Relationship between depression and diabetes in pregnancy: A systematic review

Table 1 Outcomes of included studies involving women with gestational diabetes

First author,study design	Definition/measures of depression	Timing of depression measures	Overall, nsubgroups, n	Main outcomes/findings
Abdollahi[20] Prospective, cohort	EPDS ≥ 12	Within 12 wk after delivery	n = 1449	Women with GDM had greater risk of postpartum depression than women without GDM [adjusted OR (95%CI): 2.93 (1.46-5.88), P = 0.002]
1Bener[23] Prospective, cross-sectional	EPDS ≥ 12	Within 6 mo after delivery	n = 1379 With depression, n = 243; Without depression, n = 1136	Prevalence of GDM was numerically, but not statistically, higher in women with depression (9.9%) vs women without depression (6.2%) (P = 0.051)
Berger[25] Retrospective	EPDS ≥ 13 or did not answer “No” to self-harm question	Within 4 d after delivery	Unselected, n = 322 History of mental illness, n = 215	In the unselected group, prevalence of GDM was higher in women with postpartum depression (27.3%) vs women without depression (9.0%) (P = 0.04); there was no difference in the group with previous mental illness (19.4% vs 10.2%, P = 0.14) In the unselected group, GDM was associated with postpartum depression [OR (95%CI): 12.1 (1.9-77.8)] In the unselected group, overall prevalence of depression and GDM was 0.9% (3 of 322)
Bisson[26] Prospective, case-control	EPDS ≥ 10	Approx. 30 wk gestation	n = 52 GDM, n = 26; No GDM, n = 26	Women with GDM had a greater prevalence of depressive symptoms vs women without GDM (23% vs 0%, P = 0.023) Mean (SD) EPDS score was 6.8 (4.0) for women with GDM and 4.2 (2.6) for women without GDM (P < 0.05)
Blom[27] Prospective	EPDS > 12	2 mo after delivery	n = 4941 With depression, n = 396; Without depression, n = 4545	No significant difference in the proportion of women with GDM between those who did (4/396; 1.0%) and did not (28/4545; 0.6%) have depression (P≥ 0.05) Calculated prevalence of women with both GDM and depression = 0.08% (4/4941)
Bowers[28] Retrospective	ICD9 codes 296.2, 296.3, and 311	Coded on medical history or hospital discharge record	n = 128295 With depression, n = 5815 (medical history, n = 5350); Without depression, n = 122480	Women with history of depression were more likely to have GDM than women without history of depression (5.4% vs 4.3%; P value NR) Depression was associated with significantly increased risk of GDM [OR (95%CI): adjusted for age, race/ethnicity, study site, insurance, and parity: 1.42 (1.26-1.60)]; similar results when restricted to women with history of pre-pregnancy depression [adjusted OR (95%CI): 1.36 (1.20-1.54)] Calculated prevalence of coincident GDM and depression was 313 of 128295 (0.24%)
1Burgut[29] Prospective, cross-sectional	EPDS ≥ 12	Within 6 mo of delivery	n = 1379 Qatari women, n = 837 Other Arab women, n = 542 With depression, n = 243 With history of diabetes, n = 310	GDM increased risk of depression in Qatari women [adjusted OR (95%CI): 1.65 (1.02-2.69)], but not in other Arab women [1.09 (0.63-1.91)]
Chazotte[31] Prospective	CES-D ≥ 16	Weeks 34-36 of gestation	n = 90 GDM, n = 30; High risk of preterm birth, n = 30	56.7% of women with GDM had CES-D ≥ 16; this was not significantly different vs women at low (33.3%) or at high (70%) risk of preterm birth (P≥ 0.05) Mean (SD) CES-D score was 17.0 (9.1) for women with GDM, 20.9 (9.4) for women at high risk of preterm birth, and 13.7 (7.5) for women at low risk of preterm birth (P≥ 0.05)
Crowther[33] RCT	EPDS ≥ 12	3 mo after delivery	Low risk of preterm birth, n = 30 n = 1000 Intervention2, n = 490; Routine care, n = 510	Significantly lower proportion of women in the intervention group (8%; 23/278 respondents) had EPDS indicative of depression vs women in the routine care group (17%; 50/295 respondents) (P = 0.001)
