Study Summary
In the article titled, “Physical Activity, Sedentary Behavior, and Cause-Specific Mortality in Black and White Adults in the Southern Community Cohort Study,” Matthews et al. (2014) strived to understand the causality between physical activities and adult mortality among male adults from two ethnic groups (African-Americans and Whites).
The primary exposures were sedentary lifestyles and physical activities. Adult mortality was the outcome of interest. Here, the authors used different measures to analyze the dependent variables and their influence on adult mortality in the country (Matthews et al., 2014).
The study was prospective because it was watching out for health outcomes that emerged because of environmental exposures. Its structure could also help to explain the etiology of many chronic diseases that affect many adults in America (Matthews et al., 2014).
Matthews et al. (2014) recruited 85,000 male adults from the southern part of the United States (U.S.). Particularly, the study focused on 12 key southern states – “Florida, Alabama, Mississippi, Louisiana, Arkansas, Tennessee, Georgia, South Carolina, North Carolina, Virginia, West Virginia, and Kentucky (Matthews et al., 2014, p. 395). About 86% of the respondents worked in community health organizations within their locality. These health centers mainly provided their services to low-income and uninsured people. The participants’ ages ranged from 40 years to 79 years. The overall study started in 2002 and ended in 2009. However, 14% of the respondents participated in the study for only two years (2004-2006). The study’s design was a cohort study that occurred between 2002 and 2009 (n=63,308).
Cox proportional hazard was the main statistical instrument used to estimate hazard ratios. It showed that the relationships analyzed in the study had more than a 95% confidence level. This confidence level was mainly applicable to all participants of each cohort entry (in relation to quartiles of physical activities and sedentary behaviors among respondents from the two ethnic groups sampled). The researchers came up with these findings after using age as the main time metric (Matthews et al., 2014). When formulating the findings, the researchers analyzed data regarding all respondents separately. Effect modification tests involved likelihood ratio tests to analyze the behavioral exposure of each racial cohort that caused death. This analysis excluded race-behavior interactions. The authors also assessed confounders (that affected physical activity) using backward modeling processes. Particularly, the researchers paid attention to model-building procedures that had more than a 10% estimation change (Matthews et al., 2014). The analysis further extended to include an assessment of dietary factors, such as energy nutrients and fats. Other confounding factors that they assessed using the same process included marital status, work status, and the diagnosis of life-threatening diseases/conditions (such as diabetes and hypertension). Based on their overall analysis, the researchers only included diabetes and employment status in their final models (Matthews et al., 2014). To make sure there was no possibility of pre-existing diseases clouding the results, the researchers conducted a sensitivity analysis that excluded participants that did not participate in a follow-up assessment, 12 months before the study began (Matthews et al., 2014).
Sometimes, background factors affected the quality of research findings because they introduce irrelevant factors that could affect the quality of factors (Rothman & Greenland, 2005). The main confounding effects identified by the researchers included “age, sex, race, educational level, annual household income, and marital status” (Matthews et al., 2014, p. 395). Other confounding factors included personal health, physical activities, lifestyle (drinking and smoking) and occupational factors.
To mitigate these confounding factors, the researchers made sure that most of the respondents were black people (two-thirds) (Matthews et al., 2014). Therefore, white respondents were a control group.
Effect modifiers differ from confounding factors because they are third-party factors that affect the relationship between primary exposures and the examinable effects (Brownson, Chriqui, Burgeson, Fisher, & Ness, 2010). Reverse causality is one such effect modifier. Particularly, the researchers paid attention to how racial differences could affect causality. For example, some respondents could not identify their races, as either “white” or “black.” To mitigate such effect modifiers, the researchers excluded such participants from the study. This excluded sample was 4.9% of the participants (4130 participants) (Matthews et al., 2014). Reverse causality could also occur if the researchers sampled respondents who had a history of illness. Therefore, the authors excluded respondents who suffered from heart disease, cancer and stroke. The researchers also excluded people who suffered from Parkinson’s disease, lupus, and multiple sclerosis from the study (Matthews et al., 2014). To minimize (further) the study’s effect modifiers, the researchers excluded the findings of respondents who had inclusive physical activity data. Similarly, they excluded respondents who had inadequate data regarding their sedentary lives from the final report.
The mortality rate among the respondents was 3613 (blacks) and 1394 (whites) (Matthews et al., 2014). Based on these statistics, the researchers found out that physical activity reduced the incidence of the main causes of adult mortality, such as cardiovascular diseases and cancer (Matthews et al., 2014). However, this finding was only unique to blacks because white men did not report a decreased incidence of cancer through increased physical activities. The study also found out that physical activity varied across different age groups, sex and educational levels (Matthews et al., 2014). Other factors that affected sedentary behaviors and physical exercising were smoking status, marital status, and employment status. Cancer and cardiovascular diseases were the leading causes of adult mortality. However, HIV, diabetes, and cerebrovascular diseases were other causes of death among the respondents (Matthews et al., 2014). However, many black people (than white people) died from these diseases. Diabetes, liver disease, and respiratory complications were other causes of death among white people, compared to black people. After adjusting for sex, body mass index (BMI), and other covariates in the study, the authors found out that a high level of physical activity was correlated to lower mortality rates among black people (across all causes of death) (Matthews et al., 2014). Generally, the authors found out that increased physical activity reduced the risk of death from diabetes and cardiovascular diseases by 24% (Matthews et al., 2014). However, the authors found that this percentage varied across different disease clusters. For example, they found out that there was a 19% reduction in the risk of developing cardiovascular diseases among black people. They also found out that there was a 24% decrease in the risk of developing cancer among this racial group (Matthews et al., 2014). Lastly, the researchers found out that the reverse trend was monotonic for all-cause mortality. However, cardiovascular and cancer diseases were exceptions. Based on these findings, the authors found that a sedentary life had a high correlation with an increased risk of all-cause mortality in both races. Similarly, their findings suggested that most public health campaigns should encourage people to engage in more activities that are physical because they increased their chances of living longer (Matthews et al., 2014).
