Letters

Secondary prevention of coronary heart disease

Ill defined inclusion criteria resulted in missed trials

Secondary prevention programmes may reduce overall mortality in high risk patients

Improved outcomes need to be defined

Ill defined inclusion criteria resulted in missed trials

EDITORThe review by McAlister et al of secondary prevention programmes in coronary heart disease does not adhere to some of the major principals of good practice when conducting systematic reviews of the medical literature.^1-3 These include a clearly defined research question, strict inclusion criteria so that the review can be replicated, an exhaustive search of the medical literature to find all relevant studies, and findings that can be interpreted easily by the reader and relate to clinical practice.

McAlister et al, at first glance, have chosen a huge area of the medical literature to review, encompassing both pharmacological and non-pharmacological interventions for the secondary prevention of coronary heart disease. On closer inspection, they say that it is their intention to review the literature concerned with disease management programmes for coronary heart disease. The definition of disease management programmes used is broad and is quoted as that proposed by Hunter et al as a combination of patient education, provider use of practice guidelines, appropriate consultation, and supplies of drugs and ancillary services; from the same source, Hunter et al also say that the spectrum of disease management extends from health promotion and disease prevention, through diagnosis, treatment and rehabilitation to long term care.⁴ This highlights the need to be absolutely explicit about the inclusion and exclusion criteria applied to a review in this area.

The trials included in the review are a mix of nursing and multidisciplinary team interventions, and also comprehensive cardiac rehabilitation programmes. Single modality rehabilitation programmes were excluded. Many trials of comprehensive cardiac rehabilitationwhich should be considered admissible under the umbrella of disease management programmesare missing from studies included in the review. A recent Cochrane review of exercise based rehabilitation for coronary heart disease, which was not picked up with the search strategy used by McAlister et al, cites at least 17 trials concerned with comprehensive cardiac rehabilitation that could have also been included in the review by McAlister et al.⁵ Including these trials in the current review results in a pooled odds ratio for all cause mortality of 0.87 (95% confidence interval 0.76 to 1.0).

A precise definition of disease management programmes is problematic, but using the authors' own description, it seems that deficiencies in searching and application of inclusion criteria have resulted in a review that that is difficult to interpret, is not replicable, and is potentially misleading.

Karen Rees, research fellow in systematic reviews. 
Karen.Rees{at}bristol.ac.uk

Shah Ebrahim, coordinating editor, Cochrane Heart Group. 
Shah.Ebrahim{at}bristol.ac.uk Department of Social Medicine, University of Bristol, Bristol BS8 2PR

Secondary prevention programmes may reduce overall mortality in high risk patients

EDITORIn a meta-analysis of randomised trials, McAlister et al showed that secondary prevention programmes reduce admission to hospital and improve quality of life or functional status in patients with coronary heart disease, but their effects on patients' survival remained uncertain.¹ In primary prevention through lowering blood pressure, the absolute risk reduction is positively related to the overall risk profile of the patient. ^{2 3} This implies that in combining patients at high and low risk the meta-analysis may have missed an effect in the latter group of patients. We reassessed the risk reduction in overall mortality by dividing the trials equally into two groups according to the proportion of death events in the control group, an indicator of how likely a patient may die in the absence of the prevention programmes.

Heterogeneity in the risk difference and the odds ratio on a logarithmic scale was not significant (P>0.20) in the 10 trials that provided data on mortality. The mortality in the control group of the trials varied from 1.9% to 19.1%. Although the trial by Cupple et al showed a significant difference in mortality between the intervention and control group, neither the combined overall odds ratio nor the risk difference was significant. A risk of 7% divided the trials into two groups with an equal number. The combined risk difference, according to the fixed effect method, was 0.3% (95% confidence interval -0.7% to 1.3%) in the five low risk trials and -2.7% (-4.6% to -0.8%) in the five high risk trials.⁴ The two confidence intervals did not overlap, and the difference between the patients at high and low patients was significant (P=0.01). In addition, the combined odds ratio was also significantly lower in the high risk trials than that in the low risk trials (P=0.05).

These results suggest that secondary prevention programmes may reduce the overall mortality in patients at high risk but not in those at low risk. This reanalysis provides an example in which genuine clinical heterogeneity may still exist even when the overall heterogeneity is insignificant, heightening the importance of exploring the sources of heterogeneity in meta-analysis.⁵

Jin Ling Tang, associate professor in epidemiology and community medicine. 
Wilson Tam, statistician. 
Department of Community and Family Medicine, School of Public Health, Chinese University of Hong Kong, Shatin, New Territories, Hong Kong Special Administrative Region, China

Improved outcomes need to be defined

EDITORThe paper by McAlister et al on secondary prevention programmes in coronary heart disease was highlighted in Editor's choice and This week in the BMJ with the teaser that it showed that disease management programmes in coronary heart disease prevention can improve outcomes.¹

What outcomes does the paper actually measure? Reinfarction rate? No. Overall mortality? No. McAlister et al point out that the study periods were too brief. Rate of admission to hospital? Debatable. Only two of the six reporting studies showed benefit and not to level of significance. Quality of life? Some. Again, significance not shown. Processes of care? This is an interesting one. McAlister et al suggest that this means the recording of risk factors and the prescribing of drugs.

The bottom line is the claim that improved outcomes means that the intervention groups are prescribed more drugs and that that this is a good thing, claiming that the overall benefits of the drugs are already proved. Is this what improved outcomes really are? The consumption of a larger number of drugs to treat lipids, platelets, and blood pressure readingswhere are the patients in all this? What is their experience? What about adverse reactions and side effects of the drugs? What were the improved outcomes of those who took cerivastatin and developed rhabdomyolysis?

We all want to achieve improved outcomes, but it would be more helpful if the BMJ were to be more specific in defining its terms. I know from this article that the described interventions lead to the prescribing of more drugs. I do not know whether these programmes are to be recommended.

Bob Leckridge, associate specialist. 
Glasgow Homeopathic Hospital, Glasgow G12 OYN bob.leckridge{at}virgin.net