Reference no: EM133292037

Between 1993 and 1995, investigators enrolled 1,000 pregnant women who worked outside the home and were seen at an obstetric clinic in New York City into a cohort study. Women currently experiencing depressive symptoms were ineligible for recruitment into the study. The study aim was to examine the association between how soon women returned to work following the birth of their child and post-partum depressive symptoms. Data on demographic, health status, and health behavior variables were collected at baseline.

Women were contacted 12 weeks post-partum to assess whether they were newly experiencing depressive symptoms and the amount of time they took to return to work. A total of 919 women were available for follow-up and were thus included in the analysis. The following data were obtained:

	Depressive Symptoms
Return to Work	Yes	No	Total
< 5 weeks	172	395	567
> 5 weeks	60	292	352
Total	232	687	919

1. Calculate and interpret the appropriate measure of association. The 95% confidence interval of this association was: (1.37, 2.31).

The investigators were concerned that household income level (assessed at baseline) might be influencing the observed association between when women returned to work following the birth of their child and post-partum depressive symptoms.

2. Draw a causal diagram to show the possible relationships between household income, return to work, and post-partum depressive symptoms.

The investigators stratified the data by household income level as follows:

	Lower Income Depressive Symptoms			Higher Income Depressive Symptoms
Return to Work	Yes	No	Subtotals	Yes	No	Subtotals	Total
<5 weeks	160	295	455	12	100	112	567
>  5 weeks	50	164	214	10	128	138	352
Total	210	459	669	22	228	250	919

3. Since a confounder must be associated with the outcome and exposure, calculate an odds ratio to evaluate the association between income level (3^rd variable) and amount of time before returning to work (exposure). Interpret this odds ratio.

Income	Returning to work
	<5 weeks	>5 weeks	TOTAL
Lower	(160+295) 455	(50+164) 214	669
Higher	(12+100) 112	(10+128) 138	250
TOTAL	567	352	919

4. Now evaluate whether income level is associated with incident post-partum depressive symptoms among those who returned to work >5 weeks using the appropriate measure of association.

Evaluating the association between income level and post-partum depressive symptoms among the non-exposed (returning to work ≥ 5 weeks):

	Depressive Symptoms
Income	Yes	No	Total
Lower	50	164	214
Higher	10	128	138
Total	60	292	352

5. The next step is to look at the stratum specific risk ratios.

1. Calculate the risk ratios for the association between returning to work and post-partum depressive symptoms in those with high income and those with low income. How do they compare?

1. The 95% confidence interval for the risk ratio calculated in question 5 among lower income women was 1.15-1.98. Among higher income women, the 95% confidence interval was 0.663-3.30. What conclusions can you draw with this additional information?

6. Using information from the previous questions, determine whether income confounds the observed association between when women returned to work following the birth of their child and post-partum depressive symptoms. In your answer, comment on the: a) criteria a variable must meet to be considered a confounder; b) the stratum-specific risk ratios; and c) the magnitude (expressed as a percent) of the difference between the crude and the stratum-specific risk ratios.

7. Might there still be residual confounding by income even after controlling for high and low income? Why? How could you improve control for income?

8. Can you think of other factors that might explain the observed association between returning to work and post-partum depression?

9. If women who returned to work in less than five weeks after giving birth were less likely to maintain breast feeding and breast feeding is known to impact post-partum symptoms, is breast feeding a likely confounder between time to returning to work and post-partum depression? draw a diagram!