Measure of association for the cohort study
answer the following questions in a 3-page paper: 1. A doctoral student is conducting a non-concurrent cohort study on the risk factors of a chronic disease. Tables 1 and 2 provide a summary of data that was collected on the 19,753 patients. Table 1 present the results of three Factors (A17, B24, C32)—note that these are just variable names—and they could represent variables such as age, race, sex, etc. Table two presents results by smoking level. Table 3 provides a summary of the multivariate analysis of the three variables and—in this case since the outcome is dichotomous (i.e. either you have the disease or you don't) we can use logistic regression model to study mulitple factors (such as A17 and different smoking levels) on the outcome and come up with adjusted Odds Ratios. Table 1 Risk Factor Present Risk Factor Absent Factor Total Patients with Chronic Disease Total Patients with Chronic Disease Factor A17 2785 26 16968 170 Factor B24 2009 32 17744 164 Factor C32 641 13 19112 183 Table 2 Total Patients Number of Patients with Chronic Disease Smoking None 1–10 cigarettes/day 11–20 cigarettes/day 21–30 cigarettes/day >30 cigarettes/day 5030 6061 4572 2999 1091 15 21 43 56 61 Table 3 Multivariate Analysis Factor Adjusted OR 95% CI Factor A17 0.86 0.55–1.29 Factor B24 1.50 0.97–2.27 Factor C32 1.98 1.03–3.50 A. Which measure of association would you use for this study? Why? B. Using bivariate analysis, calculate the measure of association between the chronic disease and the following theoretical risk factors: A17, B24, C32, and smoking (at the different exposure levels). Provide 95% confidence intervals. C. In narrative form, provide a description of the results for the bivariate and multivariate analysis. D. Provide discussion and interpretation of the results from the bivariate and multivariate analysis. For the multivariate analysis explain why Odds Ratios are being used with Cohort data. 2. A second doctoral student is conducting a case control study on risk factors for an infectious disease and has collected the data below. Using bivariate analysis, calculate the measure of association between the disease and risk factor studied for each level of exposure. Provide 95% confidence intervals. Provide an interpretation of the results. Factor BF Case (n=186) Control (n=277) Not Received Brand Y Brand X—50 mg Brand X—100 mg Brand X—200 mg 15 11 28 46 86 78 35 70 50 44 *If you do not have access to a statistical software package such as SPSS, the following are online calculators that can be used for this assignment: 1) Odds Ratio Calculator, 2) Relative Risk Calculator or An optional calculator is available at http://statpages(dot)org/ctab2x2.html A Brief Introduction to Epidemiology - III (Basic Statistics & Common Epidemiologic Measures). Super Course. Retrieved July 3, 2013, at: http://www(dot)pitt(dot)edu/~super1ecture/lec0891/index.htm Spitalnic S (2006). Risk Assessment II: Odds Ratio. Hospital Physician, 2006 Jan: 23-26. Retrieved May 28, 2012, at: http://www(dot)turner-white(dot)com/memberfile.php?PubCode=hp_jan06_odds.pdf Tripepi, G, Jager, KJ, Dekker, FW, Wanner, C, Zoccali, C (2007). Measures of effect: Relative risks, odds ratios, risk difference, and 'number needed to treat'. Kidney International, 72(7), 789-91. Retrieved May 28, 2012, at: ProQuest Medical Library database. (Document ID: 1338652401). Logistic Regression - Overview Retrieved May 28, 2012, at: http://www(dot)ucdmc(dot)ucdavis(dot)edu/ome/mcrtp/docs/biostatIV.ppt Estimating the Relative Risk in Cohort Studies and Clinical Trials of Common Outcomes Louise-Anne McNutt1, Chuntao Wu1, Xiaonan Xue2 and Jean Paul Hafner3 Am. J. Epidemiol. (2003) 157 (10): 940-943. Retrieved May 28, 2012, at: http://aje(dot)oxfordjournals(dot)org/content/157/10/940.full Additional Reading Croen, L, Grether, J.K., Selvin, S. (2002). Descriptive Epidemiology of Autism in a California Population: Who is at Risk? Journal of Autism and Developmental Disorders, 32(3). Retrieved May 28, 2012, at ProQuest Haselkorn, T, Bernstein, L, Preston-Martin, S, Cozen, W. Mack, W.J. (2000). Descriptive Epidemiology of Thyroid Cancer in Los Angeles County, 1972-1995. Cancer Causes and Control, 11(2). Retrieved May 28, 2012, at ProQuest.
Case
Name
Course
Instructor
Date
A. Measure of association for the cohort study
As a form of observational study, it is necessary to understand the risk factors for the disease and determine the absolute risk of the subjects. Since the study focused on possible causes through the risk factors, relative difference measures of association are applicable including the relative risk/ rate and the relative odds/ odds ratio. As the study also focuses on the risk factors that affect the chronic disease, the ratio measures help in identifying whether the disease is associated with the exposures. Furthermore, it is necessary to assess whether the association is strong. The odds ratio (OR) allows comparison on different risk factors in the cohort study. The odds ratio, risk ratio and the incident rate ratio indicate the measures of effect between exposure and outcome (Tripepi et al., 2007). An OR that is less than 1.0 indicates that risk of the outcome is lower among exposed people in comparison to the unexposed. On the other hand when OR is greater than 1.0, then the odds are higher for the exposed individuals (Tripepi et al., 2007).
B
Value95% conf intervalFactor A17Odds ratio (OR)0.9310.6011.433Factor B24Odds ratio (OR)1.7351.1622.578Factor C32Odds ratio (OR)2.1411.1593.874
value 95% conf intervalNone1-10 cigarettes/ day0.4670.230.95411-20 cigarettes/ day1.2750.6852.40421-30cigarettes/ day2.5561.4014.729>20 cigarettes/ day7.9564.37514.676C. The values of the bivariate and multivariate analysis are almost similar, but the risk factor of smoking focuses on different levels of smoking. The odds ratio takes into account the association. The OR shows that the risk that an outcome occurs upon exposure, while also comparing outcome in cases when there is no exposure.
D. The multivariate analysis results focus on more than two variables simultaneously in comparison to bivariate analysis which focuses on two variables. Hence, multivariate analysis is an extension of bivariate analysis, where there is more than one multivariate analysis. In interpreting the results of the OR in bivariate and multivariate analysis, the odds of exposure showed where there is an association between the dependent and independent variable(s). One is likely to get the disease upon exposure to Factor B24 that Factor A17 and exposure to Factor C32 increases risk of the chronic disease more than the other two factors.
When the OR is below 1, then the exposure does not influence odds of outcome, while an OR above 1 increases odds of outcome upon exposure. In Factor A17, the adjusted OR for multivariate analysis is 0.86, showing that the factor has a lower likelihood of influencing the odds of the chronic diseases. On the other hand the adjusted OR for B24 is 1.5, showing that exposure is associated with the risk...