Assessing the Role of BiasThe role of diet in promoting or preventing disease is a focus of nutritional epidemiology. In recent years, the field has embraced the idea that dietary patterns, as opposed to a single food or nutrient, are likely to have a bigger impact on health and disease prevention. The Mediterranean diet is characterized by the traditional food patterns of Mediterranean countries. The Mediterranean diet has consistently been shown to reduce the risk of all-cause mortality, cardiovascular disease mortality, and cancer mortality in several large prospective cohort studies. Suppose that we work for the American Cancer Society, which would like to update its “Guidelines on Nutrition and Physical Activity for Cancer Prevention.” Guidelines are based on current evidence, and their purpose is to provide recommendations for the general public. We are tasked with critically evaluating one recent study that examined the association between Mediterranean diet and risk of ovarian cancer in the Nurses’ Health Study, a prospective cohort study with 121,700 female nurses (Xie et al, Journal of Ovarian Research (2014) 7:112).In this study, dietary information was measured prospectively every 2 to 4 years by a self-administered semi-quantitative, 131-item food frequency questionnaire. For each food, a portion size was given, and women were asked to choose from 9 intake frequencies from never to >=6 servings per day averaging over the prior year’s intake. A Mediterranean Diet Score was calculated that considered consumption of certain fatty acids, legumes, cereals, fruits, nuts, vegetables, meat, dairy, and alcohol and ranged from 0 to 10.The following excerpts are taken from the paper. 1. “The validity of our food frequency questionnaire was evaluated by comparison with food diaries in 192 women.” Which type of bias are the study investigators trying to limit?2. If the correlation between the food frequency questionnaire and food diaries is poor, is it more likely or less likely that this type of bias is present? 3. Ovarian cancers were identified either by self-reports or data from the National Death Index. Women or their next-of-kin were asked for permission to obtain and review pathology reports to confirm the diagnosis. Which type of bias are the study investigators trying to limit by confirming the ovarian cancer diagnosis?Essentials of Epidemiology in Public Health, Fourth EditionAnn Aschengrau and George SeageIn-Class ExerciseCopyright 2020 by Jones & Bartlett Learning, LLC, an Ascend Learning Company 24. “The follow-up rate was 97%.” Which type of bias are the investigators trying to assess?5. Based on the follow-up rate, is this type of bias likely?6. Is the “Healthy Worker Effect” a concern in this study? 7. “Participants who left more than 70 items blank on the diet questionnaire were excluded from the analysis.” Which types of bias were the study investigators trying to limit?8. “Participants who left more than 70 items blank on the diet questionnaire were excluded from the analysis.” Did excluding these women create selection bias?Essentials of Epidemiology in Public Health, Fourth EditionAnn Aschengrau and George SeageIn-Class ExerciseCopyright 2020 by Jones & Bartlett Learning, LLC, an Ascend Learning Company 39. “All of this [dietary] information was collected prospectively…” What type of bias was avoided with prospective dietary data collection? 10. Examine the results in the table below. Suppose that a Mediterranean diet truly reduces the risk of ovarian cancer. What sources of bias most likely accounted for the null or nearly null findings? Hazard ratios for the association between Mediterranean Diet Score and ovarian cancer Mediterranean Diet Score<2.6 2.7-3.5 3.6-4.5 4.6-5.5 >5.5 P trendN cases 134 119 148 136 159Person-years 290,089 247,467 329,112 293,048 309,867Age-adjusted hazard ratio1.0 (ref) 0.91 0.85 0.86 0.93 0.52Multivariable-adjusted hazard ratio1.0 (ref) 0.90 0.83 0.85 0.91 0.44Essentials of Epidemiology in Public Health, Fourth EditionAnn Aschengrau and George SeageIn-Class ExerciseCopyright 2020 by Jones & Bartlett Learning, LLC, an Ascend Learning Company 4Confounding and Random ErrorPhysical activity is the centerpiece of many strategies to prevent and reduce chronic disease and improve health. The US Department of Health and Human Services (HHS) sets Physical Activity Guidelines for Americans based on the current body of evidence. Adults are encouraged to engage in 2 hours and 30 minutes of moderate intensity physical activity per week; children and adolescents are encouraged