Comparative effectiveness of exercise and drug interventions on mortality outcomes: metaepidemiological study

Naci H, et al investigated to determine the comparative effectiveness of exercise versus
drug interventions on mortality outcomes. They included 16 (four exercise and 12 drug) meta-analyses and included 305 randomised controlled trials with 339 274 participants. They observed that no statistically detectable differences were evident between exercise and drug interventions in the secondary prevention of coronary heart disease and prediabetes. Physical activity interventions were more effective than drug treatment among patients with stroke (odds ratios, exercise v anticoagulants 0.09 and exercise v antiplatelets 0.10). Diuretics were more effective than exercise in heart failure (exercise vs diuretics 4.11). They concluded that although limited in quantity, existing randomised trial evidence on exercise interventions suggests that exercise and many drug interventions are often potentially similar in terms of their mortality benefits in the secondary prevention of coronary heart disease, rehabilitation after stroke, treatment of heart failure, and prevention of diabetes.1

This study points the importance of therapeutic life style changes compared with drugs intervention. All physicians should think about this message in time of many drugs being prescribed while elderly population is increasing fast and life expectancy is longer. In this regard, statins, 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors, are very important and so, most frequently prescibed to prevent cardiovascular events in patients with high risk or even low risk of vascular disease by reducing low density lipoprotein cholesterol.2,3 Therefore, The 2013 American guidelines recommend to use high-intensity statin therapy and extend to use in more people.4 However, European worried about the potential side effects of statins in a large fraction of population for a life-long period.5 This is true to Asian people including Korean and Japanese in whom cholesterol levels and the prevalence of cardiovascular disease are still lower albeit both are increasing now.

Unwanted effects of statins have been neglected over the decade and nowadays, these off-target effects are being focused. Statin treatment are associated with deterioration of glucose homeostasis and increased risk of diabetes mellitus.6-8 Experimental and clinical studies have suggested these off-target effects are dose-dependent.9-13 A very recent study reported that after adjusting for baseline characteristics and time varying confounders, diuretics and statins were both associated with an increased risk of new-onset diabetes (hazard ratio 1.23 and 1.32, respectively).14 Most recently, renal insult by statin treatment has been suggested in contrast to the previous studies showing the renoprotective effect.15 These off-target effects are also dose-dependent. The use of high potency statins was associated with an increased rate of diagnosis for acute kidney injury in hospital admissions compared with low potency statins.

With current evidence, it is difficult to weigh the possible adverse renometabolic effects of statin on overall mortality in a quantitative manner. In the primary prevention studies, the use of statins is controversial: favoring3 or not favoring16,17 statins. According to a current meta-analysis on this issue, as compared with moderate-dose statin therapy, the number needed to harm per year for intensive-dose statin therapy was 498 for new-onset diabetes while the number needed to treat per year for intensive-dose statin therapy was 155 for cardiovascular events.12 This data support that lipid lowering effect of statins outweighs adverse renometabolic effect of statins. Of note, it would be prudent to consider current cardiorenometabolic status and natural histroy of diseases in each individual when choosing statin therapy for optimal individual patient health over the long-term. Indeed, therapeutic life style changes may be potentially as effective as many drug interventions on the secondary prevention of coronary heart disease, stroke, heart failure, and prediabetes.1

Funding: None, Disclosures: None

REFERENCES
1. Naci H, Ioannidis JPA. Comparative effectiveness of exercise and drug
interventions on mortality outcomes: metaepidemiological study. BMJ 2013;347:f5577

2. Baigent C, Blackwell L, Emberson J, et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet 2010;376:1670-81.

3. Cholesterol Treatment Trialists' (CTT) Collaborators, Mihaylova B, Emberson J, Blackwell L, Keech A, Simes J, Barnes EH, Voysey M, et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012;380:581-90.

4. Stone NJ, Robinson J, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2013, doi: 10.1161/01.cir.0000437738.63853.7a.

5. http://us5.campaign-archive2.com/?u=5722525cab46445583feb00fd&id=4c6ab38...

6. Koh KK, Sakuma I, Quon MJ. Differential metabolic effects of distinct statins. Atherosclerosis 2011;215:1-8.

7. Lim S, Sakuma I, Quon MJ, Koh KK. Potentially important considerations in choosing specific statin treatments to reduce overall morbidity and mortality. Int J Cardiol 2013;167: 1696-1702.