Dalfra[34] Prospective	CES-D ≥ 16	3rd trimester and 8 wk after delivery	n = 245 GDM, n = 176 (treated with diet, n = 109; treated with insulin, n = 68); No DM, n = 39	Mean (SD) CES-D scores at 3rd trimester were 17.0 (8.6) among women with GDM and 18.0 (8.7) among women without DM (P = 0.52) Mean (SD) CES-D scores at 3rd trimester were 16.6 (8.1) among women with GDM treated with diet and 17.7 (9.4) among women with GDM treated with insulin (P = 0.58) The severity of depressive symptoms increased from the 3rd trimester to after delivery in women with GDM [estimated mean difference in CES-D score (95%CI): 5.7 (4.2-7.3)], but decreased in women without DM [2.7 (-5.9-0.5); P < 0.0001 between groups]
Daniells[35] Prospective, longitudinal	MHI-5 ≥ 16	Weeks 30 and 36 of gestation, and 6 wk after delivery	n = 100 GDM, n = 50; No GDM, n = 50	Significantly higher proportion of women with GDM (30%) were depressed at Week 30 vs women who did not have GDM (12%) [OR (95%CI): 3.14 (1.1-8.94), P = 0.03]; however, there was no difference at Week 36 or after delivery (P≥ 0.05) Mean (SD) MHI-5 scores: Week 30: GDM, 13.9 (4.8); no GDM, 11.4 (3.8), P = 0.004; Week 36: GDM, 10.9 (3.8); no GDM, 11.7 (4.0), P = 0.31; postpartum: GDM, 11.5 (4.5); no GDM, 11.7 (4.0), P = 0.79 No significant difference in MHI-5 scores in women who were being treated with insulin (n = 7) compared with those being managed with diet only (P = 0.06; MHI-5 scores NR)
de Wit[36] Analysis of baseline RCT data	WHO-5 < 50	Early pregnancy (< 20 wk)	n = 98 obese women Depressed, n = 26	Prevalence of GDM was 13.5% of total sample of obese women and 19.2% of the subgroup with depression (NS; P value NR)
Ertel[37] Prospective, cohort	EPDS ≥ 15	Early pregnancy (< 20 wk)	n = 934	No significant association between depressive symptoms in early pregnancy and GDM measures at mid-pregnancy [adjusted OR (95%CI): for abnormal glucose tolerance associated with depression: 1.34 (0.81-2.23); for impaired glucose tolerance associated with depression: 1.53 (0.73-3.22)]
Huang[38] Prospective, cohort	EPDS ≥ 13	Mid-pregnancy (median 27.9 wk) and 6 mo (median 6.5 mo) after delivery	Prenatal, n = 2112 Postpartum, n = 1686	Prevalence of GDM was 8% among women with prenatal depression, 6% among women without prenatal depression, 7% among women with postpartum depression, and 5% among women without postpartum depression Compared with women with normal glucose tolerance, the odds of prenatal depression were significantly higher in women with isolated hyperglycemia [adjusted OR (95%CI): 1.80 (1.08-3.00)], but not in women with impaired glucose tolerance [1.43 (0.59-3.46)] or GDM [1.45 (0.72-2.91)] There was a 25% higher odds of prenatal depression per SD increase (27 mg/dL) in glucose levels [OR (95%CI): 1.25 (1.07-1.48)] Pregnancy hyperglycemia was not associated with significantly higher odds of postpartum depression
Jovanovic[39] Retrospective, claims database	ICD-9 codes 311, 296.2, 296.3, 300.4, 301.12, 309.1	Not specified, but data spanned from 21 mo before to 3 mo after delivery	n = 839792 GDM, n = 52848 No DM, n = 773751	Prevalence of depression among women with GDM was 5.3% Relative risk (95%CI): of depression in women with GDM vs women with no DM was 1.17 (1.12-1.21) Prevalence of concurrent GDM and depression was 0.4%
Katon 2011[41] Cross-sectional analysis of prospective cohort	PHQ-9	3rd trimester	n = 2398 GDM, n = 425; No DM, n = 1747	Prevalence (95%CI): of probable major depression among women with GDM was 4.5% (2.5%-6.4%) by PHQ-9 score, 5.7% (3.5%-7.9%) by antidepressant use, and 8.7% (6.0%-11.4%) by either PHQ-9 or antidepressant use, compared with the prevalence among women without DM [PHQ-9: 4.1% (3.2%-5.1%); antidepressants: 6.2% (5.1%-7.3%); PHQ-9 and antidepressants: 9.6% (8.2%-11.0%)] After adjusting for demographic characteristics, chronic medical conditions, and pregnancy variables, GDM was not associated with major [OR (95%CI): 0.90 (0.61-1.32)] or any [OR (95%CI): 0.95 (0.68-1.33)] antenatal depression
