Estimation of contribution of changes in coronary care to improving survival, event rates and coronary heart disease mortality across the WHO MONICA Project populations: appendix to a paper published in the Lancet

Constructing an evidence-based treatment score for relating changes in treatment to changes in mortality, coronary events and case fatality in the WHO MONICA Project

February 2000

Michael Hobbs¹, Markku Mähönen² and Konrad Jamrozik¹ for the WHO MONICA Project³

¹ Department of Public Health, University of Western Australia, Perth, Australia
² Department of Epidemiology and Health Promotion (MONICA Data Centre), National Public Health Institute, KTL, Helsinki, Finland
³ Annex: Sites and key personnel of the WHO MONICA Project

Correspondence to: mikeh@dph.uwa.edu.au

• Copyright notice
• Document identification:
  • WHO MONICA Project e-publications (ISSN 2242-1246), No. 21.4
  • URL:http://www.ktl.fi/publications/monica/carpfish/appendb/wts.htm
  • URN:NBN:fi-fe976568

Contents

• 1. Introduction
• 2. Requirements for a weighted treatment score (WTS)
• 3. Measuring levels of treatment for CHD in the MONICA Project
• 4. Assigning treatment benefits
• 5. Estimating the proportions of events potentially amenable to particular treatments
• 6. Deriving a weighted treatment score - practical considerations
• References
• Acknowledgements

1. Introduction

During the 1980s, the benefits of several clinical interventions used in the treatment of coronary hearts disease (CHD) were clearly established through major randomized clinical trials (RCTs) and overviews [1-9]. The population impact of these advances remains uncertain, however. The MONICA Project provides the opportunity to examine the extent to which changes in the use of common medications and procedures relates to changes in coronary event rates, mortality rates and case-fatality. The potential for such an analysis is unfortunately restricted by missing data on treatment in fatal cases in many MCCs, which makes direct analysis of the effects of treatment in individual cases impossible. The only alternative to this is an ecological analysis of changes in treatment, as observed in nonfatal cases, and endpoints across study populations. This would be facilitated by a weighted treatment score (WTS) reflecting the combined effects of levels of particular treatments, the benefits of different treatments as determined in clinical trials and the proportion of fatal or total coronary events that were potentially amenable to treatment in different clinical settings.

2. Requirements for a weighted treatment score (WTS)

A WTS should reflect the sum of the population benefits attributable to particular treatments, each of which will be derived from

Es_i = Ep_i × F_i × B_i,

where:

• Es_i = events saved by treatment i as a proportion of all events,
• Ep_i = the proportion of events that may potentially receive treatment i,
• F_i = the treatment fraction for treatment i, and
• B_i = the benefit of treatment from treatment i expressed as a proportion of events treated.

The validity of a weighted treatment score will thus depend on how accurately we are able to estimate each of the last three variables.

3. Measuring levels of treatment for CHD in the MONICA Project

As indicated above, the ability of the MONICA Project to assess the contribution of changes in treatment for CHD to changes in coronary endpoints is restricted by the scope and completeness of information collected. For example, the Project did not attempt to collect population-wide data on all aspects of the treatment of coronary heart disease. The acute coronary care record was designed primarily to collect information about the treatment of AMI. Nevertheless information collected about treatment that cases of AMI received prior to admission and at the point of discharge from hospital is assumed to provide an indirect measure of trends in treatment of subacute disease.

A further constraint is imposed by inconsistency between MCCs in periods of collection of ACC data and by variable levels of missing information about previous history of CHD and its treatment, particularly for fatal cases occurring out of hospital. The only option has therefore been to use treatment levels in nonfatal cases admitted to hospital which was generally complete, as a guide to trends in treatment.

• Recommendation 1: Because of missing data in fatal cases, measurement of levels of treatment both before and during admission will be based on levels observed in cases of nonfatal AMI.

4. Assigning treatment benefits

Benefits from common clinical interventions for CHD recorded in the MONICA Project are summarized in Table 1. These are divided into those that apply to the treatment of acute myocardial infarction (AMI) and those employed in the treatment of subacute CHD and in follow-up care of symptomatic cases.

