factors influencing long-term prognosis in male · innocase wasit felt that theterminalillness...

British Heart Journala, I975, 37, I58-i65.

Factors influencing long-term prognosis in malepatients surviving a first coronary attack

Risteard Mulcahy, Noel Hickey, Ian Graham, and Gilbert McKenzieFrom the Coronary Heart Disease Research Unit, St. Vincent's Hospital, Dublin; and the Departmentsof Medicine and of Social and Preventive Medicine, University College, Dublin; and theDepartment of Social and Preventive Medicine, Queen's University, Belfast

Three hundred and sixty-four men who survived afirst episode of acute coronary insufficiency or myocardialinfarction for 28 days were admitted to a coronary heart disease secondary prevention programme between IJ7anuary I96I and 3I December 1971. Of these, 252 have been followed far at least 4 years. The 4-yearmortality was 13.5 per cent (34 patients). The average annual mortality was 3.4 per cent but an excess ofdeaths occurred during the first year offollow-up.Of iI characteristics measured during the acute attack, only severity of the attack was significantly

associated with poor 4-year survival.Cigarette consumption after infarction was significantly less among those surviving the 4-year period when

compared with decedents. Follow-up systolic and diastolic blood pressure levels were significantly lower amongdecedents. No significant differences were noted in serum cholesterol levels and in mean weight. The presenceofpost-infarction angina did not affect the prognosis.

Epidemiological research has yet to provide a de-finite risk factor profile of the coronary-proneindividual. Many factors have been incriminated tovarying degrees, including hypertension, hyper-lipidaemia, cigarette smoking, obesity, family his-tory, water hardness, and sociopsychological fac-tors (Cady, I967; Dudley, Beldin, and Johnson,I969; Rosenman et al., 1970; Chapman et al., I97I;Heady, 1973). Few controlled single or multiplefactor population prospective trials have been under-taken to quantify rigorously the hypothesis thatreduction in coronary heart disease incidence canbe achieved through risk factor intervention.

Although the pathogenetic mechanisms ofatheroma have not been identified, secondary pre-vention programmes may provide the means bywhich the prognosis of individuals with newlydiagnosed coronary heart disease can be improved.Successful rehabilitation and improved survival inpatients with coronary heart disease are the twinobjectives of secondary prevention. In addition, aframework is established within which the clinicianmay test the efficacy of various treatment regimenson survival.

Received 4 July 1974.

AimsThe object of this report is to record the survivalexperience of our male patients who have survived afirst coronary attack and to identify initial and long-term determinants of prognosis among them. Themethods of our follow-up study have been describedpreviously (Mulcahy and Hickey, I972).

Subjects and methodsThree-hundred and sixty-four consecutive male patientsunder 6o years who survived a first attack of acute cor-onary insufficiency or myocardial infarction for at least28 days were admitted to a rehabilitation and secondaryprevention programme between I January I961 and3I December 1971. Of these, 259 entered the study at orbefore 3I December I968 and were, therefore, theo-retically eligible for inclusion in this 4-year follow-upstudy. The 259 initial and fourth year follow-up re-cords were extracted from the secondary preventionfollow-up file. At the time of writing the follow-upstatus of 3 cases was indeterminable. A further 4 caseswere excluded from the analysis as 'outliers" (Cox andSnell, i968), leaving in total, 252 patients - the index

'In these 4 cases a number of the biochemical measurementsdeparted so far from the mean that their inclusion in theanalysis would have seriously biased the results.

on June 20, 2021 by guest. Protected by copyright.

http://heart.bmj.com

/B

r Heart J: first published as 10.1136/hrt.37.2.158 on 1 F

ebruary 1975. Dow

nloaded from

http://heart.bmj.com/

Prognosis of coronary heart disease I59

cases - satisfying the medical and statistical criteriafor inclusion in the analysis.Although nominally the follow-up period is 4 years,

operationally some variation was experienced. Theaverage follow-up period for the 2I8 survivors was48.89 (±5 months) and 8o per cent were followed upbetween 42 and 54 months. The remaining 20 per centwere all examined between 36 and 6o months. The sur-vivorship distribution of the 34 deaths is discussedseparately in the results, section A. The term 4-yearfollow-up refers to an average follow-up period of 48.89months.