to participate in 60 minutes of physical activity daily.Suppose that we work for HHS, which is interested in understanding how body size and physical activity during different periods in adolescence may impact health. We are tasked with critically evaluating a recent study that examined the association between adolescent body size, physical activity, and risk of breast cancer (Oh et al. Interaction between early life body size and physical activity on risk of breast cancer. Int J Cancer 2015; 137:571-581). This investigation was a prospective cohort study of female nurses aged 25-42 who were free of breast cancer at enrollment in 1989. In 1997, participants were asked to report their childhood and adolescent body size (at ages 5, 10, and 20) using a 9-level figure drawing (1 being very thin and 9 being overweight). Participants have been followed-up since 1989 for incidence of breast cancer. 1. Based on the description in the introductory paragraph, which potential confounding variables were controlled through the use of restriction? 2. The following table shows some baseline characteristics of the study population according to average childhood body size. The characteristics listed are considered possible risk factor for breast cancer.Average Body Size at ages 5-10 yearsCharacteristic 1 1.5-2 2.5-3 3.5-4 4.5+Age (years) 42.5 41.8 42.0 42.3 42.5Birth weight >=3.9 kg, % 10.0 11.5 13.9 16.2 18.7Age at menarche 12.8 12.6 12.3 12.1 12.0Total physical activity at ages 12-17 (MET-hr/wk)56.3 54.5 51.9 48.4 46.0Age at first birth (years) (among parous women)26.3 26.5 26.5 26.5 26.5Family history of breast cancer, %9.6 9.7 9.3 9.3 9.5Essentials of Epidemiology in Public Health, Fourth EditionAnn Aschengrau and George SeageIn-Class ExerciseCopyright 2020 by Jones & Bartlett Learning, LLC, an Ascend Learning Company 5State which variables are potential confounders and which variables are not potential confounders according to these data. Give the rationale for your answer.3. In the multivariate regression, investigators compare participants with body size=1 (index group) to participants with body size ≥ 4.5 (reference group). Table 1 shows the crude and adjusted risk ratios. The age-adjusted RR controlled only for age. The MV-adjusted (multivariable-adjusted) RR controlled for many potential confounders at the same time.Crude and adjusted RR for the association between adolescent body size and breast cancerRR for breast cancer 95% CICrude 1.79 1.28-2.19Age-adjusted 1.64 1.31-2.06MV-adjusted 1.64 1.29-2.033. a. Which of the following is the best interpretation of the crude RR?i. In these data, women with adolescent body size≥ 4.5 had 1.79 times the risk of breast cancer compared to those with adolescent body size=1 over the study period.ii. In these data, women with adolescent body size=1 had 1.79 times the risk of breast cancer compared to those with adolescent body size≥ 4.5 over the study period, adjusting for numerous potential confounders.iii. In these data, women with adolescent body size=1 had 1.79 times the crude risk of breast cancer compared to those with adolescent body size≥ 4.5 over the study period.3. b. Which of the following is the best interpretation of the age-adjusted RR?i. In these data, women with adolescent body size=1 had 1.64 times the crude risk of breast cancer compared to those with adolescent body size≥ 4.5 over the study period.ii. In these data, women with adolescent body size=1 had 1.64 times the risk of breast cancer compared to those with adolescent body size≥ 4.5 over the study period, adjusting for age.Essentials of Epidemiology in Public Health, Fourth EditionAnn Aschengrau and George SeageIn-Class ExerciseCopyright 2020 by Jones & Bartlett Learning, LLC, an Ascend Learning Company 6iii. In these data, women with adolescent body size≥ 4.5 had 1.64 times the risk of breast cancer compared to those with adolescent body size=1 over the study period, adjusting for age.iv. In these data, women with adolescent body size=1 had 1.64 times the risk of breast cancer compared to those with adolescent body size≥ 4.5 over the study period, adjusting for numerous potential confounders.3. c. State in words your interpretation of the 95% confidence interval for the MV-adjusted RR. 3. d. Based on the 95% CI alone, is the MV-adjusted RR statistically significant? 3. e. What is the magnitude of confounding by this collection of potential confounders?3. f. Is this collection of potential confounders actually confounding the association between adolescent body size and risk of breast cancer?a. Yesb. Noc. Maybe
