Imagine you're a manager in one ofManitoba's rural or northern RegionalHealth Authorities. You receive datashowing that in one or more areas in yourRHA, when residents are hospitalized,over 90% of them don't go to their localhospital. What do you do? Downsize thehospital? Expand its services? Close it?Convert it to another kind of health carefacility that better suits local needs?

Well, most likely you'd look at the mat-ter much more closely. But the point is,that's the kind of information managersin rural and northern RHAs are lookingfor, a starting point from which they canbegin tailoring health care services tomeet the needs of their residents. It's thekind of information the Manitoba Centrefor Health Policy and Evaluation providesin this report on rural hospitals.

Manitoba Health asked MCHPE todevelop a method of assessing rural hospi-tal performance. With over two billiondollars budgeted for Manitoba's healthcare system yearly, politicians, health caremanagers and the public increasinglywant information about health care oper-ations—hospitals in particular. Our reportoffers Manitobans a preliminary look athow rural hospitals function in thisprovince.

To RHAs it offers answers to questionssuch as: Are hospitals meeting the hospi-talization needs of their region? Howacute is the care they provide? Whichareas have the greatest need for hospitalservices? Are some areas using more hos-pital services than expected? Are hospitals

discharging their patients efficiently? Arehospital beds full?

Answering these questions presentedquite a challenge.

Challenges and New ApproachesVirtually none of the existing literature onhospital performance assessment was use-ful for our analysis of Manitoba's ruraland northern hospitals. Most such previ-ous studies focussed on urban or teachinghospitals. Their applicability to rural hos-pitals is limited: smaller hospitals general-ly deal with less severe cases, have lowervolumes, and a much higher proportionof non-specialist physicians and nurses.They also have different economies ofscale: small hospitals might be far moreefficient than a large urban hospital atdealing with, say, uncomplicated pneumo-nia, but less capable of treating severeheart failure.

So this work, with its focus solely onrural and northern hospitals, appears tobe the first of its kind. We developed a setof indicators tailored especially for ruralhospitals that cover multiple aspects oftheir performance.

A key feature of our method is theinclusion of a population-based perspec-tive. We look at how much hospital carepopulations are using. This is thenweighed against a population's need forhospital services, arrived at using popula-tion characteristics that are clearly associ-ated with the need for health care—likeage, gender, socioeconomic status andpremature mortality rates. For this study,

Assessing the Performance of Rural and Northern HospitalsTHE MANITOBA CENTRE FOR HEALTH POLICY AND EVALUATION

JULY2000

we looked at physician service areas (PSAs)from rural and northern RHAs in Manitoba,and ranked all 51 by their need for health care services.

In addition, we developed three indicatorsspecifically for this report. Intensity of serviceslooks at the proportion of cases involvingsurgery or deliveries, the complexity of typicalmedical cases, and the proportion of cases thathad a length of stay greater than one day.Another measure, discharge efficiency, calcu-lates how long a hospital's stays should be,given the type of cases it treats, then comparesthem to how long stays actually are. A thirdindicator looks at the share each institutionhad of local hospitalizations. In other words,for every 100 people hospitalized in a particu-lar community, how many of them stayed intheir local hospital?

We also included a fourth measure commonin hospital assessments, occupancy rates.

Each hospital was ranked (a) in relation tothe other hospitals in rural Manitoba and (b) in relation to hospitals of similar size andfunction—major rural, intermediate rural,small rural, small multi-use and northern iso-lated. And since many hospitals in ruralManitoba handle a low volume of cases, webased our performance analysis on a three-yearaverage (fiscal years 96/97-98/99) rather than a single year.

