Transfer” to anotherhospital j . . . . . . . . . . . ..nyes DNo

To be asked only if thisis a hospital complexA patient transferred fromone unit of this complexto another unit where theunits mointain separateand distinct records?. . . . . . mYes I_JNo ONA

b. If “No” to any, ask:

Is a separate listing of these cases available?

m Yes = No

(Describe the list, and/or explain the circumstances.)

La.

1.

2.

3.

4.

5.

6.

7.

There are ather classes.of patients we wish toEXCLUDE from our sample. Are any of the followingincluded in the list of discharges --

Well-newborn infantsdischarged at the sametime as the mother? . . . . . . . =Yes ONa UNl

Persons dead on arrival? . . . nYes ONO

Fetal deaths? . . . . . . . . . . nYes nNa

Outpatients? . . . . . . . . . . . UYes UNo

Patients who die in theemergency room, nothaving been admittedas inpatients? . . . . . . . . . . UYes O~[o

Emergency patients whodie in the operating roomnot having been admittedprior to death? . . . . . . . . . . ayes DNo

Patients transferred fromone service or roam toanother but not actual Iydischarged fram thehospital?. . . . . . . . . . . . . . Oyes mNo

8. Patients given a pass toleave the hospital for ashort period of time butnot actually dischargedfrom the hospital?. . . . . . . . ny~. !_JNo

b. If “Yes” to any, ask;How can they be identified in the list?(Describe)

I. Summary description of lists to be used in sampling frame

a. What discharges appear an the primary list?

b. What discharges appear on the supplemental list(s)?

US60MM-DC

52

Section D - A AND B PROCEDURE HOSPITAL SAMPLE SELECTION

NTERVIEWER: 5. About one week before our visit, we wil I telephone you

Draw the sample of discharges for calendar months July, to confirm our appointment. If you ore busy or obsent,August, and September and prepare am original sad one who should I tolk to obout this?

copy of form PHS-4734-5 as instructed in the SamplingInstructions. Also note any difficulties you encounteredapplying these instructions. Name

Leave a copy of the list of sample cases from which you Titlewill be abstracting data from the face sheets on yournest visit.

1. When would be the best time of doy for someone to come. 6.0. lit the future, would you prefer thot we work with

someone in your deportment other then yourself?and do the abstrodting?

(Time of oay~ m Yes D No}.. Do you hove o preference os to doy of the week or time

of month for our visit?I

b. (E “Yes”) With whom would you like uz to work?

(Day)Name

(Time of month)

~. INTERVIEWER: Telephone No.

Enter date and time for next visit Room No.

(Date) ,INTERVIEWER:

a.m.p.m. (Time)

L At thot time, will you be oble to hove the medicalAt B Procedure hospitals, leave with the MedicalRecord Librarian (or her representative) tbe irr-

records of the sample coses availoble ond the foce . .srructlons for abstracting medical Items; I.e.,

sheets checked for completeness? final diagnosis, operations, etc.

D Yes a No (Arrange .uitable time forvieit anti con-ad entry inquaaticm 3.)

Remarks

—

FORM NHS-HOS-2X ( 10-28-64)

53

Section E - C PROCEDURE HOSPITAL SAMPLE SELECTION

1. When would bethebest time formyvisitnextmon~

VTERVIEWER:

Proceed with the sampling, working with the Medical(Time of month)

Record Librarian. You maywant toaskthe MedicaI 2. Do you hovea preference astodayof week or —

Record Librarian to draw the sampleas you observetime of month for my visit?

and verify. Answer any questiona she might haveabout the sampling procedure. Review the abstract (Day)

form with the Medical Record Librarian and observeaa sheabsttacta a few records .Explain the procedure

(Time of month)

for transmitting completed abstrscts. Leave s copy 3. INTERVIEWER:

of the Manual for Medical Record Librarians. Explain Establish date andtime for next visit.that you will return next month to assiat in drawingthe nextsample and tohelpin sbstractingthe required (Date)

information; but, after that youwillonlyretum once a.m.

or twice a year to be sure that theprocedare is work- p.m. (Time)

ing the same in all hospitals in the survey and to 4.1fyau are busy orabsent, wlioshouldl talk —

answer any questions that she might have. to about this?

Name

Title

~emarks

iection F - SAMPLE ABSTRACTING OF MEDICAL RECORDS FACE SHEETS (A, B, and E Procedure Hospitals Onl)

Before I leave, I would like ta look at a few of the medical

records - iust the foce sheets - ta see if the information which

wv need is on the face sheet and ta see whether there will be any

problems in copying the information to the obstract form.

INTERVIEWER:

Abstract a minimum of six medical record face sheets andtransmit the abstract forms (form PHS-4734- 1 or 47342)along with the other material and two blank copies of theface sheet. If any information requested on the abstract‘form is missing from the face sheer, inquire where suchinformation might be obtained.

If the face sheet appears to contain abbreviations orcoded entries in lieu of written-out entries, obtain a listof definitions for these abbreviations and codes.