Katon 2014 (VA)[40] Retrospective, VA database	ICD-9 codes 296.2-296.39	Up to date of delivery	n = 2288 GDM, n = 118 No GDM or hypertensive disorder, n = 1966	Prevalence of depression was 9.3% in women with GDM and 8.8% in women without DM (no statistical analysis)
Katon 2014 (PPD)[42] Retrospective, hospital database	PHQ-9	2nd or 3rd trimester and 6 wk after delivery	n = 1423	Prevalence of GDM did not differ between women with postpartum depression (19.3%) and women without postpartum depression (20.7%) (P = 0.89) GDM was not a risk factor for postpartum depression [OR (95%CI): 0.68 (0.40-1.13), P = 0.13] Prevalence of concurrent GDM and depression was 1.12%
Keskin[43] Prospective, cohort	BDI ≥ 17	24-28 wk gestation	n = 89 GDM, n = 44 No GDM, n = 45	Prevalence of depression did not differ between women with GDM (80%) and women without GDM (83%) (P = 0.4)
Kim[44] Prospective, longitudinal	CES-D (cut-off NR)	Week 12-20 of gestation and 8-12 wk after delivery	n = 1445 GDM, n = 64; No GDM, n = 1233	No difference in the proportion of women with depressive symptoms in the GDM (14.1%) vs no GDM (13.5%) group (P > 0.05) After adjustment, GDM was not associated with an increase in depressive symptoms between pregnancy and postpartum [adjusted OR (95%CI): 1.22 (0.54-2.77)] Calculated prevalence of both GDM and depression = 0.62%
Ko[45] (Korean) Prospective, cohort	Postpartum depression model (dissertation by JI Bae, Ewha Womans University)	Weeks 24 and 28 of gestation	n = 68 Coaching program group, n = 34 Control group, n = 34	Women with GDM who participated in a 4-wk educational coaching program had a greater decrease in depression scores [mean (SD) change from baseline: -3.77 (6.50)] than women with GDM who did not participate in the program [mean (SD) change from baseline: 1.23 (6.76)] (P = 0.043)
Kozhimannil[46] Retrospective, cohort	ICD9 codes 296.2, 296.3, 300.4, 301.12, 309.1, and 311	During the 6 mo before and up to 1 yr after delivery	n = 11024 With GDM, n = 346 (taking insulin, n = 163); No DM, n = 10367	Prevalence of depression in women with GDM taking insulin was 16.0% vs 13.7% among women with GDM not taking insulin (P value not reported) Relative to women without diabetes, risk of depression was higher in both women with GDM taking insulin [adjusted OR (95%CI): 1.85 (1.19-2.87)] and in women with GDM not taking insulin [adjusted OR (95%CI): 1.69 (1.09-2.62)]
Levy-Shiff[66] Prospective	BDI	2nd trimester	n = 153 GDM, n = 51 No DM, n = 49	No significant difference in depression during 2nd trimester between GDM [mean (SD) BDI score 6.70 (4.46)] and controls [6.59 (5.88), P≥ 0.05] For sample as a whole, higher levels of cognitive assessment of pregnancy as a challenge was associated with lower depression (P < 0.05)
Liu[47] Prospective	Survey asking if diagnosed or discussed with HCP	Postpartum (mean 9.7 mo)	n = 3748 White, n = 1043 Asian/Pacific Islander, n = 425 Hispanic, n = 1253 Black, n = 1027	Prevalence of GDM was 7.6% in white (P < 0.05 vs all other ethnic groups), 14.9% in Asian/Pacific Islander (P < 0.05 vs other ethnic groups), 10.1% in Hispanic (P < 0.05 vs white and Asian/Pacific Islander groups), and 10.1% in black (P < 0.05 vs white and Asian/Pacific Islander groups) populations Prevalence of pre-existing depression was 2.8% in white (P < 0.05 vs all other ethnic groups), 12.4% in Asian/Pacific Islander (P < 0.05 vs all other ethnic groups), 7.6% in Hispanic (P < 0.05 vs all other ethnic groups), and 5.5% in black (P < 0.05 vs all other ethnic groups) populations No association between GDM and PPD; African Americans with GDM had decreased likelihood of PPD compared with those without GDM [OR (95%CI): 0.1 (0.0-0.5)] Weighted percentage of women with PPD with or without GDM was 10% vs 7.5% in white women (P < 0.05), 18.6% vs 14.4% in Asian/Pacific Islander (P≥ 0.1), 13.8% vs 9.8% in Hispanic (P≥ 0.1), and 1.1% vs 10.4% in black women (P≥ 0.1)
