401175 Analytic Approaches in Epidemiology: Health Survey

1.The data in the table below are taken from a national health survey, a representative cross-sectional prevalence survey estimating 12-month incidence of a range of risk factors for pancreatitis in addition to socio-demographic variables. Similar questions regarding diabetes status and pancreatitis were asked of participants. An extract of the data is given below, stratified by males and females.

Males

Females

Discuss statistical interaction and confounding by sex for the following effect measures:

Absolute measure

-Risk difference

Relative measures

-Risk ratio

-Odds ratio 

2. A prospective cohort study of individuals living in Western Sydney was conducted, investigating the effect of diabetes on the development of pancreatitis over a follow up period of 12 months. The data at the end of the follow-up period for males and females are given below.

Males

Females

Assume no loss to follow-up and that all participants were followed over the follow-up period.

Discuss statistical interaction and confounding by sex for the following effect measures:

-Risk difference

-Risk ratio

-Odds ratio

In your discussion compare your estimates with those from Question 1. Comment on similarities or differences on effect estimates, and likely reasons for these similarities or differences

3. Propose a causal diagram of the relationship between diabetes and cardiovascular, based on the variables provided above. Provide a narrative description (1 paragraph only) of the causal diagram incorporating findings from your analysis in Question 2.

1.

Retrospective measure

Male

Risk ratio= (7/998)/(28/2597)= 0.65 < 1

Odd Ratio = (7/991)/(28/2569)=0.648 <1

Risk difference= (7/998)-(28/2597) = 0.0038

Female

Risk ratio= (4/488)/(24/3949)= 1.34 > 1

Odd Ratio=(7/484)/(24/3925)=2.36 >1

Risk difference= (4/488)-(24/3925)= 0.0038= 0.002

Prospective measure

Male

Risk ratio= (41/675)/(9/450)= 3.03 >1

Odd Ratio=(41/634)/(9/441)=3.16 > 1

Risk difference= (4/634)-(9/441) = 0.0038= 0.014

Female

Risk ratio= (94/301) / (45/450) = 3.12 >1

Odd Ratio = (94/187) / (45/405) = 4.52 > 1

Risk difference= (4/634)-(9/441) = 0.0038 = 0.014

2.

As per the first retrospective study, the risk ratio in male is 0.65 which is less than 1 and the risk ratio in female is 1.34. Therefore, as per the first retrospective measure, the probability of having diabetes decreases with male patients with pancreatitis. However, the probability of having diabetes increases with female patients with pancreatitis. Therefore, there is a clear difference in likeliness of having diabetes between male and female patients. The risk differences for male and female are approximately similar. However, as per the prospective measure for both male and female the probability of having diabetes increases with female patients with pancreatitis. At the same time in both of these studies the risk differences of males and females are similar. It clearly indicates that, the first retrospective measure on male participants are biased. There are certain amount of confounding that made the first retrospective measure on male different from the other situation.

3.

It can be estimated that there are some patient who have other regulatory physical condition specially who are male that influenced the retrospective study. However, for both male and female the occurrence of case disease which was pancreatitis in retrospective study is less than the Prospective study. From the similar risk differences in the above risk differences found in first and the second study, it can be said that, the outcomes are significant and viable for the data monitoring. It should be also considered that, cardiovascular diseases are the common results of the pancreatitis. In first study, there might be some patients of cardiovascular disease within the control group, who did not have diagnosed with pancreatitis yet (Scirica et al., 2013). At the same time, in second study for both male and female the risk difference is similar which is 0.014. However, in first study the RD (risk differences) between male and female are slightly different. Therefore there are some obvious confounding in the first study. It can be also interpreted that, Cardiovascular disease is the physical issue that results from the pancreatitis.

References

Bellin, M. D., Beilman, G. J., Dunn, T., Pruett, T., Chinnakotla, S., Ngo, A., ... & Hering, B. J. (2013). Islet autotransplantation to preserve beta cell mass in selected patients with chronic pancreatitis and diabetes mellitus undergoing total pancreatectomy. Pancreas, 42(2), 317.

Bexelius, T. S., Ljung, R., Mattsson, F., & Lagergren, J. (2013). Cardiovascular disease and risk of acute pancreatitis in a population-based study. Pancreas, 42(6), 1011-1015.

Ewald, N., & Hardt, P. D. (2013). Diagnosis and treatment of diabetes mellitus in chronic pancreatitis. World journal of gastroenterology: WJG, 19(42), 7276.

Scirica, B. M., Bhatt, D. L., Braunwald, E., Steg, P. G., Davidson, J., Hirshberg, B., ... & Cavender, M. A. (2013). Saxagliptin and cardiovascular outcomes in patients with type 2 diabetes mellitus. New England Journal of Medicine, 369(14), 1317-1326.

Singh, S., Chang, H. Y., Richards, T. M., Weiner, J. P., Clark, J. M., & Segal, J. B. (2013). Glucagonlike peptide 1–based therapies and risk of hospitalization for acute pancreatitis in type 2 diabetes mellitus: a population-based matched case-control study. JAMA internal medicine, 173(7), 534-539.