British Heart Journal, I972, 34, 897-goo. Br Heart J: first published as 10.1136/hrt.34.9.897 on 1 September 1972. Downloaded from

The cost-benefit analysis of a coronary care unit

P. C. Reynell and M. C. Reynell From the Bradford Royal Infirmary

A five-bedded coronary care unit located in a general medical ward in a district general has been subjected to a cost-benefit analysis and the results expressed in costper life saved, assuming that the only benefit of the unit was from cardiac arrest with survival to leave hos- pital. Direct costs were £320 per life saved. Indirect costs were also incurred because the existence of the unit resulted in the admission ofpatients who would otherwise have been treated at home, though this was mitigated by a reduction in mean length of stay during thefirstfive years of the p unit's existence. Net indirect costs were £220 per life saved, giving a total cost of £540 per life saved. It is suggested that this kind of outcome analysis by survival could be used to assess the per- formance of other types of coronary care unit and of other forms of expensive medical treatment.

In comparing one type of medical care with Salaries In most coronary care units another and assessing priorities within the salaries are by far the largest single item of Health Service, it is desirable to have some cost. Doctors and nurses are resources which to are not only expensive but in short supply, measure of cost-effectiveness. Attempts copyright. apply the techniques of cost-benefit analysis and this was the main reason which prompted to the medical services may fail because of the us to establish a unit within a general medical difficulty of expressing benefits in financial ward. In theory no extra doctors, nurses, or terms. There are, however, certain situations technicians were employed to staff the unit, in in which it is possible to analyse but in practice the nursing complement was outcome in terms of lives saved, and the usually a little higher than that of a compar-

$coronary care unit is one of these. able male medical ward without a coronary http://heart.bmj.com/ We have therefore carried out a cost-benefit care unit. Over the year the average difference analysis of the work of a coronary care unit was approximately one staff nurse during the using cost per life saved as a measure of day and one second-year nurse at night with effectiveness. replacements for sickness and holidays. We The five-bedded unit sited in a general have taken this into account in assessing costs. medical ward of a district general hospital has The coronary care unit undoubtedly involved been described previously (Reynell, I969). No an increase in work-load on the ward, and we

major structural alterations were required. believe that this increase amounted to more on September 26, 2021 by guest. Protected * The unit was designed to economize as far as than the working time of one nurse. This possible in the use of scarce nursing staff and could mean that other in the ward does not normally admit patients over 65 years were relatively neglected and it might imply old. Patients normally spend three days on the a hidden deficit on the benefit side of the the coronary care unit and are then transferred cost-benefit equation. We do not believe that to a general medical bed. There were I20 this happened, but can produce no objective general medical beds in the hospital. One was evidence to support our belief. We consider *sacrificed to make space for resuscitation that the extra work-load was successfully equipment, but otherwise the number of beds absorbed by the existing medical and nursing has remained constant throughout the period staff. In economic terms this means more under consideration. efficient use of labour. In human terms it means that well-motivated doctors and nurses Direct costs (Table i) were willing to work harder. In these inflationary times, we should point Drugs, etc We have taken into account out that costs refer throughout to those pre- drugs, oxygen, infusions, etc. used exclusively vailing in I970. for acute coronary care. As these were not al- Received 9 December I971. ways accounted for separately from those used 898 Reynell and Reynell Br Heart J: first published as 10.1136/hrt.34.9.897 on 1 September 1972. Downloaded from