Last follow-up was taken as the last outpatient visit.In the decedent group, last follow-up examination tookplace some time before death. In no case was it felt thatthe terminal illness influenced the last follow-up findingsappreciably.The 252 index cases were classified into 3 severity

groups as follows:a) Acute coronary insufficiency, 66 (26.2%)b) Myocardial infarction, 140 (55.6%)c) Myocardial infarction (complicated), 46 (I8.2%)

Patients with acute coronary insufficiency had typicalcardiac pain and serial electrocardiographic abnormali-ties without enzyme changes or other indications ofheart muscle necrosis (Mulcahy and Hickey, I966).

Patients with myocardial infarction had typical cardiacpain with characteristic electrocardiographic abnor-malities of infarction and/or enzyme changes (Mulcahyand Hickey, I966). They were designated complicated if,during the acute stage, one or more of the followingconditions presented: i) supraventricular tachyarrhyth-mias; 2) ventricular tachycardia; 3) second-degree orcomplete heart block; 4) prolonged hypotension; 5)cardiogenic shock; 6) cardiac failure (frank left or rightventricular failure or pulmonary congestion in the x-ray);7) pericarditis, with or without post-infarction syndrome;8) thromboembolism, including systemic as well aspulmonary embolism; 9) Bundle-branch block, tran-sient or persistent, with QRS complexes of O.I2S ormore.

Additional complications are now included in thiscontinuing study but they were not recorded before thedevelopment of coronary care facilities.Our report concentrates on the measurements, height,

weight, systolic and diastolic blood pressure, serumcholesterol, number of cigarettes smoked per day, life-time smoking or cigarette index (Mulcahy, Hickey, andMcDonald, I966), anginal status, age, family history,and severity of initial myocardial infarction. The sameinformation was available at fourth year follow-up or, inthe case of decedents, at last follow-up.

ResultsA) Length of survival of decedentsThirty-four index cases (I3.5%) had died and 2I8(86.5%) were alive at the end of the 4-year follow-up period. Table i shows the observed survivaltime distribution for the decedent group. Eight

patients (23.5%) died within 6 months of initialinfarction, and I4 (4I%) were dead within i year.The hypothesis of equal risk of death in each yearafter initial attack was easily rejected (X2 = I0.87,d.f. = 3, P o.o5), though an aberrantsampling fluctuation in the '25 to 31 months cate-gory' contributed substantially to the excess mor-tality observed in the tail. Nevertheless, it is con-cluded that this statistical model provides a reason-able basis upon which to make inferences about thesurvival distribution of decedents. In simple terms,this statistical technique confirms that the mortalityrate after a first coronary attack is not constant overthe follow-up period, and that mortality experiencemay tend to improve with length of survival.

B) Comparison of initial characteristics insurvivors and decedentsTable 2 shows, for both groups, the mean valuesobtained for eight continuously distributed meas-urements made at initial examination. No statis-tically significant differences were observed, and itis concluded that taken individually these variablesare not important determinants of 4 years' survival.The distribution of anginal status was examined.

Of those with angina before their illness, I9 per centdied, while II.6 per cent of those without anginadied. The 7.4 per cent difference was not statis-tically significant at the 5 per cent level (X2 = 2.22,d.f. = i, P >0.05). Similarly, no significant differ-ence was observed in the death rates of each of thefour categories (negative I2.9%, doubtful II.6%,positive 2I.I%, strongly positive I4.8%) of familyhistory (X2=2.12, d.f.=4, P>o.o5).

Table 3 shows the distribution of severity of theinitial attack in both survivors and decedents.Patients experiencing a complicated myocardial in-farction had a worse prognosis, I4 (30.4%) beingdead at fourth year follow-up compared with i6(II.4%) in the uncomplicated myocardial infarc-tion group and 4 (6.i%) in the acute coronary in-sufficiency group (X2 = 14.95, d.f. = 2, P

I6o Mulcahy, Hickey, Graham, and McKenzie

TABLE I Distribution of survival-times in 34patients who died within 4 years of entry to study

Time Observed % Expected* %from entry frequency frequencyto study(mth)

I- 8 23.5 8.32 24.57- 6 I7.6 7.51 22.113- 2 5.9 5.50 i6.2I9- 3 8.8 4.02 ii.825- 7 20.6 2.94 8.73I- 3 8.8 2.I5 6.337- 3 8.8 I.57 4.643- I 2.9 L.I5 3.449- I 2.9 o.85 2.5Total 34 (I00) 34 (Ioo)

A= 0.052(I o.oo85) - see appendix.* Derived from the doubly truncated exponential law (seeAppendix).

comparison of the average discriminant scoresobtained failed to reach statistical significance(X2 = 24.56, d.f. = I5, P > 0.05). It is considered that,of the intial factors, only severity of the initialattack shows a statistically significant relation withsurvivorship status at 4 years.