8. Lim S, Sakuma I, Quon M, Koh KK. Differential metabolic actions of specific statins: clinical and therapeutic considerations. Antioxid Redox Signal 2013 Aug 7. [Epub ahead of print]

9. Yada T, Nakata M, Shiraishi T, Kakei M. Inhibition by simvastatin, but not pravastatin, of glucose-induced cytosolic Ca2+ signalling and insulin secretion due to blockade of L-type Ca2+ channels in rat islet beta-cells. Br J Pharmacol 1999;126:1205-13.

10. Koh KK, Quon MJ, Han SH, et al. Simvastatin improves flow-mediated dilation but

reduces adiponectin levels and insulin sensitivity in hypercholesterolemic patients. Diabetes

Care 2008;31:776-82.

11. Koh KK, Quon MJ, Han SH, Lee Y, Kim SJ, Shin EK. Atorvastatin causes insulin resistance and increases ambient glycemia in hypercholesterolemic patients. J Am Coll Cardiol 2010;55:1209-16.

12. Preiss D, Seshasai SR, Welsh P, et al. Risk of incident diabetes with intensive-dose compared with moderate-dose statin therapy: a meta-analysis. JAMA 2011;305:2556-64.

13. Carter AA, Gomes T, Camacho X, Juurlink DN, Shah BR, Mamdani MM. Risk of incident diabetes among patients treated with statins: population based study. BMJ 2013;346:f2610.

14. Shen L, Shah BR, Reyes EM, Thomas L, Wojdyla D, Diem P, et al. Role of diuretics, β blockers, and statins in increasing the risk of diabetes in patients with impaired glucose tolerance: reanalysis of data from the NAVIGATOR study. BMJ 2013;347:f6745.

15. Dormuth CR, Hemmelgarn BR, Paterson JM, James MT, Teare GF, Raymond CB, et al; Canadian Network for Observational Drug Effect Studies. Use of high potency statins and rates of admission for acute kidney injury: multicenter, retrospective observational analysis of administrative databases. BMJ 2013;346:f880.

16. Culver AL, Ockene IS, Balasubramanian R, et al. Statin use and risk of diabetes mellitus in postmenopausal women in the Women's Health Initiative. Arch Intern Med 2012;172:144-52.

17. Goldstein MR, Mascitelli L. Do statins cause diabetes? Curr Diab Rep 2013;13:381-90.

We agree with Nunan and colleagues that causal inferences from randomized trials can be jeopardized by limitations in methodological quality in their design, conduct, analysis, and reporting, leading to underestimation or overestimation of the true intervention effect, i.e., bias. Unsurprisingly, none of the exercise trials identified in our review had double blinding (it would be impossible to mask patients to exercise interventions), and only a small number of trials blinded the assessors. However, meta-epidemiological reviews have demonstrated that methodological quality attributes (including blinding) do not seem to influence much the magnitude of objective outcomes such as mortality (this is in contrast to subjective outcomes where limitations in blinding and allocation concealment result in exaggerated treatment effects) [1, 2].

The important consideration in our network meta-analysis is whether the randomized trials of drug and exercise interventions have a balanced distribution of relative treatment effect modifiers [3]. We considered conceptual heterogeneity in terms of trial eligibility criteria, follow-up duration, blinding practices, and patient characteristics reflecting baseline disease severity. Patient population characteristics were broadly similar in all conditions except for stroke where patients participating in exercise trials had less severe disability compared to those included in drug trials [4]. All drug trials included in our review had participants with ischemic strokes.

There was generally more variability within individual meta-analyses than there was between meta-analyses of drug and exercise interventions in terms of patient population and trial characteristics. In cases where differences were observed between the trials of drug and exercise interventions (differences in terms of trial follow-up duration and blinding practices), these differences did not translate to statistical heterogeneity in effect size estimates. We have also statistically evaluated the consistency of direct and indirect evidence included in every closed loop in our network meta-analyses, and confirmed the lack of evidence of inconsistency.

Byatt and colleagues questioned why ACE inhibitors showed no mortality benefit in our analysis. Our objective was not to test the mortality benefits of exercise and drug interventions alone – this was done in the primary meta-analyses included in our review. Our analysis on ACE inhibitors was based on the 1-year survival benefits of ACE inhibitors reported by Flather and colleagues [5]. The magnitude of mortality benefits in our analysis closely parallel those in this earlier meta-analysis. The wider credible interval reflects our conservative approach in adopting random-effects Bayesian network meta-analyses, which often results in wider 95% uncertainty estimates. This also explains why the 95% CrI around the mortality benefit of exercise interventions in coronary heart disease crosses 1.00 unlike the original meta-analysis on which it is based.