Manoudi[48] Prospective, cross-sectional	MINI; HAM-D	NR	n = 187 GDM 2.7%	Proportion of patients with major depressive episode who also had GDM was 2.6% (same as overall population, which was 2.7%)
Mautner[49] Prospective	EPDS	24th-37th week of gestation; 2-5 d postpartum; 3-4 mo postpartum	n = 40 GDM, n = 11 No GDM, n = 29	Mean (SD) EPDS scores in late pregnancy [7.55 (5.48)], immediately postpartum [7.00 (3.74)], and 3-4 mo postpartum [6.36 (5.63)] were not different in women with GDM compared with women without pregnancy complications [mean (SD) EPDS scores 6.41 (4.37), 4.69 (4.43), and 5.48 (4.88) in late pregnancy, immediately postpartum, and 3-4 mo postpartum, respectively] (P≥ 0.05)
Mei-Dan[50] Retrospective, health administration database	ICD-9, ICD-10CA, and/or DSM-IV (ICD codes NR)	Within 5 yr before pregnancy	n = 437941 With pre-pregnancy depression, n = 3724 No known mental illness, n = 432358	Prevalence of GDM during the index pregnancy was 3.4% in women with pre-pregnancy depression and 4.7% in women with no known mental illness (no statistical analysis) Prevalence of GDM and pre-pregnancy depression was 0.029%
Natasha[51] Prospective, case-control	MADRS ≥ 13	Approx. 25 wk gestation	n = 748 GDM, n = 382 No GDM, n = 366	Prevalence of depression was higher in women with GDM (25.92%) than in women without GDM (10.38%) (P value NR) There were significant associations between depression and current GDM (P < 0.001) and between depression and a history of GDM (P < 0.018) Mean (variance) MADRS scores were significantly higher in women with GDM [8.33 (7.23)] than women without GDM [4.42 (5.89)] (P value NR) Relative to women without GDM, women with GDM were more likely to have mild (MADRS score 13-19; adjusted OR: 3.07 or 4.06)3 or moderate (MADRS score 20-34; adjusted OR: 3.94) depression (P < 0.001)
Nicklas[52] Baseline description of RCT cohort	EPDS > 9	Mean (SD) 7.0 (1.7) wk postpartum (range, 4-15 wk)	n = 71	24 (34%) women with GDM had EPDS > 9 at postpartum visit [mean (SD) score 11.4 (2.2)]; cesarean delivery (P = 0.005) and greater gestational weight gain (P = 0.035), but not history of depression (P = 0.97), were associated with PPD
O'Brien[53] Retrospective, records review	EPDS ≥ 10	Mean (SD) 13.6 (8.2) wk gestation	n = 362 With depression, n = 256 Without depression, n = 106	No difference in prevalence of GDM between women with EPDS < 10 (14.6%) and those with EPDS ≥ 10 (15.0%) (P≥ 0.05)
Ragland[54] Prospective, cross-sectional	BDI > 13	During pregnancy	n = 50 GDM, n = 22	Mean BDI score among women with GDM was 13.7 9 (41%) women with GDM had BDI > 13
4Räisänen 2013[56] Retrospective, registry review	ICD10 codes F31.3, F31.5, F32-34	Up to 6 wk postpartum or a history of depression	n = 511422	Prevalence of GDM: 11.2% of women without any depression (n = 492103), 13.8% of women with history of depression but not PPD (n = 17881), 17.4% of women with PPD but no history of depression (n = 431), and 17.6% of women with both history of depression and PPD (n = 1007) (P ≤ 0.001) Among women with history of depression, increased prevalence of PPD was associated with GDM [OR (95%CI): = 1.62 (1.23-2.14)]
4Räisänen 2014[55] Retrospective, registry review	ICD10 codes F31.3, F31.5, F32-34	Up to hospital discharge after delivery	n = 511938	Prevalence of GDM: 11.2% of women without any depression (n = 493037), 13.4% of women with history of depression but not during pregnancy (n = 14781), 14.5% of women with depression during pregnancy but no history of depression (n = 2189), and 17.6% of women with both depression during pregnancy and history of depression (n = 1931) (P ≤ 0.001) An increased prevalence of depression during pregnancy was associated with GDM [adjusted OR (95%CI): 1.29 (1.11-1.50)]