C) Comparison of foliow-up characteristicsin survivors and decedentsTable 4 indicates significant mean differences be-tween survivors and decedents with respect tosystolic blood pressure, diastolic blood pressure, andcigarette smoking. The mean values of both bloodpressure measurements are raised in survivors and

decedents at last follow-up but the rise was signi-ficantly greater in the survivors.On average, survivors were smoking 8 cigarettes a

day less than decedents (P

Prognosis of coronary heart disease I6I

cigarettes a day). Statistically significant increaseswere recorded in both blood pressure measurements.From the last two columns ofTable 5 it is apparent

that the average increase in blood pressure measure-ments in the survivors was significantly greater thanthe corresponding increase in the decedents. Nostatistically significant difference was found onaverage in the amount by which both survivors and

TABLE 3 Distribution of severity of initial attackin both survivorship groups

Survivorship groupSeverity Dead Alive Total

Acute coronaryinsufficiency 4 (6.i) 62 (93-9) 66 (IOO)

Myocardial infarction(uncomplicated) i6 (II.4) 124 (88.6) I40 (IOO)

Myocardial infarction(complicated) I4 (30.4) 32 (69.6) 46 (Ioo)

Total 34 (I3.5) 2i8 (86.5) 252 (IOO)

x2= I4.95, d.f. = 2, P < 0.05.

non-survivors reduced their serum cholesterol andweight. However, the survivors did significantlyreduce their cigarette consumption by I6.44 cig-arettes a day compared with a reduction of 9.i8cigarettes a day in the non-survivors.

Anginal status was next examined. Of the 79patients with angina at last follow-up, I5.2 per cent(I2 cases) were dead. The rates for comparison areI9.0 per cent (initial examination) and I5.2 percent (follow-up examination). A non-significantresult is obtained and no important increase in theproportion of deaths was observed in those patientsexperiencing angina at follow-up.The variables in Table 5 and anginal status were

used in another discrimination function analysis.Bartlett's test revealed a statistically significantdifference (X2 = 25.76, d.f. =6, P

x62 Mulcahy, Hickey, Graham, and McKenzie

DiscussionPell and D'Alonzo (I964) reviewed I0 studies re-porting long-term survival after acute coronarydisease. All these studies differed in such importantaspects as diagnostic criteria, sex and age distri-bution of patients, and follow-up methodology.The 5-year survivorship varied from 49 to 83 percent.

In a more recent study (Geismar et al., I973) a 5-year survivorship of 56 per cent among 642 patientswith coronary heart disease who survived to leavehospital was reported. In a study of 598 men whosurvived an infarction for 4 weeks, Zukel et al.(i969) reported a 5-year survivorship of 8o per centwhile the HIP study (Weinblatt et al., i968) showedan 8i per cent survivorship over 4- years amongsimilar patients.Our study reports an 86.5 per cent survivorship

over 4 years in men under 6o years with first myo-cardial infarction or acute coronary insufficiency.Our results are similar to those of Zukel et al. (I969)and the HIP group, which, perhaps more than anyother studies, resemble our own in relation to agedistribution, diagnostic criteria, and time of startof follow-up.About 40 per cent of our decedents died during

the first year of follow-up. A higher death rate inthe first year has been reported by Weinblatt et al.(1968) and this fact may have important impli-cations from the secondary prevention point ofview. It is at least worth speculating that, if secon-dary prevention measures are effective, they maynot affect mortality for some time after they areinstituted. In such an event one might find that thebeneficial influences of these measures might not bedetected until some time after the heart attack.To achieve valid results, a controlled randomized

study ofthe effects ofsecondary prevention measureson survival in patients with coronary heart diseasewould be required, but for practical and indeedethical reasons such a programme is virtually im-possible to organize. However, by examining sur-vival curves among particular groups within theentire cohort it should be possible to detect theeffect of various initial and follow-up attributes onlong-term survival. It might be feasible to establishprognostic criteria for patients according to thenature of the coronary attack and to their attributesat the time of the attack. Similar criteria might alsobe established based on risk factor changes in re-sponse to a secondary prevention programmeoccurring during the follow-up period.We have shown that, of ii initial factors studied,

only the severity of the initial attack influencedlong-term prognosis. Unlike the HIP study (Franket al., I968) we did not find that the initial systolic

or diastolic blood pressure status influenced long-term survival.