* indicative benefits based on RCTs [1-9];
** for at least two years following AMI;
# depending on severity and location of CA obstruction

Treatment during AMI can necessarily affect fatal events and case fatality only, whereas the treatment of subacute disease (as with beta-blockers (BBs) and antiplatelet therapy (APT) following AMI) may affect either or both fatal and nonfatal events equally. Thus treatment of subacute disease may affect coronary event rates without necessarily having any effect on case fatality. Ideally, separate treatment scores might be developed for fatal and nonfatal events. The practical difficulty of this will be discussed later.

While Table 1 can be used as guide to assign benefits of treatment in a treatment score, a number of qualifications and uncertainties remain as discussed below.

4.1 What is the benefit of revascularization procedures (CARPs) and which of these should be included in a treatment score?

The benefits of coronary artery bypass grafting (CABG), compared with medical treatment, vary greatly with the site and severity of obstruction and the number of occluded vessels (from 0 to 70% reduction in risk of death for up to five years), with minimal benefits in peripheral one vessel disease. The average benefit shown in the meta-analysis of RCTs of CABG compared with medical treatment was of 20% reduction in the risk of death. Since the RCTs of CABG were performed, both surgical techniques and medical treatment have improved and the absolute and relative benefit of the former is uncertain. While percutaneous transluminal coronary angioplasty (PTCA) has been shown to be as effective as CABG in equivalent cases [9], it will in general not be performed in more severe cases of multi-vessel disease and we would therefore not expect it to have the same overall average benefit as CABG. Studies based on procedure registries, however, do provide evidence of progressive substitution of PTCA for CABG, even in cases of three-vessel disease [10]. In addition to these sources of uncertainty we might also intuitively assume that the marginal benefits of CARPs might decrease as procedure rates increase, if this signifies greater availability of resources for less severe cases.

• Recommendation 2: For the purpose of this study, the reduction in risk of death from CHD with CARPs is assumed to be 20%
• Recommendation 3: The benefit of CARP will be based on the levels of treatment with CABG or PTCA (not CABG + PTCA)

4.2 Do the benefits of CARPs apply equally to fatal and nonfatal events?

The RCTs of CABG have used fatal endpoints only. Some registry-based studies have shown that the benefits apply also to total events, but we have no firm evidence that this is the case [11]. Some authors suggest that precipitation of AMI through the rupture of atherosclerotic plaques may not be prevented by CARPs, but when this occurs, death is less likely because of improved circulation [11].

• Recommendation 4: Because of uncertainty about the effect of CARPs, on nonfatal cases, the WTS will be relate primarily to fatal events.

4.3 What treatment benefits should be attributed to the use of BBs and APT following AMI?

The earlier overviews of BBs following AMI found that the treatment benefit was restricted to "one or two" years, which would then apply to a possibly minor proportion of deaths that eventually occur in persons who have had a previous AMI. The overall impact of BBs on event rates could thus be substantially diluted. More recently the extended follow-up of subjects in the Timolol trial suggests that the benefit is at least maintained [12,13]. However this does not necessarily mean that benefits continue to accrue. If this is the case we should probably still assume only a two-year benefit from BBs after AMI.

The same point has been made in relation to the use of antiplatelet agents following AMI in the overview of trials of APT [5]. However the situation is less clear, because the benefit of APT in other chronic vascular disease has not been qualified by period of use.

• Recommendation 5: The short-term treatment benefits of BBs and APT following AMI observed in trials will be discounted by 50% to allow for attenuation of benefits over longer periods of time.

4.4 What allowances should be made for time-dependent effects of treatment?

The risk of death following AMI declines exponentially with time from onset. Delays in treatment therefore mean that many ultimately fatal cases will not be protected. It is therefore not possible to attribute benefits of treatment accurately without knowing when treatment was started - information that was not recorded in the MONICA Project. As thrombolytic therapy (TT) is generally given within a few hours of admission to hospital, it seems reasonable to use the risk estimates given in RCTs to estimate benefits of treatment. But in the case of APT, BBs and ACE inhibitors, treatment may not be started at the time of admission. The extent to which this changes over time could bias estimates of treatment. For example, in a sample of cases in the Perth MONICA Register in which the time of commencement of treatment was determined, delays in treatment became less with calendar time (as awareness of the results ISIS-2 increased). Thus in the later years of the study, APT was used earlier and therefore more effectively as well as more frequently.

