Drug combinations and all cause mortality in heart disease: Authors' reply

BMJ 2005; 331 doi: https://doi.org/10.1136/bmj.331.7509.160 (Published 14 July 2005) Cite this as: BMJ 2005;331:160
  1. Julia Hippisley-Cox, professor of clinical epidemiology and general practice (julia.hippisley-cox{at}nottingham.ac.uk),
  2. Carol Coupland, senior lecturer in medical statistics
  1. Division of Primary Care, School of Community Health Sciences, Nottingham NG2 7RD
  2. Division of Primary Care, School of Community Health Sciences, Nottingham NG2 7RD

    EDITOR—There can be no substitute for a properly conducted randomised controlled trial in determining effectiveness of different treatments. All observational studies, however carefully designed and analysed, are subject to unmeasured confounding, as we point out in our paper. We think our study adds to the work on the Polypill by Wald and Law in providing more direct evidence on combinations of individual treatments as used in clinical practice in the United Kingdom.

    Firstly, we conducted an analysis restricted to patients without diabetes, congestive cardiac failure, or myocardial infarction, to deal with confounding by indication. As we reported in the paper, the results of the restricted analysis showed little change in the odds ratios for the various combinations of treatment except for angiotensin converting enzyme (ACE) inhibitors alone. For example, the adjusted odds ratio for the combination of statins, aspirin, and β blockers was 0.20 (95% confidence interval 0.11 to 0.37) and that for statins, aspirin, β blockers, and ACE inhibitors was 0.36 (95% confidence interval 0.15 to 0.88). This shows that confounding by indication was not a substantive source of bias in our analysis.

    Secondly, Snyder requested more data on the pattern of usage of different ACE inhibitors, which are shown in the table. This pattern reflects the rank order for prescriptions dispensed nationally according to UK prescribing analysis and cost data. This shows that a variety of ACE inhibitors were in use at the time of the study and that no particular preparation dominated the findings. We do not think that we have sufficient power to undertake further analyses comparing individual ACE inhibitors or an analysis by dose, particularly in combination with the other treatments.

    Pattern of angiotensin converting enzyme inhibitor use in cases and controls

    View this table:

    Thirdly, Vos raises the issue that the PROSPER study did not show a reduction in all cause mortality with pravastatin. However, mortality due to coronary heart disease fell by 24%.1 Also, the weight of the evidence from randomised controlled trials supports the hypothesis that statins improve survival for patients with ischaemic heart disease.25


    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    View Abstract

    Sign in

    Log in through your institution

    Free trial

    Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
    Sign up for a free trial