Wirback T, Möller J, Larsson JO, Galanti MR, Engström K. Social factors in childhood and risk of depressive symptoms among adolescents–a longitudinal study in Stockholm, Sweden. Int J Equity Health. 2014 Nov 11;13:96.
1) What are the main exposures of interest? What is the outcome? What is the study design?
2) What potential effect measure modifier(s) are the authors concerned about?
3) Summarize the findings of this study in your own words.
4) What “real mechanistic features in the causal framework(s)” could be potentially causing the differential effects the authors report on?
5) Table 4 shows the main exposures of interest by each category of the effect measure modifier. The authors calculated a “synergy index” which quantifies departure from additivity.
Here’s how synergy indices are calculated:
Synergy index = RR11 – 1
(RR10 – 1) + (RR01 -1)
Where:
RR11 = exposed to exposure AND effect modifier
RR10 = exposed to exposure but NOT effect modifier
RR01 = exposed to effect modifier but NOT exposure
1 is the null value for the synergy index. Values above 1 indicate synergy.
Show how the synergy index for “Parental employment status” was calculated.
6) Which exposures appear to have a statistically significant synergistic departure from additivity? (use Table 4)
7.) Examine whether there is multiplicative gender interaction with each of each social factors listed in Table 4. Show calculations state whether you think there is a departure from multiplicity, and if there is state whether you think it is synergistic or antagonistic.
8) The authors state: “Potential misclassification of outcome may have occurred because of inadequate ability among adolescents to recognize and report symptoms, and this misclassification might theoretically differ between socioeconomic groups but particular between boys and girls.” What phenomenon are they talking about here?
9) Do you have concerns about potential selection bias in this study?
10) Why did the authors adjust for gender in Tables 2 and 3?
Sample Solution