One approach to this problem might be to examine levels of treatment in fatal cases by days from onset of the event to death to determine when this levels out. A second option would be to assume that when FT was given, APT would have been commenced at about the same time (we found very few instances in which APT was not used in cases receiving FT). We could then apply the combined benefit for FT and APT to the proportion of cases given both FT and APT and a discounted benefit for APT to the remaining cases given APT. However, this would add greatly to the complexity of the analysis.

• Recommendation 6: Discounting the benefits of TT, APT and BB for delays in initiating treatment of AMI will not be attempted because of the added complexity that this would add to the analysis. The possibility that treatment benefits will be exaggerated should be recognised.

4.5 What benefit do we attribute to the use of ACE inhibitors?

Major RCTs and overviews of the early use of ACE inhibitors in AMI were published near the end or after the period of event monitoring of most MCCs [14].

Early use of ACE inhibitors is will therefore not be considered for inclusion in the WTS. It is assumed that during the MONICA Project these drugs were used principally later in hospital care for treatment of heart failure. The benefit of ACE inhibitors would therefore also apply mainly to late deaths. Compared with FT and APT, the pattern of use during the acute event of ACE inhibitors in fatal and nonfatal cases is strikingly different, with levels of treatment in fatal cases at least as high as in nonfatal cases. This is consistent with initiation of treatment later in the course of hospital treatment in contrast with those therapies that are commenced soon after admission.

• Recommendation 7: The treatment benefit of ACE inhibitors observed in RCTS of their use in heart failure (about 20% reduction) should be applied to later deaths only. This is assumed to be one quarter of the deaths potentially amenable to treatment with TT and APT.

4.6 Should the benefits of treatment with the same drug before and during hospital admission for MI both be counted?

Table 2 shows that 41% of hospital cases (and 55% of fatal cases managed in hospital) had a previous history of CHD. By 1991-93 the level of prior use of APT had risen to at least 50% in nearly half of the MCCs, while in the same proportion of MCCs, the use of BBs had risen to 40%. The question arises as to whether cases who had received treatment before admission to hospital would obtain the expected benefit from the same drugs used in hospital? It could be argued, for example, that such cases already represent treatment "failures".

* = column percentage
# = percentage of total deaths

• Recommendation 8: For simplicity, no allowance will be made for possible double counting of treatment effects of the same drugs used before and during treatment of AMI.

5. Estimating the proportions of events potentially amenable to particular treatments

Table 1 summarized the benefits of treatments given in the acute and subacute phases of CHD. The proportions of fatal or nonfatal coronary events that are potentially amenable to treatment in these different clinical settings must also be determined.

5.1 Events amenable to treatment of subacute CHD

The proportion of fatal or nonfatal events that are potentially amenable to reduction through treatment of subacute CHD (including follow-up care of symptomatic cases) should be identifiable in MONICA through the recording of previous history of CHD at the time of admission for AMI. Unfortunately this could not be recorded consistently in all MCCs, particularly in the case of deaths occurring out of hospital, for which access to named records was denied in some centres. It is therefore possible to make only a general estimate of the distribution of cases by previous history from the data of selected MCCs with few missing data for this variable. This is shown in Table 2 which indicates that 41% of all events (38% of non-fatal events and 45% of all fatal events) occur in persons with a previous history of CHD.

• Recommendation 9: The proportion of coronary deaths assumed to be amenable to the effects of treatment of subacute CHD (including follow-up care of symptomatic cases) is assumed to be 45%

Levels of treatment for selected drugs (BBs and APT) before admission in cases with a previous history of CHD have also been examined in selected MCCs with few missing data in fatal cases. Treatment fractions were found to be similar in fatal and nonfatal cases. We will therefore assume that levels of treatment in nonfatal cases can be used as a proxy for treatment in all cases.