TABLE I Direct costs of coronary care unit TABLE 2 Indirect costs of coronary care per annum unit per annum Salaries £2000 Extra admissions p.a. x mean length of stay x in- costs per day Drugs, oxygen, infusions, etc. £700 (i6o) (I6-7 days) (£2-50) Depreciation of capital equipment £565 = 6680 p.a. Repairs, etc. £580 Deduct £3845 Reduced length of stay x annual admissions x in-patient costs per day (3-i days) (520) (£2-50) =£C4030 p.a. Total =X2650 p.a. in the rest of the ward, it was necessary to rely on the ward sister's estimate of consumption. Depreciation on capital equipment (de- annum. Under this heading we include cater- fibrillators, oscilloscopes, electrocardiograph, ing, laundry, power-light-heat, medical pacemaker, mobile suction, anaesthetic trol- records, drugs, diagnostic radiology, path- ley, special beds, and wall sphigmomano- ology, and pharmacy. We omitted those items meters) The capital costs were those of cost not incurred by these particular pa- prevailing in I966 and replacement costs tients. Costs were obtained from the weekly would now be somewhat higher. Approximate itemized costs per hospital in-patient per week life spans were estimated for each item and provided by the hospital finance department. they were assumed to depreciate in a linear The in-patient costs as defined were £2-50 per rather than an exponential manner. patient per day. The total cost per annum is calculated by multiplying in-patient costs per Repairs These included replacement of day by the number of patients (i6o) and by leads, disposable electrodes, electrocardio- the mean length of hospital stay of all patients graph paper, and other items of expendable admitted to the unit during the fifth year copyright. equipment accounted for in the department (I6'7 days). of rather than the ward. The admission of extra patients led to in- creased pressure on medical beds and this re- sulted in a reduction in the mean length of Indirect costs (Table 2) stay of all patients admitted to the unit of 3-I The establishment of a coronary care unit may days over the five-year period. We therefore involve indirect costs by altering the pattern feel entitled to deduct in-patient costs for 3-I http://heart.bmj.com/ of delivery of medical care within the catch- days for each patient admitted to the unit in ment area of the hospital. In particular, it is a year. This reduced calculated indirect costs likely to result in the admission of patients to £2650 per annum. To the best of our who were previously treated at home. The knowledge our patients were none the worse annual admission rates to the unit during the first five years of its existence (Fig.) provide strong presumptive evidence that this was so. The figures suggest that it was over a year FIG. The increase in admissions to the on September 26, 2021 by guest. Protected before there was much change in the referring coronary care unit during the first five years habits of local practitioners. Admissions then of its existence. By the fifth year there is an rose steeply for two years, but thereafter ap- estimated excess of I6o admissions per annum. peared to be reaching a plateau. Fortunately this pattern of admission results in an S-shaped curve which makes it easy to esti- 500-/ ' mate that by the 5th year admissions exceeded by about i6o per annum the number which E would have occurred if there had been no coronary care unit. This involves the assump- tion that the admission rate of suspected myo- u400 / cardial infarcts in patients under the age of 65 0 would have remained constant over the five- .E year period if no coronary care unit had been established. It is therefore necessary to include relevant 300 2 3 4 5 in-patient costs for i6o extra patients per Year The cost-benefit analysis of a coronary care unit 899 Br Heart J: first published as 10.1136/hrt.34.9.897 on 1 September 1972. Downloaded from for being discharged earlier. Pressure on beds our 40 survivors, 9 were resuscitated at least may also have led to a small decrease in once by the emergency resuscitation service length of stay in the case of some other pa- while in a general medical bed. However, this tients admitted to the general medical wards, service is based on and administered by the but we cannot calculate this with any pre- coronary care unit and staffed by the same cision from figures available to us and we doctors and may be regarded as an extension believe that any resulting economic benefit of the unit. In calculating benefit, we have was very small. assessed our current performance as 12 lives The establishment of a coronary care unit saved per annum, the average of the last two may have altered the delivery of medical care years. We thus arrive at a cost of approxim- in the area in other ways, such as increased ately £540 per life saved (direct costs £320; load on ambulance services and decreased indirect costs £220). load on domiciliary services, but we do not The increased accuracy of diagnosis have access to data enabling us to estimate achieved by hospital admission rather than these items and we do not believe that they domiciliary care of patients with suspected would amount to any substantial positive or may well have influ- negative net balance. enced time spent off work, return to former employment, etc. Twenty-five per cent of admissions to our unit do not have myocardial Benefits infarction or coronary insufficiency. This is We propose a new unit of benefit, namely cost presumably a net benefit which cannot easily per life saved. A better unit would be man/ be expressed in quantitative terms. year/life saved, but this would require reliable survival tables and we do not yet have suffici- ent data to construct these with any accuracy. Discussion We have assumed that the only benefit of a We pay lip service to the concept that the coronary care unit is resuscitation from car- individual human life