Rumbold[57] Prospective	EPDS ≥ 12	Late pregnancy (for GDM)	n = 212 GDM (or glucose intolerance of pregnancy), n = 25 Negative OGCT, n = 95 Positive OGCT/negative OGTT, n = 29	No difference in proportion of women with EPDS score ≥ 12 in the GDM group (19%) compared with other groups (P≥ 0.05)
Silveira[58] Prospective, cohort	EPDS ≥ 13	Early (mean 12.4 wk gestation) and mid (mean 21.3 wk) pregnancy	n = 1115 GDM, n = 52 No glucose abnormality, n = 953	Prevalence of GDM did not differ between women with at least minor depression (EPDS ≥ 13) and women without depression (4.6% vs 5.6%) (P = 0.58) Prevalence of GDM did not differ between women with probably major depression (EPDS ≥ 15) and women without major depression (4.1% vs 5.6%) (P = 0.51)
Singh[59] Retrospective	BDI ≥ 10; self-reported medical history	During pregnancy	n = 152 History of depression, n = 39 No history of depression, n = 113	Of 39 women with history of depression, 15 (38%) had GDM Of 113 women with no history of depression, 67 (59%) had GDM (P value not reported)
Sit[60] Prospective	DSM-IV (SCID)	Past or current diagnosis	n = 186 Past MDD, n = 41 Current MDD, n = 39 Bipolar disorder, n = 45 No psychiatric disorder, n = 61	Mean (SD) glucose concentration after OGCT was 100 (25.0) mg/dL and did not differ among groups (P = 0.564) Rate of abnormal OGCT was 7% (13 of 186) and did not differ among the groups (P = 1.000) Only 3 women with abnormal OGCT were confirmed as having GDM (group not specified)
Song[61] (Chinese) Prospective	Self-rating Depression Scale ≥ 41	During pregnancy	n = 104 GDM, n = 50 No GDM, n = 54	Incidence of depression was 22% in women with GDM, significantly higher than in women without GDM (7.4%) (P < 0.05) Among women with GDM, mean (SD) insulin concentration 1 h after OGTT was significantly lower in women with depression [58.3 (32.4) mIU/mL, n = 11] than in those without depression [102.1 (65.2) mIU/mL, n = 39] (P < 0.05)
Sundaram[62] Prospective, exploratory	Survey of PPD diagnosis; survey of symptoms based on PHQ-2	Postpartum	Up to 61733 pregnancies	In analysis of data from 22 states, GDM was not a significant predictor of PPD symptoms [OR (95%CI): 1.13 (0.93-1.30), n = 45642, P = 0.14) or diagnosis [OR (95%CI): 0.96 (0.64-1.52), n = 5919, P = 0.89)
Walmer[63] Retrospective, electronic medical records	ICD-9 codes 296.2, 296.3, 309.0, 309.1, 311, 300.4	Postpartum	n = 18888 pregnancies (14988 women) GDM, n = 696 pregnancies (659 women)	After adjusting for age, pre-eclampsia, and preterm birth, GDM was significantly associated with increased risk of PPD [adjusted OR (95%CI): 1.46 (1.16-1.83), P = 0.001]; however, the association was not significant after adjusting for other clinical and demographic characteristics [adjusted OR (95%CI): 1.29 (0.98-1.70), P = 0.064] In subanalyses of ethnic/racial groups, GDM was significantly associated with PPD in black and white women, but not Hispanic women, after adjusting for age, pre-eclampsia, and preterm birth; the associations were not significant after full adjustment GDM was significantly predictive of mental health disorder (including depression, anxiety, and others) within 3 mo postpartum [adjusted OR (95%CI): 1.38 (1.04-1.85), P = 0.028]
Whiteman[64] Retrospective, maternal and infant database	ICD-9-CM codes 293.83, 296.2, 296.3, 300.4, 301.12, 309.0, 309.1, 311	Up to hospital discharge after delivery	n = 1057647	GDM was significantly associated with increased risk of depression [adjusted OR (95%CI): 1.44 (1.26-1.65)] (P value NR) Obesity was also associated with increased risk of depression, but there was no significant, additive interaction between GDM and obesity

¹The Bener et al and Burgut et al publications describe the same study, although different subgroups analyses are reported;

²Intervention comprised dietary advice, blood glucose monitoring, insulin therapy as needed, and usual care;

