Guideline 57: Cardiac rehabilitation - Supporting material

Costing from a randomised trial of secondary prevention clinics in primary care

Campbell NC, Thain J, Deans HG, Ritchie LD, Rawles JM, Squair JL (1998) Secondary prevention clinics for coronary heart disease: randomised trial of effect on health. BMJ 316:1434-7.

Campbell NC, Ritchie LD, Thain J, Deans HG, Rawles JM, Squair JL (1998) Secondary prevention in coronary heart disease: a randomised trial of nurse led clinics in primary care. Heart 80:447-52.

In summary, nurse-led secondary prevention clinics were found to improve secondary prevention, improve health related quality of life and reduce hospital admissions (by about 28%).

Cost analysis

Direct and indirect health service costs attributable to the clinics are shown below. Also reported are the effects of the clinics on the use of health services (primary and secondary care) and estimates of costs attributable to changes in use.

1.1 Direct costs of the clinics

Nurse time formed the largest part of the direct clinic costs (table 1). In all, the study incurred 915 hours of nurse time, representing one hour and 22 minutes per patient invited to the clinics (673) or one hour and 40 minutes per patient who attended (551). Other costs included clinic materials (nurse manuals, relaxation tapes, exercise books, other consumables) and the costs of the training days. (No estimate has been included for accommodation because no additional accommodation was required.)

Table 1 Direct costs of the clinics

1.2 Indirect costs in primary care

The clinics encouraged increased use of laboratory tests for cholesterol and variations in prescribing (table 2). There were additional costs from increased lipid testing and prescribing; of prescribing costs, £68 of the £80 increase per intervention group patient and all of the £38 increase per control group patient were accounted for by statins.

There were no changes in general practitioner consultation rates. The median number of general practitioner consultations in three months for both intervention and control groups at baseline was one (P=0.107). It remained one in both groups at outcome (P=0.488).

Overall, the estimated additional cost to primary care incurred by the clinics was £68 per patient. If the findings of this study were applied throughout Grampian (population 539 698 (ISD, 1996)), we could expect 12,600 (2.34%) patients with identifiable coronary heart disease in general practice, and a total cost in primary care of approximately £857,000.

Table 2 Indirect primary care costs in intervention and control groups

1.3 Indirect costs in secondary care

Of 540 intervention group patients, 132 (24%) were admitted to hospital during the year prior to the study, and 106 (20%) during the study year. The corresponding figures for 518 control group patients were 137 (26%) and 145 (28%). Adjusting for age, sex, general practice and baseline performance, the odds ratio of requiring admission to hospital for intervention group patients was 0.64 (0.48 to 0.86, P=0.003). The difference was explained only partly by "cardiac" admissions: there were 36 (7%) in the intervention group and 49 (9%) in the control group during the study year. It was not due to differences in non-fatal myocardial infarctions: 13 (2%) in the intervention group; 12 (2%) in the control group.

The actual reduction in hospital admissions in the trial averaged 1.09 fewer nights in hospital per patient. Assuming a cost of £199 per bed night (ISD, 1996), the intervention has potential to save £216 per patient (table 3). If the findings of this study were applied throughout Grampian, we might expect 12 600 patients with coronary heart disease (2.34% of the total population of 539 698) to be identified and 38 fewer patients in hospital every night. This reduction in numbers is large enough to make savings realisable (by, for example, closing wards). The total reduction of 13 700 nights in hospital per year could release an estimated £2720 000 of resources (or savings).

Table 3 Indirect hospital costs in intervention and control groups.

This estimate of cost savings is, however, dependent on the actual cost savings in the study being translated into practice. The cost saving was mostly due to fewer patients in hospital but partly to a shorter mean length of stay. The latter was not statistically significant - before the study the median length of stay in hospital was seven nights in the intervention and six in the control group (P=0.435) and, during the study, it was six in both groups (P=0.408).

We could calculate a more conservative estimate of cost savings by assuming that the mean length of stay was the same in both groups (nine nights) and that the reduction in numbers of patients requiring hospital admission was at the lower limit of 95% confidence in the study (OR=0.86). In this case, the mean difference in hospital admission between intervention and control groups would be 0.27 nights per patient, and the potential saving £54 per patient. In Grampian, this would translate into nine or ten fewer patients in hospital every night or 3400 fewer nights in hospital per year.

Discussion

Economic evaluations have been reported for the Belfast, OXCHECK and Family Heart studies, all of which used similar nurse-led interventions. (O'Neill et al, 1996; Langham et al, 1996; Wonderling et al, 1996). Table 4 summarises the costs from all four studies.

Table 4. Costs (£) per patient attributable to nurse-led interventions for secondary and primary prevention.

* Equipment, training, travelling, support, laboratory tests
** Difference between intervention and control groups.

Some caution is needed in comparing the four studies because costs were collected and calculated in different ways. Importantly, in the Belfast, Oxcheck and family heart studies, costs were calculated per patient screened, whereas in the Northeast Scotland study, they were calculated per patient invited for screening. If adjustments were made for the rate of attendance following invitation (82%), direct costs per patient who attended nurse led clinics in the Northeast Scotland study would rise to £32.

The direct costs were similar in three of the four studies. The exception was the Family Heart Study whose intervention was considerably more labour intensive (Wonderling et al, 1996). There appear to be more differences in indirect costs but these have broad 95% confidence intervals which, in the Belfast, OXCHECK and Family Heart Studies include zero (O'Neill et al, 1996; Wonderling et al, 1996). Only in the Northeast Scotland study do the indirect costs reflect differences (in prescribing and hospital admission rates) that were statistically significant. The increased drug costs are in line with the considerable emphasis on and increase in drug prescribing, especially statins. The cost saving from fewer hospital admissions is particularly striking and needs explanation. In particular, could this have been a chance finding? Hospital admissions have a particularly high unit cost (£199 per night), so small changes lead to big differences in costs with broad confidence intervals. If adjusted to the lower limit of 95% confidence, however, there would still have been savings in hospital costs of £54 per patient. Improved implementation of secondary prevention would have been unlikely to reduce cardiac events so quickly and, indeed, the reduction in cardiac admissions was smaller. It is possible that improved fitness and wellbeing led to fewer admissions for other causes, but this does not appear to have occurred in the Belfast study. More detailed analysis of the reasons for hospital admissions is needed before the effect found can be attributed confidently to an effect from the secondary prevention clinics.

[Guideline Index]

 

Scottish Intercollegiate Guidelines Network,
Elliott House, 8-10 Hillside Crescent,
Edinburgh EH7 5EA
Web contact duncan.service@nhs.net
Last modified 16/09/11
© SIGN 2001-2011