Although Byatt and colleagues also state that no time scale was given for the mortality benefits reported in our study, these are provided in our online appendices. Average trial follow-up duration was generally similar across drug and exercise meta-analyses. For each condition, average trial follow-up duration was never the shortest for exercise trials (24 months in trials of coronary heart disease; ~6 months in trials of stroke; ~12 months in trials of heart failure; and > 2 years in trials of pre-diabetes).

Muhlhauser and Meyer suggest that meta-analyses of complex interventions may be obsolete due to heterogeneity. We do not share this view. Heterogeneity can be helpful in identifying important relative treatment effect modifiers. Regardless, in our review, there was little statistically detectable between-study heterogeneity among the trials of exercise-based interventions. We acknowledge that modest heterogeneity is almost always difficult to exclude. However, the observed consistency of the results may suggest that factors that are typically considered relative treatment effect modifiers do not in fact influence much the treatment effects. Moreover, there is no evidence to suggest that exercise interventions are confounded by intensification of care, which is likely due to successful trial practices ensuring that the type, intensity, and frequency of drug interventions were similar in exercise and control arms of randomized trials. Finally, objective outcomes such as mortality are largely immune to many potential biases in methodological quality attributes of trials [1, 2].

Muhlhauser and Meyer also suggest that the randomized trials of exercise interventions “are likely to be performed by people who want to proof of benefit of exercise.” To the extent that this is a problem, it would be important to point out that the randomized trials of drugs are almost exclusively performed by pharmaceutical companies that have clear financial conflicts of interests in the findings of their randomized controlled trials. It has been shown time and again that pharmaceutical companies exclusively sponsor trials that have favorable conclusions for their products [6, 7]. Industry sponsorship creates asymmetries in the evidence base, with some companies selectively comparing their products against others that they deem straw man competitors – in a consistent manner with the recent concern voiced by Light and Lexchin, postulating that much of the research and development activities of pharmaceutical companies are driven by marketing and sales motives [8].

As Hopayian suggests, more research is needed – particularly to investigate the “formulation” and “dose” of different types exercise interventions that may work as well or better than commonly prescribed drugs. We hope that our paper prompts increased discussion and data collection to eventually foster improved comparisons between drug and exercise interventions. In addition to the need to invest in large randomized controlled trials directly comparing drug and exercise interventions in a wide range of illnesses, we hope that researchers undertake similar reviews; expand the breadth and depth of our analysis; and embark upon a research agenda that compares the benefits and harms of pharmacological and non-pharmacological interventions. We agree with the importance of systematically exploring the impact of exercise interventions on additional outcomes beyond mortality. To this end, we look forward to the findings of the systematic review that Nunan and colleagues are undertaking [9].

Huseyin Naci and John PA Ioannidis

References:

[1] Savović J, Jones HE, Altman DG, Harris RJ, Jüni P, Pildal J, Als-Nielsen B, Balk EM, Gluud C, Gluud LL, Ioannidis JP, Schulz KF, Beynon R, Welton NJ, Wood L, Moher D, Deeks JJ, Sterne JA. Influence of reported study design characteristics on intervention effect estimates from randomized, controlled trials. Ann Intern Med. 2012 Sep 18;157(6):429-38.

[2] Wood L, Egger M, Gluud LL, Schulz KF, Jüni P, Altman DG, Gluud C, Martin RM, Wood AJ, Sterne JA. Empirical evidence of bias in treatment effect estimates in controlled trials with different interventions and outcomes: meta-epidemiological study. BMJ. 2008 Mar 15;336(7644):601-5.

[3] Jansen JP, Naci H. Is network meta-analysis as valid as standard pairwise meta-analysis? It all depends on the distribution of effect modifiers. BMC Med. 2013 Jul 4;11:159.

[4] Naci H, Ioannidis JPA. Comparative effectiveness of exercise and drug interventions on mortality outcomes: metaepidemiological study. BMJ 2013,347:15577.