Xemarka

54

Exhibit J. Memorandum of Agreement

CONFIDENTIAL.Ihe mdic.1 record in f.rm.cim that i. .btaimd IM6U &i.“r

UI is coil.cmd under uAority of Pubk Law652c.fth E4dICO~ss(70 2tmt,48% 42U.S.C. 242c). AU i.fm-matim which would padt idcan IC.u.m of m kdivid.ml or of M e;t.btkbmmt wiff be held srticdy curdi-d.nti.1, will be used otiy bypct.mm engaged in ad for the purpo.e. of &e sucrey and will nor bedisclomd or released mother perao .x used h anyocher rmmose (22 FR 16S7).

U.S. DEPAR’IMEHTOFHEAL1’H, EDUCATION, AND WELFARS

Public Health ServiceNational Center for Health Sfxtlstics

Washington, D.C.

CONTMCT PROPOSAL FOR SERVICES IN CONNECTIONWITH SELSCTING AND COPTING HOSPITAL RECORDS

I. Service to be provided

❑ A.

❑ B.

•1 c.

❑ D.

The hospital makes available to the National Center for Health Statistics a lfsttng Of patient dischargesfrom which NCHS will eelect a sample of case records and the NCHS representative will abstract therequired data.

The hospital makes avaifable to the National Center for Health Statistics a fisting of patient dischargesfrom which NCHS will eeiect a sample of case records. The hospital abstracts the medfcal foformationand makes the case records available to NCHS for completing the abstract form.

The hospital selects a sample of patient discharges from a iisting of such discharges and abstractsrequired data for NCHS.

Other (Specify in detail)

II. Reimbursement of Cost

❑ A. Payment will be made to the hospital in the total sum of $ for fiscal year

❑ B. Payment will be made to the hospital at the rate of cente per abstracted record.

❑ C. No payment will be made to the hospital.

❑ D. Other (Specify)

III, Schedule of Payment

❑ A. Payment is to be made at the end of the fiscal year.

a B. Other (Specify)

IV. Payee will be (exact informationto appear on contract and check):

V. Hospital coordinator ofthis project will b=

SIGNATURS OF AUTHORIZEDREPRESENTATIVE SIGNATURE OF AUTHORIZEDREPRESENTATIVEOF HOSPITAL OF NCHS

Date

Comments

PHS-47W410-64

Form &+vo.cdRudget Sureau No. 6s.R6z3. R2

55

-J U. S. GOVERNMENT P372NT3NGC3FFfCE :1972 543-879/35

Series 1.

Series 2.

Series 3.

Series 4.

Sen”es 10.

Series 11.

Series 12.

Series 20.

Series 21.

Series 22.

*

OUTLINE OF REPORT SERIES FOR VITAL AND HEALTH STATISTICS

Public Health Sorvico Publication No.1OOO

Programs ad collection procedures. —Reports which describe the general programs of the NationalCenter for Health Statistics ahd its offices and divisions, data collection methods used, definitions, andother material necessary for understanding the data.

Reports number 1-4

Data evaluation and methods research .-Studies of new statistical methodology including: experimentaltests of new survey methods, etudies of vital statistics collection methods, new analytical techniques,objective evaluaticms of reliability of collected data, contributions to statistical theory.

Reports number 1-19

Analytical studies.— Reports presenting analytical or interpretive studies based on vital and health sta-tistics, carrying t,he analysis further than the expository types of reports in the other series.

Reports number 1-4

Documentq and. committee reports.— Final reports of major committees concerned with vital and healthstatistics, and documents such as recommended model vital registration laws and revised birth anddeath certificates.

Reports number 1-6

Data From the Health Interview Survey. -Statistics on illness, accidental injuries, disability, use ofhospital, medical, dental, and other services, and other health-related topics, based on &ta collected ina continuing national household intervjew survey.

Reports number 1-33

Data From the Health Examination Survey .—Statistics based on the direct examination, testing, andmeasurement of national samples of the population, including the medically defined prevalence of spe-cific diseases, and distributions of the population with respect to various physical, physiological, andpsychological measurements.

Reports number 1-18

Data From the Health Records Survey. —Statistics from records of hospital discharges and statisticsrelating to the health characteristics of persons in institutions, and on hospital, medical, nursing, andpersonal care received, based on national samples of establishments providing these services andsamples of the residents or patients.

Reports number 1-5

Data on mortality . —Various statistics on mortality other than as included in annual or monthly reports-special anal yses by cause of death, age, and other demographic variables, also geographic and tinseries analyses.

Reports number 1-4

Data on natality, marriage, and divorce. —Various statistics on natality, marriage, and divorce otherthan as included in annual or monthly reports-special analyses by demographic variable:, also geo-graphic and time series analyses, studies of fertility.

Reports number 1.9

Data From the National Natality and Mortality Surveys. —Statistics on characteristic of births anddeaths not available from the vital records, based on sample surveys stemming from these records,including such topics as mortality by socioeconomic claas, medical experience in the last year of life,characteristics of pregnancy, etc.

Reports number 1-3

/

For a list of titles of reports published in these series, write to: National Center for Health StatisticsU.S. Public Health ServiceWashington, D.C. 20201