is infinitely precious copyright. diac arrest with survival to leave hospital. and that no expense must be spared when life Some 3 per cent of patients with proven myo- is in jeopardy, but advances in life-saving cardial infarction (WHO criteria) admitted to medical treatment are becoming increasingly our unit suffer a cardiac arrest from which costly and it is obvious that infinite financial they are resuscitated and survive to leave hos- resources are not likely to be made available. pital. This figure almost certainly under- More and more we are likely to be faced with estimates the true number of lives saved. the harsh criteria of cost effectiveness in http://heart.bmj.com/ Other measures such as the early recognition allocating priorities in medicine. and treatment of dangerous dysrhythmias and Previous attempts have been made to apply of left ventricular failure, and endocardial the technique of cost-benefit analysis to as- pacing of some patients with complete heart pects of the Health Service (Brooks, I969; block may sometimes be life-saving proce- Fein, I958; Holtman, I964; Klarman, I965; dures, but there is no way of proving this. Weisbrod, I96I), but it has always proved We prefer to take into consideration only difficult to express the benefits of medical what we can measure. The number of sur- care in financial terms. Attempts have been on September 26, 2021 by guest. Protected vivors from cardiac arrest during the first five made to express results in terms of working years of the unit's existence is shown in days saved, but what monetary value does one Table 3. There is some indication that the put on the work of a housewife or the survival number per annum is increasing. This is of a pensioner ? The application of strict partly due to the increased number of ad- economic criteria might imply that saving the missions, but we like to think that it is also life of a pensioner is a negative benefit, a due to the encouragement of general practi- judgement to which no doctor would sub- tioners to admit these patients early and to scribe. Too often one is reduced to putting a streamlining of resuscitation procedures. Of price tag on judgements which are essentially social and political (Stringer, 1970). There are, however, certain situations in TABLE 3 Number of patients surviving car- medicine in which it is possible to estimate diac arrest to leave hospital duringfirstfive with reasonable accuracy the number of lives years of unit's existence saved by expensive forms of medical treat- ment. We have in mind not only different Year I 2 3 4 5 Total types of coronary care units, but also intensive Survivors 3 8 5 10 I4 40 care units, renal dialysis, organ transplanta- tion, and certain types of expensive , goo Reynell and Reynell Br Heart J: first published as 10.1136/hrt.34.9.897 on 1 September 1972. Downloaded from in fact any life-threatening situation in which pose-built and independently staffed coronary the survival of treated patients can be com- care units save more lives at an acceptable pared with that of a comparable group for extra cost, but there is no evidence that this whom the treatment was not available. Money is so. is not always the most important element of 'cost'. Even at its most mechanized, medicine is a labour-intensive industry and scarcity of References skilled manpower is sometimes an even greater Brooks, R. G. (I969). Cost-benefit analysis of patients limiting factor. One can visualize similar cost/ treated at a rheumatism centre. Annals of the benefit equations expressed as nurse-hours Rheumatic Diseases, 28, 655. per life saved where this is appropriate. Our Fein, R. (I958). Economics of Mental Illness. Basic Books, New York. own unit was located in a general medical ward Holtman, A. G. (I964). Estimating the demand for in order to economize in scarce nursing staff public health services: the alcoholism case. Public rather than to save money, and our results Finance (The Hague), I9, 35!. could be expressed as one extra nurse/month Klarman, H. E. (I965). Syphilis control program. In Measuring Benefits of Government Investments, per life saved, though it is obvious that the Ed. by R. Dorfman. The Brookings Institution, small increase in nursing cover was not the Washington D.C. only, or even the principal, factor responsible Lawrie, D. M. (I969). Long-term survival after ven- for the saving of lives. tricular fibrillation complicating acute myocardial infarction. Lancet, 2, I085. In an outcome analysis by survival one McNamee, B. T., Robinson, T. J., Adgey, A. A. J., should ideally take into account the quality Scott, M. E., Geddes, J. S., and Pantridge, J. F. and duration of survival. We do not yet have (1970). Long-term prognosis following ventricular sufficient follow-up data to do this, but our fibrillation in acute ischaemic heart disease. British MedicalJournal, 4, 204. results to date show an 82 per cent one-year Reynell, P. C. (I969). Coronary care unit in a general survival and a 70 per cent two-year survival medical ward. British Medicaljournal, 2, 502. among survivors from cardiac arrest. Our Stringer, J. (I970). Cost and benefit assessment in experience accords with that of others that medical care. In Resources in Medicine, p. 23. Ed. by Jane Collins. King Edward's Hospital Fund for copyright. most of these patients return to work and that London. long-term prognosis does not differ signifi- Weisbrod, B. A. (I96I). Economics of Public Health. cantly from that of all patients discharged University of Pennsylvania Press, Philadelphia. from hospital after myocardial infarction (Lawrie, I969; McNamee et al., 1970). A cost Requests for reprints to Dr. P. C. Reynell, of C540 per survivor does not appear to be Bradfield Royal Infirmary, Duckworth Lane, excessive. It may be that more elaborate pur- Bradford 9, Yorkshire. http://heart.bmj.com/ on September 26, 2021 by guest. Protected