5.2 Events amenable to treatment during the acute phase

A major problem in estimating the benefits of treatment in acute cases is that of identifying fatal hospital cases that, under normal clinical conditions, would not be considered for treatment. For example, cases of cardiac arrest in persons in hospital for other reasons, those patients with AMI who die soon after arrival at hospital or deaths occurring in persons with chronic CHD or heart failure. The extent to which selection for treatment occurs is illustrated in Table 3, which is based on data from selected MCCs in with missing data on treatment was missing in no more than 10% of fatal cases. Treatment levels of beta-blockers (BBs), fibrinolytic treatment (FT) and antiplatelet treatment (APT) were three to four times higher in nonfatal cases than fatal cases. These differences are far greater than can be explained by the effects of treatment per se and must therefore be due to case selection. An alternative way of making the same point is that case fatality in treated cases, also shown in Table 3, accounts for between 30% and 40% of total coronary deaths in hospital, and is comparable to the case fatality observed in controls of similar age in clinical trials. In general, these differences are seen regardless of the extent of missing data. If benefits of treatment were estimated by applying levels of treatment in nonfatal cases to all deaths in hospital, the deaths avoided would be greatly exaggerated.

An alternative method of determining treatment benefits would be to apply the treatment benefits from trials directly to the observed number of treated fatal cases. However even in the MCCs with the largest numbers of events, the number of such cases is small and estimates would be unreliable. Missing data relating to treatment in fatal cases in some MCCs would create an even greater problem.

The best option would seem to be to discount total deaths in hospital in all MCCs by the average proportion that they contribute in MCCs without high levels of missing data. In practice this will be between 30% and 40%. The acute treatment score will thus apply to say, 40% of inpatient deaths or about 12- 15% of total deaths.

• Recommendation 10: The proportion of total coronary deaths amenable to treatment during hospital admission for AMI is assumed to be 15%.

6. Deriving a weighted treatment score - practical considerations

The previous sections have outlined some of the complexities and uncertainties in developing a WTS. In particular we have noted:

• The need to use treatment levels in cases of nonfatal definite AMI admitted to hospital as a proxy measure of treatment in fatal cases and as a proxy measure of levels of treatment of subacute CHD in the wider population.
• Information about previous history of CHD, essential for determining the possible benefits of treatment before admission to hospital, is deficient in many MCCs, particularly in fatal cases.
• Inpatient treatment is relevant to mortality and case fatality but cannot directly affect coronary event rates. There is therefore a theoretical argument for having separate treatment scores for fatal and nonfatal events.
• The effects of CARPs on non-fatal events is uncertain – they may therefore have a greater impact on mortality rates and case fatality than on coronary event rates.
• The duration of benefit of beta-blockers and antiplatelet agents following AMI (and therefore the population impact) is uncertain.
• The benefits of treatment in hospital are time-dependent – a factor that cannot be allowed for easily in MONICA data.
• Assigning benefits for treatment with APT and BBs both before and during admission for AMI could lead to overestimation of benefits.

It would be possible to develop methods to overcome some of these problems. For example, discounting treatment benefits for delays in treatment in hospital; and developing separate scores for fatal and nonfatal cases. Sensitivity analyses could be also undertaken of the effects of varying the weights assigned for different treatments (as in the case of CABG and PTCA or for the use of BBs before admission to hospital). However given the deficiencies in the data that cannot be rectified, making these adjustments might imply precision in measuring treatment that cannot be justified and would add greatly to the complexity of analysis of the relationship between trends in treatment and different endpoints.

Despite the problems in assigning accurate treatment benefits, the one fact that stands out from Table 2, is that treatment of subacute disease has the potential to affect deaths more than treatment during the acute event by a factor of at least three. This fact alone justifies the development of a WTS. In the score recommended here, (Table 4), we have therefore emphasized this potential difference and placed less emphasis on assigning precise treatment benefits. For simplicity we have assumed a 20% reduction in the risk of death with CARPs and 10% reduction with treatment with BBs and APT for subacute disease, and 20% reductions with the use of APT, TT and ACE inhibitors and 10% reduction with the use of BBs during acute care. The benefit of ACE inhibitors has been discounted, however, because it is assumed that during the MONICA Project, its benefits applied only to late deaths related to heart failure.

Out of hospital treatment correction factor

The treatment fraction for treatments before admission to hospital should be measured only in those cases with a previous history CHD. Unfortunately the latter information was not consistently available in all MCCs, even in cases of nonfatal definite AMI. In Centres with adequate data, levels of treatment with APT and BBs before admission to hospital were in general at least twice as high in cases with a previous history of CHD compared with the level when all cases are included in the denominator. In Table 4 we have included a "correction factor" to be applied to the observed level of previous treatments based on all cases. This is conservative and will tend to lead to underestimation of treatment benefits in subacute CHD.