Increasing age was associated with increasingmortality over 4 years in our group, but the associa-tion was not significant. In a study of patients under40 years with coronary heart disease a worse long-term prognosis was noted in the younger survivors(Gertler et al., I964), while in the HIP study (Franket al., I968) the mortality was higher in olderpatients. In the seven countries study (Keys, I970)age had no obvious effect on survivorship. It may bethat the older the age group the more likely one is tofind that age is positively associated with a lessfavourable long-term prognosis.The presence of angina before the initial attack

had no significant influence on prognosis in ourexperience though the HIP study (Weinblatt et al.,I968) showed angina to have an adverse effect onlongterm mortality.As in the HIP study (Weinblatt et al., I968) we

found no relation between current and lifetimesmoking noted at initial attack and 4-year survival.Initial cholesterol figures bore no relation to long-term prognosis, a finding that supports those ofother workers (Keys, I970; Shanoff, Little, andCsima, 1970).The presence of complications such as left ven-

tricular failure and prolonged hypotension remainsthe only detectable initial determinant of prognosisin these patients. This finding has been reported inthe past and, in general, investigators have reportedpoor long-term survival experience among thosewith cardiac failure, dyskinesis, life-threateningarrhythmias, and cardiomegaly during the acutestage (Honey and Truelove, I957; Pell andD'Alonzo, I964; Beard et al., I967; Weinblatt etal., I968; Norris et al., I973). In addition, someworkers (Pell and D'Alonzo, I964; Sievers, I964)find that the adverse effect of complications duringthe acute stage on long-term survival is apparentonly for the initial I tO 2 years of survival, and thatpatients with initially complicated myocardial in-farcts who survive this early period appear to have aprognosis no worse than that of those who survivean uncomplicated myocardial infarct.We have recorded the nature of the different

complications during the acute stage in all ourpatients. The number of patients who have diedduring follow-up is not large enough yet to identifythe individual complications which may be the bestdeterminants of poor prognosis, but preliminaryexamtion of our data suggests that frank pul-monary oedema during the acute stage is a highlysignificant pointer to high subsequent mortality.

In the later stages of this study since the adventof coronary care facilities we have been able to



/B


ebruary 1975. Dow

nloaded from



record a greater variety of complications during theacute stage, but the influence of these recentlyrecorded complications must await further follow-up.

Very little information is available about theeffect of alteration of risk factors on survival inpatients with myocardial infarct or acute coronaryinsufficiency. We have shown that our decedentsand survivors had significantly reduced cigarettesmoking, but the reduction in smoking experiencewas significantly greater in survivors compared todecedents. As far as we know this favourable ex-perience among those who reduce their smokingconsiderably has not been previously reported.There was a small but significant reduction in

cholesterol levels in our patients but, while chol-esterol reduction was greater among survivors, itwas not significantly so compared to decedents.

Appropriate dietary or drug treatment will lowerserum cholesterol in hypercholesterolaemic subjects.However, whether such a reduction will influencesubsequent coronary heart disease mortality has yetto be decided. A study by Miettinen et al. (I972)suggests that cholesterol reduction is effective inreducing mortality in a primary prevention trial,and a few secondary prevention trials (Morrison,I960; Leren, I967; Bierenbaum et al., I973) sup-port this finding. However, the methods of thesesecondary prevention trials have been criticized.Among other factors, weight as a variable was notcontrolled and the treatment groups describedshowed a reduction in weight while the controlgroup did not.

In 3 other secondary prevention trials (Ball et al.,I965; Rose, Thomson, and Williams, I965; Med-ical Research Council, I968) the effect of obesity as a

variable was reduced or eliminated in both controland treatment groups. None of these trials showed asignificant difference in mortality in treated and un-treated patients, despite a reduction in serumcholesterol in the treated group.There is as yet no evidence that control of high

blood pressure in survivors with coronary heartdisease has any beneficial effect in preventing furthercoronary attack. Survivors and decedents showedsimilar mean systolic and diastolic blood pressurelevels during the initial hospitalization. Both groupsshowed similar rises in blood pressure at firstfollow-up examination (Table 6). At last follow-upexamination, however, while the blood pressurefigures of the survivors were sustained, those of thedecedents showed a significant fall towards thelower levels noted in hospital during the initialillness (Table 6).The rise in blood pressure in both groups at first

follow-up examination can be attributed to theknown fact that inpatient blood pressure figures aresignificantly lower than casual outpatient figures.However, the subsequent fall in blood pressureamong decedents cannot easily be explained. Thesepatients may have been successfully treated forhypertension or they may have sustained furthernon-fatal myocardial infarcts to account for thesefindings. The significant fall in blood pressuresduring follow-up decedents at least merits furtherstudy and suggests the need for a careful evaluationof the indications for treatment of hypertension inpatients with coronary heart disease.

