Description of the study design and population

Describe the evidence presented in the paper you selected. Specifically address the following questions:

The case-control study was conducted with an objective to see the relationship of physical exercise with coronary heart disease (CHD) in urban areas of Indian Population (Rastogi et al 2004). The urban hospitals of New Delhi and Bangalore were selected to take a sample of 350 cases and 700 controls. The cases and controls were matched for age, gender and hospital. The analysis was done using conditional logistic regression to control for confounders due to matching and also other confounders. The results found that 48 percent of controls and 38 percent of cases participated in some kind of leisure-time physical activity. The participants with highest level of physical activity had lowest risk of developing CHD. On the other hand; people with increased levels of sedentary lifestyles, had increased risk of developing CHD. Thus the paper concluded that leisure-time exercise such as 35-40 minutes per day of brisk walking had a protective effect on heart as compared to sedentary lifestyle. The paper recommended that physical activities in daily lives should be promoted in urban India. 

-What was the exposure or intervention?

The exposure or intervention was leisure time exercise including about 35-40 minutes of brisk walking. 

-What was the outcome?

The outcome was acute Myocardial Infarction. 

-What was the study design?

The study design was case-control study. 

-What was the study population?

The study population was patients of urban hospitals in New Delhi and Banglore. 

-What were the main findings?

People with exposure to highest level of leisure time exercise, have relative risk of 0.45 (95% CI 0.31-0.66) as compared to non-exercise group. It means leisure-time exercise had protective effect for Cardio-vascular disease risk. On the other hand people with greater than 3.6 hours per day of sedentary activity had about 1.88 times (95% CI 1.09-3.20) higher risk of developing CHD, as compared to people with less than 70 minutes per day of sedentary activity per day. 

To what extent can the observed association between the exposure and outcome be attributed to non-causal explanations? Consider the following questions in your critique: 

-Are the results likely to be affected by selection and/or measurement bias?

Although the patients in both the groups were matched by age, gender and hospital; but still there are chances of selection bias as the control subjects were relatively healthy and were with minor ailments as compared to cases which were having diagnosis of Acute Myocardial Infarction (AMI). It is very common to have selection bias in case-control studies (van Rein et al 2014). The selection of controls could be inappropriate and there could be bias in selection of cases itself. In case of AMI, the disease is very severe and the chances of death are high. Thus in the present study, only surviving patients were selected which had comparatively less severe illness, which may result in bias (van Rein et al 2014). This is also called “survivor bias in case-control studies” (ibid).

Details of exposure (leisure time exercise) and outcome (coronary heart disease)

The author has also discussed selection bias in discussion section of the paper. Author argued that since controls were chosen from seven different out-patient clinics and in-patient wards; there are chances that association exists in one particular group and not in others which may dilute the results and induce bias in the results. The author also refers to that cases who survived were only interviewed, although there were 25 cases that did not survive and thus not included in the study. The author has also argued the possibility that only health conscious individuals gave consent for the study, which may have induced bias in the study. The author has also told that controls in the study were more educated and had lower incomes than cases; and this may have induced selection bias in the study..

-Are the results likely to be affected by confounding?

The confounders which were adjusted include age, gender, cigarette and bidi smoking, BMI, WHR, alcohol intake, education, or income; but there is still possibility of other unknown confounding factors which might have affected the results, such as use of smokeless tobacco. Professor Neil Pearce in his article has stated that matching in case-control studies do not remove confounding rather it can introduce the confounding (Pearce 2016). This is because, while attempting matching for confounding factors, matching is also done unintentionally for exposure itself (ibid). Further matched case-control design should also include matched analysis (ibid). Professor Pearce also argued in his paper that; if there is an association between matching factor and exposure; then matching can actually introduce confounding if not controlled in the analysis (ibid). In this particular paper; various potential confounders such as age, family history of CHD, cigarette smoking, bidi smoking, were controlled in the analysis. Thus every possible effort was done to remove confounding.

The author has also argued in his paper that physical activity may also be protective for some of the ailments in control group, which may have induced confounding bias in the study. 

-Are the results likely to be affected by chance variation?

The results were significant at 95 percent level of significance. Thus there were only 5 percent chances of chance variation in the results.