FindingsOur analysis of hospital performance benefittedimmensely from the division of rural hospitalsinto five different categories. Since hospitalswere compared to other institutions of similarsize and function, it allowed for fairer andtherefore more meaningful comparisons to bedrawn. This was especially helpful in compar-ing usage rates and intensity of services

p There are a lot of empty beds in Manitoba'srural hospitals. Occupancy rates across the 68hospitals we studied averaged less than 60%.Twenty hospitals had rates below 50%. Onlyone rural hospital reported an occupancy rateabove 80%.

p There are important differences in the ratesat which rural residents use their local hospi-

tals. Only 11 of 68 hospitals provided morethan 50% of the hospitalizations their area residents used. On the other hand, 18 hospitalsprovided less than 10% of area residents' hos-pitalizations, meaning over 90% were hospital-ized elsewhere.

p Residents of about one PSA in five used hos-pital services at a rate more than 10% higherthan their "need" profile suggested.

p Usage rates are related to hospital type. Few local residents used small rural and smallmulti-use hospitals (Fig. 1). In contrast, north-ern isolated hospitals had above average usagerates, demonstrating how dependent northernresidents are on them. Major rural hospitalshad the largest share of local use.

p Intensity of services is closely related to hos-pital type. Virtually all of the hospitals scoringin the high range of our intensity scale weremajor rural or intermediate rural hospitals.The lowest intensity scores were recorded bysmall rural and northern isolated hospitals.

p After adjusting for differences in hospitaltype, no one type of hospital appeared less

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

BEN

ITO

REST

ON

ROSS

BU

RN

MA

NIT

OU

MA

CGRE

GO

R

WH

ITEM

OU

TH

BEN

ITO

REST

ON

ROSS

BU

RN

MA

NIT

OU

MA

CGRE

GO

R

WH

ITEM

OU

TH

Occupancy Rates Share of Hospitalizations

1. Occupancy vs. Share of Hospitalizations at Small Multi-Use Hospitals: 96/97-98/99

efficient than another. However, within eachhospital grouping—such as intermediaterural—some hospitals were more efficient indischarging patients than others.

p Discharge efficiency is unrelated to theintensity of services the hospital provides.

p Occupancy rates can be a misleading indica-tor for judging hospital performance. Manyhospitals that recorded high occupancy ratesalso had low intensity scores and/or poor dis-charge efficiency.

Interpreting the DataFirst of all, we must emphasize that while thisassessment is in many respects a report card,it is in no way an action plan. It gives RHAsthe lowdown on how their hospitals are per-forming and how much use area residents aremaking of them. It does not offer specific rec-ommendations on what to do with, say, a hos-pital that ranks low on discharge efficiency andlocal usage. Such decisions are beyond thescope of this report and are better left up toeach RHA with its local perspective on areaneeds. What this study offers is a first steptoward making those decisions.

That being said, it is clear that the numberand function of Manitoba's rural hospitalsneeds to be reassessed. And we can illustratehere some possible ways RHAs might use thisstudy to that end. At the same time we'll pointout some possible conclusions and potentialmis-conclusions that might be drawn from anyor all of the indicators used in this hospitalassessment.

For instance, hospitals performing well on avariety of indicators might be used as bench-marks, a standard to aim for in trying toimprove all rural hospitals. If an institutioncombines high intensity services with gooddischarge efficiency and high occupancy rates,one can assume they are doing a lot of thingsright. Even more so if their level of use corre-sponds to the community's need. To RHA managers such facilities may provide a model:"What are they doing that our other hospitalsare not?"

On the other side of the coin, hospitals thataren't performing well will draw a great deal ofattention. In improving the system, these arethe likely targets of change. This is where it is

particularly important to consider not just oneindicator, but all indicators, to try to get thefull story on how a hospital is functioning.

For example, one might look at northernisolated hospitals, which have 23% occupancyrates—much lower than the other types ofhospitals—and conclude there is little need forthem. But they also account for more thanone-third of hospitalizations of local area resi-dents, a usage rate much higher than many"fuller" rural hospitals to the south (Fig. 1).The most obvious reason for this is that thenext hospital is hundreds of kilometres away(Fig. 2). Whatever the reason, it's clear thatresidents depend on these hospitals; low occu-pancy rates don't tell the whole story.