Just as anginal status before the attack is a poordeterminant of long-term prognosis, so is theanginal status of patients after the attack. Whilethose with residual angina fared less well, the

TABLE 6 Intersurvivorship group comparisons of mean blood pressure (systolic and diastolic) levels at inzitial,first follow-up, andfourth year follow-up examination*

Survivorship Blood pressure (mmHg) (kPa)group (A) (B) (C)

Initial examination Ist follow-up examination 4th year examination

(Systolic I26.43 ( I.2I) I46.89 ( i.o8) I48.07 (+.45)Alive J .6.8I (o0.I6) '9.54 ( 0.14) 19.69 ( O.I9)DiastolUc 79.69 (±0.70) 91-53 (±0.52) 91.22 (±0.74)

L I0.59 ( 0.09) 12.17 (± 0.07) 12.13 (± 0.09)Systolic I28.03 (±3.2I) I45.56 (±2.27) I33.17 (±3.30)t

Dead 17.03 (±0.43) 19.36 (± 0.30) 17.71 (± 0.44)Diastolic 8I.26 (± I.95) 94.13 (± 1.59) 85.97 (± 2.34)tIo.8o (±0.26) 12.51 (+ 0.2I) I1.43 (± 0.3)

* Standard errors in parentheses.t Last follow-up.Blood pressures in pascal units (kPa) are given in italics.



/B


ebruary 1975. Dow

nloaded from


x64 Mulcahy, Hickey, Graham, and McKenzie

mortality rate in those with and without angina wasnot significantly different. There is no informationavailable from other sources about the influence ofresidual angina on prognosis.

We are grateful to Professor Peter Frogatt for his as-sistance.

This work is supported by The Medical ResearchCouncil of Ireland and The Irish Heart Foundation.

ReferencesBall, K. P., Hanington, E., McAllen, P. M., Pilkington,

T. R. E., Richards, J. M., Sharland, D. E., Sowry,G. S. C., Wilkinson, P., Clarke, J. A. C., Murland, C.,and Wood, J. (I965). Low-fat diet in myocardial infarc-tion. Lancet, 2, 50I.

Beard, 0. W., Hipp, H. R., Robins, M., and Verzolini, V. R.(1967). Initial myocardial infarction among veterans: ten-year survival. American Heart Journal, 73, 3I7.

Bierenbaum, M. L., Fleischman, A. I., Raichelson, R. I.,Hayton, T., and Watson, P. B. (I973). Ten-year exper-ience of modified-fat diets on younger men with coronaryheart-disease. Lancet, I, I404.

Cady, L. D. (I967). Epidemiology of coronary heart disease.American Journal of Cardiology, 20, 692.

Chapman, J. M., Coulson, A. H., Clark, V. A., and Borun,E. R. (I97i). The differential effect of serum cholesterol,blood pressure and weight on the incidence of myocardialinfarction and angina pectoris. Journalof Chronic Diseases,23, 631.

Cox, D. R., and Snell, E. J. (I968). A general definition ofresiduals (with discussion). Journal of theRoyal StatisticalSociety, Series B, 30, 248.

Dudley, E. F., Beldin, R. A., and Johnson, B. C. (I969)Climate, water hardness and coronary heart disease.J'ournal of Chronic Diseases, 22, 25.

Fisher, R. A. (I938). The statistical utilization of multiplemeasurements. Annals of Eugenics, 8, 376.

Frank, C. W., Weinblatt, E., Shapiro, S., and Sager, R. C.(I968). Prognosis of men with coronary heart disease asrelated to blood pressure. Circulation, 38, 432.

Geismar, P., Iversen, E., Mosbech, J., and Deyev, K. (I973).Longterm survival after myocardial infarction: a nationalfollow-up study of 642 patients in Denmark. InternationalJournal of Epidemiology, 2, 257.

Gertler, M. M., White, P. D., Simon, R., and Gottsch, L. G.(I964). Longterm follow-up study of young coronarypatients. American Journal of the Medical Sciences, 247,I45.