Chance variation is also called chance error or random error. It is inherent in any research based on statistical predictions. It is the difference between the predicted value and actual value. In other words it is the probability by which the estimates differ from the true value. In a normal distribution curve, if we take a range up to 1.96 standard deviations above and below the estimated mean; there are 95 percent chances that true value will fall in that range; which leave only 5 percent chances of any variation which is called chance variation. There will be 2.5 percent chances that true value will be above this range and 2.5 percent chances will be that true value is below this range. Thus total percentage of chance variation will be 5 percent (Sowey, & Petocz, 2017). 

Main findings of the study

Do you think there is evidence of a causal association between the exposure and the outcome? This question relates to the internal validity of the study.

Consider the following questions in your critique:

The scientific research is considered to be internally valid if it could minimise Systematic errors or bias; and the cause-effect relationship is not a spurious relationship. There are several criteria defined by various epidemiologists from time to time; which validates the evidence of causal association between the exposure and the outcome; within the study. Some of these criteria are as follows. The cause must precede the effect i.e. there should be temporal relationship between cause and effect. The covariation between cause and effect should be high. It means, there should be a clearly visible change in outcome by changing the exposure. There should be a dose-response relationship between exposure and the outcome. It means higher the change in exposure, higher should be the change in outcome or vice-versa (Neuman 2016). These criteria for internal validity of study are further discussed in detail as follows. 

-Is there a temporal relationship between exposure and outcome?

Yes, there was a temporal relationship between exposure and outcome. It was a case-control study where exposure preceded the outcome. The cases of AMI and controls were recruited in the study. They were then asked about their physical activity and leisure-time exercise to which they were exposed in the past. Thus exposure or non-exposure to leisure-time exercise preceded the development or non-development of AMI.

There is an increasing interest over the years for the life-course framework to chronic diseases epidemiology (Lynch & Smith 2005). The life-course approach also includes temporal relationship where exposure precedes outcome and is accumulated throughout life to precipitate in old age or middle age in the form of chronic diseases such as AMI (ibid). Exposure to various determinants such as sedentary lifestyle at different life-course stages influence specific biological processes and thus the development of chronic diseases such as AMI (ibid). 

-Is there a strong relationship between the exposure and the outcome?

Yes, the relationship between leisure time exercise and AMI was very strong as the P value was less than 0.0001 for the relationship. People with 35-40 minutes of brisk walking had 55 percent lower risk of developing AMI as compared to controls who did not exercise. 

-Is there a dose-response relationship between exposure and the outcome?

Yes, there was a dose-response relationship between exposure and the outcome. People in the highest level of physical exercise had lowest risk of developing AMI and this trend was significant at p<0.0001. 

Evaluation of non-causal explanations for observed association

-Are the results consistent within the study?

Yes, the results were consistent within the study.

Age and Sex adjusted analysis also showed that leisure-time physical exercise lowered the risk of AMI. After adjusting for confounders like cigarette/ bidi smoking, the leisure time exercise had protective effect on CHD risk. The results were also consistent in multi-variate analysis. 

Do the findings accord with other evidence? Specifically: 

-Are the findings consistent with other evidence, particularly evidence from studies of similar or more powerful study design?

Yes, the findings are consistent with other evidence as shown in the article itself. The article refers to one prospective study from US on women that concluded that more than 3 hours per week of leisure time physical activity has a protective effect on heart. Another cohort study on US men concluded that individuals doing more than 30 minutes per day of moderate-intensity physical exercise had 20 percent lower chances of developing CHD. The selected paper refers to another US-based cohort study on post-menopausal women, which concluded that walking daily have a protective effect on heart.

Also the findings are consistent with evidence from some recent studies. In 2014, Andersen and colleagues found that leisure time physical activity has a protective role towards the risk of developing AMI and the two have a dose-response relationship (Andersen et al 2014). Also the INTERHEART study in China found a protective role of leisure-time physical activity for AMI as compared to sedentary lifestyles (Cheng et al 2014). In Copenhagen City Heart study, it was found that leisure-time physical activity has a protective effect in post-MI patients (Saevereid et al 2013). A recent systematic review and meta-analysis by Claes et al also showed that home-based physical exercise is protective for cardiovascular rehabilitation (Claes et al 2017). Some other studies have also found that exercise-based rehabilitation improves quality of life and functional capacity of heart (Peixoto et al 2015). 

-Are the results plausible in terms of a biological mechanism?

