Comparison of placenta consumers’ and non-consumers’ postpartum depression screening results using EPDS in US community birth settings (n=6038): a propensity score analysis

Table 3 Adjusted comparisons for 2 definitions of postpartum depression as measured by the Edinburgh Postnatal Depression Scale (EPDS), comparing women who ate their placenta to women who did not, using bootstrapped propensity-score matched data from MANA Stats, birth years 2016–2018. Absolute risks (proportions), adjusted odds ratios, and the corresponding 95% confidence intervals were calculated from the bootstrapped samples

Outcome	Ate Placenta % (95% CI)	Did not eat placenta % (95% CI)	Model 1^a aOR (95% CI)	Model 2^b aOR (95% CI)
Main analysis	n = 1876	n = 1876
EPDS ≥ 11	9.9% (9.6 – 10.2)	8.4% (7.8 – 9.0)	1.20 (1.10 – 1.31)	1.19 (1.09 – 1.30)
EPDS ≥ 13	5.5% (5.3 – 5.8)	4.8% (4.4 – 5.3)	1.15 (1.03 – 1.29)	1.14 (1.02 – 1.28)
Assuming unknown is yes	n = 2532	n = 2532
EPDS ≥ 11	9.2% (8.8 – 9.6)	7.8% (7.3 – 8.3)	1.19 (1.09 – 1.29)	1.18 (1.09 – 1.28)
EPDS ≥ 13	5.3% (4.9 – 5.6)	4.5% (4.1 – 4.8)	1.19 (1.06 – 1.33)	1.17 (1.05 – 1.31)
Assuming unknown is no	n = 1932	n = 1932
EPDS ≥ 11	10.0% (9.7 – 10.3)	8.9% (8.1 – 9.6)	1.13 (1.03 – 1.25)	1.13 (1.03 – 1.25)
EPDS ≥ 13	5.6% (5.3 – 5.8)	5.1% (4.5 – 5.6)	1.10 (0.98 – 1.25)	1.10 (0.97 – 1.25)
Only counting raw	n = 1080	n = 1080
EPDS ≥ 11	10.8% (10.5 – 11.2)	8.3% (7.2 – 9.4)	1.33 (1.17 – 1.56)	1.34 (1.16 – 1.57)
EPDS ≥ 13	6.3% (5.9 – 6.5)	4.8% (4.0 – 5.7)	1.31 (1.08 – 1.61)	1.30 (1.07 – 1.61)

^aModel 1 used propensity score-matched data, and controlled additionally for the actual propensity scores, to adjust for any residual confounding subsequent to the frequency matching procedure
^bModel 2 controlled additionally for “significant” psychosocial history (see footnote to Table 1), history of anxiety or depression treated with drugs or inpatient therapy, and history of anxiety or depression not treated with drugs or inpatient therapy. These variables were all included in the propensity score model, and based on Table 1, indeed ended up equally matched between the two groups, but always the possibility of residual confounding, especially for such important covariables

ISSN: 1471-2393