• Recommendation 11: Treatment levels before admission to hospital in persons with previous history of CHD are assumed to be twice the observed levels or treatment before admission in all cases.

References

1. Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis 1985;27(5):335-71.
2. ISIS-1 (First International Study of Infarct Survival) Collaborative Group. Randomised trial of intravenous atenolol among 16 027 cases of suspected acute myocardial infarction: ISIS-1.  Lancet 1986;2(8498):57-66.
3. Yusuf S, Wittes J, Friedman L. Overview of results of randomized clinical trials in heart disease. I. Treatments following myocardial infarction. JAMA 1988;260(14):2088-93.
4. Yusuf S, Wittes J, Friedman L. Overview of results of randomized clinical trials in heart disease. II. Unstable angina, heart failure, primary prevention with aspirin, and risk factor modification. JAMA 1988;260(15):2259-63.
5. Antiplatelet Trialists' Collaboration. Collaborative overview of randomised trials of antiplatelet therapy--I: Prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. [see comments] [published erratum appears in BMJ 1994 Jun 11;308(6943):1540]. BMJ 1994;308(6921):81-106.
6. Yusuf S, Zucker D, Peduzzi P, Fisher LD, Takaro T, Kennedy JW, et al. Effect of coronary artery bypass graft surgery on survival: overview of 10-year results from randomised trials by the Coronary Artery Bypass Graft Surgery Trialists Collaboration [see comments] [published erratum appears in Lancet 1994 Nov 19;344(8934):1446]. Lancet 1994;344(8922):563-70.
7. Hampton JR. Coronary artery bypass grafting for the reduction of mortality: an analysis of the trials. Br Med J (Clin Res Ed). 1984;289(6453):1166-70.
8. FTT Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative overview of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Fibrinolytic Therapy Trialists' (FTT) Collaborative Group [published erratum appears in Lancet 1994 Mar 19;343(8899):742] [see comments]. Lancet 1994;343(8893):311-22.
9. Pocock SJ, Henderson RA, Rickards AF, Hampton JR, King SB 3rd, Hamm CW, et al. Meta-analysis of randomised trials comparing coronary angioplasty with bypass surgery [see comments]. Lancet 1995;346(8984):1184-9.
10. Weintraub WS, Jones EL, King SB 3d, Craver J, Douglas JS Jr., Guyton R, et al. Changing use of coronary angioplasty and coronary bypass surgery in the treatment of chronic coronary artery disease. Am J Cardiol 1990;65(3):183-8.
11. Weintraub WS, Mauldin PD, Becker E, Kosinski AS, King SB 3rd. A comparison of the costs of and quality of life after coronary angioplasty or coronary surgery for multivessel coronary artery disease. Results from the Emory Angioplasty Versus Surgery Trial (EAST). Circulation 1995;92(10):2831-40.
12. The Norwegian Multicenter Study Group. Timolol-induced reduction in mortality and reinfarction in patients surviving acute myocardial infarction. N Engl J Med 1981;304(14):801-17.
13. Bellin EY. Six-year follow-up of the Norwegian multicenter study on timolol after myocardial infarction [letter]. N Engl J Med 1986;314(16):1052.
14. AIRE Study Investigators. Effect of ramipril on mortality and morbidity of survivors of acute myocardial infarction with clinical evidence of heart failure. The Acute Infarction Ramipril Efficacy (AIRE) Study Investigators [see comments]. Lancet 1993;342(8875):821-8.

Acknowledgements

The MONICA Centres are funded predominantly by regional and national governments, research councils, and research charities. Coordination is the responsibility of the World Health Organization (WHO), assisted by local fund raising for congresses and workshops. WHO also supports  the MONICA Data Centre (MDC) in Helsinki. Not covered by this general   description is the ongoing generous support of the MDC by the National  Public Health Institute of Finland, and a contribution to WHO from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland, USA for support of the MDC and the Quality Control Centre for Event Registration in Dundee. The completion of the MONICA Project is generously  assisted through a Concerted Action Grant from the European Community. Likewise appreciated are grants from ASTRA Hässle AB, Sweden, Hoechst AG, Germany, Hoffmann-La Roche AG, Switzerland, the Institut de Recherches Internationales Servier (IRIS), France, and Merck & Co. Inc., New Jersey, USA, to support data analysis and  preparation of publications.