Yes, the results are plausible in terms of a biological mechanism. The leisure-time physical activity results in lipid lowering in Atherosclerotic plaques. It also reduces thrombotic potential and increases fibrinolytic potential (Libby 2013). The study has also mentioned underlying biological mechanisms due to which physical activity has beneficial effects on CVD risk. These include reduced blood pressure, increased HDL (High-Density Lipoproteins), increase in insulin sensitivity, improvement in endothelial function, and reduction in atherogenic cytokine production. 

Discussion on bias and confounding

Are the findings externally valid, that is generalisable? Specifically:

The external validity of a research refers to the extent to which the results of the study could be generalised across heterogenous populations. The sampling bias may be a threat to external validity of research where the sample is not true representative of exeternal populations (Pearl 2017). It is important to strengthen the reporting of findings on external validity so that context of application of findings could be understood i.e. whether the results apply to local settings or group settings or wider country settings. This is important to translate research in to practice as the interventions will also be applicable to similar context or settings (Steckler, & McLeroy, 2008).

This particular research was conducted on hospital patients from Delhi and Banglore. Thus results of the study could be generalised to urban metros of India only, that too particularly Delhi and Banglore. If another study is conducted by taking controls outside hospitals, from general public, the results could be different. Thus to be further specific, the results of the study could be generalised to urban hospital patients of New Delhi and Banglore. 

-Can the findings be applied to the source population from which the study population was derived?

The study population was derived from urban hospitals of New Delhi and Banglore. Sample was sufficient to generalise the findings to the source population; but as the study was done on patients selected from hospitals, the generalisation of results to general public outside hospitals is questionable. 

-Can the study results be applied to other relevant populations? 

The study results are specific to New Delhi and Banglore as the selected sample was representative of hospital patients of these two cities; and thus the results could not be applied to other relevant populations.

References 

Andersen, K., Mariosa, D., Adami, H. O., Held, C., Ingelsson, E., Lagerros, Y. T., … & Sundström, J. (2014). Dose-response relations of total and leisure-time physical activity to risk of heart failure: a prospective cohort study. Circulation: Heart Failure, CIRCHEARTFAILURE-113.

Cheng, X., Li, W., Guo, J., Wang, Y., Gu, H., Teo, K., … & Yusuf, S. (2014). Physical activity levels, sport activities, and risk of acute myocardial infarction: results of the INTERHEART study in China. Angiology, 65(2), 113-121.

Claes, J., Buys, R., Budts, W., Smart, N., & Cornelissen, V. A. (2017). Longer-term effects of home-based exercise interventions on exercise capacity and physical activity in coronary artery disease patients: A systematic review and meta-analysis. European journal of preventive cardiology, 24(3), 244-256.

Libby, P. (2013). Mechanisms of acute coronary syndromes and their implications for therapy. New England Journal of Medicine, 368(21), 2004-2013.

Neuman, W. L. (2016). Understanding research. Pearson.

Pearce, N. (2016). Analysis of matched case-control studies. bmj, 352, i969.

Pearl, J. (2017). The Eight Pillars of Causal Wisdom (Lecture notes for the UCLA WCE conference, April 24, 2017).

Peixoto, T. C., Begot, I., Bolzan, D. W., Machado, L., Reis, M. S., Papa, V., … & Guizilini, S. (2015). Early exercise-based rehabilitation improves health-related quality of life and functional capacity after acute myocardial infarction: a randomized controlled trial. Canadian Journal of Cardiology, 31(3), 308-313.

Rastogi, T., Vaz, M., Spiegelman, D., Reddy, K. S., Bharathi, A. V., Stampfer, M. J., … & Ascherio, A. (2004). Physical activity and risk of coronary heart disease in India. International journal of epidemiology, 33(4), 759-767.

Saevereid, H. A. S., Schnohr, P. S., & Prescott, E. P. (2013). Speed and duration of walking and other leisure time physical activity and the risk of heart failure: the Copenhagen City Heart study. European Heart Journal, 34(suppl 1), P3646.

Sowey, E., & Petocz, P. (2017). A Panorama of Statistics: Perspectives, Puzzles and Paradoxes in Statistics. John Wiley & Sons.

Steckler, A., & McLeroy, K. R. (2008). The importance of external validity. American Journal of Public Health, 98(1), pp. 9–10.

van Rein, N., Cannegieter, S. C., Rosendaal, F. R., Reitsma, P. H., & Lijfering, W. M. (2014). Suspected survivor bias in case–control studies: stratify on survival time and use a negative control. Journal of clinical epidemiology, 67(2), 232-235.