2. Distances Between Rural & Northern Hospitals

KILOMETRES

25 5075100

WANT THE COMPLETE REPORT? YOU CAN DOWNLOAD IT FROM OUR WEB SITE: www.umanitoba.ca/centres/mchpe

OR CONTACT MCHPE: PH. (204) 789-3805; FAX (204) 789-3910S101-750 Bannatyne Avenue, Winnipeg, Manitoba, Canada R3E 0W3.

High occupancy rates can be equally mis-leading. Many hospitals that reported highoccupancy rates received low scores on the dis-charge efficiency and intensity measures. Theirhigh occupancy rates may, compared to otherhospitals, reflect a practice of admittingpatients who are not as sick, or perhaps keep-ing their patients longer.

Even with two or more indicators, it's stillpossible to jump to a wrong conclusion. Atfirst glance a hospital with unusually lowscores might be seen as simply inefficient.However, given the comparatively low numberof patients some rural hospitals treat each year,it could also be the result of just one or twounusually long stays in one year. Or, it couldreflect an unusual event like the departure of aphysician.

So it's worth repeating: hospitals should notbe assessed on the basis of a single indicator,several need to be taken into account. A poorscore on one or two indicators signals that fur-ther investigation is required.

This seems especially true when looking atshares of hospitalizations. True, if in a givenarea, 90% of the hospitalizations don't takeplace in the local hospital, it strongly suggeststhat the local hospital isn't really needed by thecommunity. But does that mean the hospitalisn't performing well? Maybe. That it should beclosed? Maybe. It could just as likely mean thatthe needs of the community have changed, andtherefore so should the role of the hospital.

For example, in another MCHPE report,Alternatives to Acute Care (1996), we foundthat in some cases, a large proportion ofpatients using small rural hospitals were long-stay elderly patients. Instead of residing in longterm care facilities (which may not have beenavailable), they were having extended stays ashospital patients. The hospitals were in factserving as nursing homes. So here is a situa-tion where an RHA might consider changingthe role of the facility—from hospital to nurs-ing home—to make it better serve the appar-ent needs of the community.

Many other factors can influence a commu-nity's need for its hospital. Even something

like a new highway. Perhaps a once-busy smallrural hospital now sees residents "voting withtheir feet" to bypass it because a much largerfacility is suddenly just as easy to get to. Thisin turn might snowball: the local hospital isunder-used; routine procedures aren't so "rou-tine" anymore; residents' confidence in thehospital erodes; which leads to even less use;and so on.

This is where the RHAs might start to con-sider "constellations of hospitals." If a hospitalhas become redundant with another not faraway, then maybe it should be converted tosomething else. Here is where local input onwhat makes sense is vital. Perhaps a communi-ty health care centre with more primary caredoctors or nurses would better serve. Or per-haps an emergency centre. Instead of threesmall hospitals each offering the same limitedservices, they can, as a constellation of hospi-tals or health care facilities, offer a broaderrange of complementary services.

Of course it may well be that some under-used and low scoring hospitals will be targetedfor closing. And doubtless, concerns will beraised. A look at Saskatchewan's experienceshould allay some of those concerns. There,they closed over 50 small rural hospitals in theearly nineties. Follow-up studies in theseregions showed that health, as indicated bymortality rates, actually improved after hospi-tal closures—more so in these areas than else-where in the province. So, the closing of smalllocal hospitals didn't imperil the lives of resi-dents. In fact, their life expectancy increased.

Whatever decisions RHAs make about thefuture of their rural hospitals, they now have atool that can help them. This assessment pullsinto focus a much larger picture of how eachhospital is performing than was previously pos-sible. Coupled with their local perspective, it'sup to each RHA to take it from there.

Summary by RJ Currie based on the Report:Assessing the Performance of Rural andNorthern Hospitals in Manitoba: A First Lookby David Stewart, Charlyn Black, PatriciaMartens, Sandra Peterson and David Friesen