Critical Analysis
Random errors often occur when a study’s findings are inconsistent with the findings of similar tests (Brownson et al., 2010). The term “random” comes from the unpredictability of the findings. Although most measurements are prone to random error, our case study shows that unpredictable dependent and independent variables could cause have caused inconsistent findings. Different interpretations of the instrumental reading could also have caused inconsistent findings. Partly, environmental factors could fan these problems. Precision concepts closely align with random errors. Stated differently, high precision measures often lead to lower variability levels.
Besides random errors, selection bias could also have affected the sampled results by lowering the possibility of randomization, which would have increased the probability of extrapolating the findings across a large geographical area (Matthews et al., 2014).
Using race as a classification metric in the study could potentially have undermined the credibility of the information obtained from the study because the researchers relied on self-reporting measures to classify people according to different racial profiles. Other researchers have experienced the same problem by noting inconsistencies in their findings. For example, Bryan, Tremblay, Perez, Ardern, and Katzmarzyk (2006) say Wolf and Walsh (two researchers) reported significant differences in physical reporting by race. Matthews et al. (2014) admit that this challenge could have affected the quality of their findings because although they found comparable validity for white and black males in the Southern United States, their validity coefficients were low. Part of the problem could have stemmed from measurement errors for analyzing physical activities and sedentary behaviors among the respondents. Research shows that such errors could easily attenuate observable risk estimates (Ibrahim, Alexander, Shy, & Farr, 1999).
The small number of deaths across both racial groups could also have undermined the credibility of the research findings because it lowers the statistical power for comparing the relationship between physical activities and adult mortality among blacks and whites. Nevertheless, more than 1,400 deaths noted among the respondents (about physical activities and sedentary lifestyles) provided a statistically significant correlation between the dependent and independent variables.
The discussion section of the paper adequately addressed the strengths and limitations of the study. For example, the representative cohort sample of more than 60,000 people is among the largest samples used to understand the relationship between physical activities and adult mortality (Matthews et al., 2014). The long follow-up time (6.4 years) also increases the study’s reliability because it provides an accurate measure of the effects of lifestyle choices on human health. This reliability emerged from the 5000 deaths reported during the study period (Matthews et al., 2014). Measures to control the confounding factors in the study also contributed to the study’s reliability because there was less interference from these factors. Demographic and dietary factors are a few confounding issues that the authors controlled this way. Using trained interviewers to sample some of the respondents also improved the credibility of findings received from the respondents because they were able to extract information that would have otherwise been difficult to obtain. Particularly, using these professionals was instrumental in assessing a broad range of physical activities that could affect the overall health of the respondents
Matthews et al. (2014) concentrated their study on the Southern United States. Although they used a largely representative sample of more than 60,000 adults, they did not select all the respondents randomly (Matthews et al., 2014). For example, they sampled many respondents who worked in the health sector. Therefore, it would be difficult to extrapolate the findings of this study to people from other economic sectors. Similarly, it would be difficult to generalize the findings of the study to people from other parts of the United States because the study mainly focused on understanding the relationship between physical exercises and adult mortality among black and white males in the southern United States. To generalize the findings across the country, the researchers would have to sample adult men across the country.
Given the study’s findings, the authors’ conclusions were appropriate. They found out that the decline in physical activities within America, partly contributed to the high number of adult mortalities in the country (Matthews et al., 2014). Similarly, they found out that the increase in chronic conditions within the country could stem from the decline in physical activities within the country (Matthews et al., 2014). In the same way, the study found out that there were insignificant differences in the above relationship, across both races (blacks and whites). Their findings provide much-needed empirical data to understand the importance of physical activities in promoting overall health standards.
Future studies should investigate the same relationship across other states in the country to find out if the same findings would suffice nationally. Using respondents from southern states alone is insufficient to draw up a holistic understanding of the relationship between physical activities and adult mortality among black men. Therefore, future studies should include respondents from other racial backgrounds. Particularly, these studies should investigate the relationship between the same dependent and independent variables among Hispanics because they live in (almost) similar conditions, and fall with the same economic cluster, as African Americans do. This way, there would be fewer confounding factors when investigating the study issue.
References
Brownson, R. C., Chriqui, J. F., Burgeson, C. R., Fisher, M. C., & Ness, R. B. (2010).Translating epidemiology into policy to prevent childhood obesity: The case for promoting physical activity in school settings. Annals of Epidemiology, 20(6), 436–444.
Bryan, S., Tremblay, M., Perez, S., Ardern, C., & Katzmarzyk, P. (2006). Physical Activity and Ethnicity Evidence from the Canadian Community Health Survey. Canadian Journal of Public Health, 97(4), 271-276.
Ibrahim, M., Alexander, L., Shy, C., & Farr, S. (1999). Cohort studies. ERIC Notebook. Web.
Matthews, C. E., Cohen, S. S., Fowke, J. H., Han, X., Xiao, Q., Buchowski, M. S., &… Blot, W. J. (2014). Physical activity, sedentary behavior, and cause-specific mortality in black and white adults in the Southern Community Cohort Study. American Journal of Epidemiology, 180(4), 394-405.
Rothman, K. J., & Greenland, S. (2005). Causation and causal inference in epidemiology. American Journal of Public Health, 95(1), 144-150.