Table 2. Quality^a of studies included in systematic review of maternal depression and child growth in developing countries, 1996–2010.

Study design	Selection			Exposure	Comparability
	Representative study sampleb	Response rate ≥ 80%c		Depression assessed using diagnostic testd	Adjustment for 2 or more demographic variablese
Prospective longitudinal cohort
Patel et al. 200331	No	Yes		No	No
Rahman et al. 200433	Yes	Yes		Yes	Yes
Tomlinson et al. 200635	Yes	No		Yes	No
Santos et al. 201038	Yes	Yes		No	Yes
Case–control
Adewuya et al. 200834	No	NA		Yes	Yes
Anoop et al. 200429	NA	NA		Yes	Yes
Baker-Henningham et al. 2003f,25	No	Yes		No	Yes
Carvalhaes et al. 200236	Yes	Yes		No	Yes
de Miranda et al. 199637	No	NA		No	Yes
Rahman et al. 200432	No	NA		No	Yes
Cross-sectional
Black et al. 2009g,24	No	No		No	Yes
Harpham et al. 200530	Yes	Yes		No	Yes
Harpham et al. 200530	Yes	Yes		No	Yes
Harpham et al. 200530	Yes	Yes		No	Yes
Harpham et al. 200530	Yes	No		No	Yes
Stewart et al. 200823	No	Yes		No	Yes
Surkan et al. 200839	Yes	Yes		No	Yes

NA, not available (data were either incomplete or not reported); OR, odds ratio.

^a Study quality was assessed using a checklist adapted from the Newcastle–Ottawa Scale for assessing the quality of nonrandomized studies in meta-analyses.

^b Women and children included in the study were representative of the community.

^c Individuals who refused to participate and those lost to follow-up were included in calculating the response rate.

^d Individuals who were not assessed using a diagnostic test for depression were assessed using a measure of depressive symptoms or of common mental disorders.

^e The analysis of the relationship between maternal depression and child growth was adjusted for at least two demographic variables.

^f In this study, multivariate adjustments were made for more than two demographic variables. However, the data used in our meta-analysis were based on crude estimates because maternal depressive symptoms were not included in the final adjusted model due to the stepwise procedure.

^g In this study, multivariate adjustments were made for growth outcomes as continuous variables but only the crude OR was presented in the paper. For the meta-analysis, ORs were based on an adjusted analysis of data obtained from the authors of the original study.