[5] Flather MD, Yusuf S, Køber L, Pfeffer M, Hall A, Murray G, Torp-Pedersen C, Ball S, Pogue J, Moyé L, Braunwald E. Long-term ACE-inhibitor therapy in patients with heart failure or left-ventricular dysfunction: a systematic overview of data from individual patients. ACE-Inhibitor Myocardial Infarction Collaborative Group. Lancet. 2000 May 6;355(9215):1575-81.

[6] Lathyris DN, Patsopoulos NA, Salanti G, Ioannidis JP. Industry sponsorship and selection of comparators in randomized clinical trials. Eur J Clin Invest. 2010 Feb;40(2):172-82.

[7] Stamatakis E, Weiler R, Ioannidis JP. Undue industry influences that distort healthcare research, strategy, expenditure and practice: a review. Eur J Clin Invest. 2013 May;43(5):469-75.

[8] Light DW, Lexchin JR. Pharmaceutical research and development: what do we get for all that money? BMJ. 2012 Aug 7;345:e4348.

[9] Nunan D, Mahtani KR, Roberts N, Heneghan C. Physical activity for the prevention and treatment of major chronic disease: an overview of systematic reviews study protocol. Systematic Reviews 2013,2:56.

The obvious intention of the authors of this meta-analysis to stress the need for more and better effectiveness and comparative studies on exercise is well appreciated. However, the methodological approach used in this metaepidemiological study does not generate valid information. The results are biased and misleading.

The authors compare complex interventions with single component drug interventions. Evaluation of complex interventions requires specific considerations [1]. Meta-analysis of complex interventions may be obsolete due to clinical heterogeneity and incomplete information [2]. The exercise studies included in this meta-analysis suffer from all problems inherent to complex behavioural interventions. Exercise is by itself a complex intervention and often part of multicomponent treatment regimens. The intervention is not blinded, most studies lack an appropriate sham intervention, and studies are likely to be performed by people who want to proof benefit of exercise. Without appropriate control interventions exercise treatment is inevitably confounded by intensification of care which usually improves adherence to drugs and other therapies. Exercise studies may include healthier patients than drug trials. Additional to these shortcomings the studies included in the present meta-analysis demonstrate a substantial risk of bias. The related systematic reviews report that most exercise studies are characterized by unclear randomisation procedure, lack of blinding, high drop-out rates and no intention-to-treat analysis. Publication bias is likely. Accompanying treatments including dosing and adherence to prescribed therapies are rarely reported. In contrast, most drug trials had a low risk of bias and as single component interventions were less likely to be prone to confounding.

Due to the substantial limitations of the quality of exercise intervention trials the conclusions of the present meta-analysis are not justified. Systematic reviews have to acknowledge the quality of the included studies and with regard to exercise the specific methodological challenges related to complex interventions.

1)Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, Petticrew M: Developing and evaluating complex interventions: the new Medical Research Council guidance. BMJ 2008; 337:a1655.
2)Lenz M, Steckelberg A, Richter B, Mühlhauser I. Meta-analysis does not allow appraisal of complex interventions in diabetes and hypertension self-management: a methodological review. Diabetologia 2007; 50: 1375-1383

Ingrid Mühlhauser, MD, PhD, professor
Unit of Health Sciences and Education
University of Hamburg
Martin-Luther-King Platz 6
D-20146 Hamburg
Germany
Email: Ingrid_Muehlhauser@uni-hamburg.de

Gabriele Meyer, PhD, professor
Faculty of Medicine
Institute for Health and Nursing Science
Martin-Luther-University Halle-Wittenberg
Magdeburger Straße 8
D-06110 Halle (Saale)
Germany
Email: Gabriele.Meyer@medizin.uni-halle.de

While Naci and Ioannides’ network analysis of meta-analyses comparing the effects of exercise against medication on mortality is an important landmark, there are a few problems that should be considered.

First, one inclusion criterion is debatable. When more than one meta-analysis was available, they chose the most recent. An alternative would have been to choose on the basis of quality. At the very least, it is desirable to see a quality assessment of the meta-analyses and a consideration of how that might have affected the results.

Second, lumping all forms of exercise without consideration of the “formulation” and “dose”, to borrow terms from drug prescribing, left the study open to misclassification bias, with a tendency to underestimate the effects of the efficacious forms of exercise. The authors allude to this themselves.

Third, I wonder if they have over-extended their conclusion with “our analysis suggests that exercise potentially had similar effectiveness to drug interventions”. What it demonstrated was that, barring two exceptions, there was no difference between the two options. To demonstrate similarity, a non-inferiority study would be needed.