Heady, J. A. (I973). A cooperative trial on the primary pre-vention of ischaemic heart disease using clofibrate:design, methods, and progress. Bulletin of the WorldHealth Organization, 48, 243.

Honey, G. E., and Truelove, S. C. (1957). Prognostic factorsin myocardial infarction. Lancet, I, 1209.

Keys, A. (I970). (Editor.) Coronary Heart Disease in SevenCountries. American Heart Association, Monograph No 29.New York.

Leren, P. (1967). The effect of plasma cholesterol loweringdiet in male survivors of myocardial infarction. A con-trolled clinical trial. Acta Medica Scandinavica, Suppl.466.

Medical Research Council, Report of a Research Committee(I968). Controlled trial of soya-bean oil in myocardialinfarction. Lancet, 2, 693.

Miettinen, M., Turpeinen, 0., Karvonen, M. J., Elosuo, R.,and Paavilaihen, E. (1972). Effect of cholesterol-loweringdiet on mortality from coronary heart-disease and othercauses. Lancet, 2, 835.

Morrison, L. M. (I960). Diet in coronary atherosclerosis.Journal of the American Medical Association, 173, 884.

Mulcahy, R., and Hickey, N. (I966). Cigarette smokinghabits of patients with coronary heart disease. BritishHeart_Journal, 28, 404.

Mulcahy, R., and Hickey, N. (I972). Coronary Heart DiseaseRehabilitation and Secondary Prevention Programme: AResearch Study. St. Vincent's Hospital, Dublin.

Mulcahy, R., Hickey, N., and McDonald, M. (I966). A noteon a quantitative method of recording cigarette smokingexperience. IrishJournal of Medical Science, No. 487, 30I.

Norris, R. M., Caughey, D. E., Deeming, L. W., Mercer, C. J.,and Scott, P. J. (I973). Prognosis following acute myo-cardial infarction. New Zealand MedicalJournal, 77, 12.

Pell, S., and D'Alonzo, C. A. (I964). Immediate mortalityand five-year survival of employed men with a first myo-cardial infarction. New England Journal of Medicine, 270,9I5.

Rose, G. A., Thomson, W. B., and Williams, R. T. (I965).Corn oil in treatment of ischaemic heart disease. BritishMedicalJournal, I, I531.

Rosenman, R. H., Friedman, M., Straus, R., Jenkins, C. D.,Zyzanski, S. J., and Wurm, M. (I970). Coronary heartdisease in the Western collaborative group study. J3ournalof Chronic Diseases, 23, I73.

Shanoff, H. M., Little, J. A., and Csima, A. (I970). Studies ofmale survivors of myocardial infarction. XII. Relation ofserum lipids and lipoproteins to survival over a io-yearperiod. Canadian Medical Association Journal, 103, 927.

Sievers, J. (I964). Myocardial infarction. Clinical features andoutcome in three thousand thirty-six cases. Acta MedicaScandinavica, 175, Suppl. 406.

Weinblatt, E., Shapiro, S., Frank, C. W., and Sager, R. V.(I968). Prognosis of men after first myocardial infarction:mortality and first recurrence in relation to selected para-meters. American Journal of Public Health, 58, I329.

Zukel, W. J., Cohen, B. M., Mattingly, T. W., and Hrubec,Z. (I969). Survival following first diagnosis of coronaryheart disease. American Heart Journal, 78, 159.

Requests for reprints to Dr. Risteard Mulcahy, Cor-onary Heart Disease Research Unit, St. Vincent's Hos-pital, Elm Park, Dublin 4.

AppendixThe probability density fumction of the doublytruncated exponential distribution is defined by:

P(T = t/A) = A.exp (-A-t)0 (A) (I)For A greater than zero and t lying between theconstants Ci and C2 (C2 > Ci) and where the de-nominator is given by:

(A)=exp (-A-Ci)-exp (-A'C2)



/B


ebruary 1975. Dow

nloaded from



The method of maximum likelihood was em- (T A)Aexp (-At)ployed to estimate A and yielded, in these data R(tR, A)(Table i), a value of 0.I52I (± o.oo85) when Ci and where R (t, A) = exp (- At) - cap (- AC2). Equa-C2 are one and 54 months, respectively. tion (2) is a non-decreasing function of time, rising

Corresponding to equation (i) the hazard or age- from approximately A (when t-GCi) to infinityspecific death rate is given by: (when t-*C2).



/B


ebruary 1975. Dow

nloaded from

factors influencing long-term prognosis in male · innocase wasit felt that theterminalillness...

Documents