³Note that the adjusted OR for mild depression is variously reported as 3.065062 or 4.06 in the publication;

⁴The Räisänen et al 2013 and 2014 publications use the same database within the same time period (2002-2010) and, therefore, the study populations are almost identical. BDI: Beck Depression Inventory; CES-D: Center for Epidemiologic Studies Depression scale; CI: Confidence interval; DM: Diabetes mellitus; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; EPDS: Edinburgh Postnatal Depression Scale; GDM: Gestational diabetes mellitus; HAM-D: Hamilton Depression Scale; HCP: Healthcare professional; ICD: International Classification of Diseases; ICD-9-CM: International Classification of Diseases, 9^th revision, Clinical Modification; ICD-10CA: Enhanced version of International Classification of Diseases, 10^th revision, for use in Canada; MADRS: Montgomery-Åsberg Depression Rating Scale; MDD: Major depressive disorder; MHI 5: Mental Health Inventory-5; MINI: Mini International Neuropsychiatric Interview; NR: Not reported; NS: Not significant; OGCT: Oral glucose challenge test; OR: Odd ratio; PHQ: Patient Health Questionnaire; PPD: Postpartum depression; RCT: Randomized controlled trial; SCID: Structured Clinical Interview for DSM-IV; SD: Standard deviation; VA: Veterans affairs; WHO-5: World Health Organization Well-Being Index.

Table 2 Outcomes of included studies involving women with pre-existing type 1 or type 2 diabetes

First authorstudy design	Definition/measures of depression	Timing of depression measures	Overall nSubgroups, n	Main outcomes/findings
Berger[25] Retrospective	EPDS ≥ 13 or did not answer “No” to self-harm question	Within 4 d after delivery	Unselected, n = 322 History of mental illness, n = 215	Prevalence of pre-existing DM did not differ between women with or without postpartum depression in either the unselected group or the group with history of mental illness Of 5 women with pre-existing DM, none had depression
Callesen[30] Prospective, cohort	HADS ≥ 8	8 wk gestation	n = 148 Type 1, n = 118 Type 2, n = 30	Women with DM and depression were more likely to have preterm delivery (54% vs 16%, P = 0.003) and less likely to be nulliparous (23% vs 54%, P = 0.03) than women with DM without depression
Dalfra[34] Prospective	CES-D ≥ 16	3rd trimester and 8 wk after delivery	n = 245 Type 1, n = 30; No DM, n = 39	Mean (SD) CES-D scores at 3rd trimester were 19.1 (9.6) among women with Type 1 DM and 18.0 (8.7) among women without DM (P = 0.67) The severity of depressive symptoms increased from the 3rd trimester to after delivery in women with Type 1 DM [estimated mean difference in CES-D score (95%CI): 6.6 (2.9-10.2)], but decreased in women without DM [2.7 (-5.9-0.5), P < 0.0001 between groups]
Jovanovic[39] Retrospective, claims database	ICD-9 codes 311, 296.2, 296.3, 300.4, 301.12, 309.1	During pregnancy and/or within 3 mo after delivery	n = 839792 Type 1, n = 1125 Type 2, n = 10136 No DM, n = 773751	Prevalence of depression was 5.2% and 8.3% among women with type 1 and type 2 DM, respectively Prevalence of concurrent type 1 DM and depression was 0.006% Prevalence of concurrent type 2 DM and depression was 0.086% Relative risk (95%CI): of depression in women with type 1 DM vs women with no DM was 1.16 (0.86-1.56) Relative risk (95%CI): of depression in women with type 2 DM vs women with no DM was 1.84 (1.70-2.00)
Katon 2011[41] Cross-sectional analysis of prospective cohort	PHQ-9	3rd trimester	n = 2398 Pre-existing DM (type NR), n = 226; No DM, n = 1747	Prevalence (95%CI): of probable major depression among women with pre-existing DM was 5.8% (2.7%-8.8%) by PHQ-9 score, 8.9% (5.1%-12.6%) by antidepressant use, and 13.3% (8.8%-17.7%) by either PHQ-9 or antidepressant use, compared with the prevalence among women without DM [PHQ-9: 4.1% (3.2%-5.1%); antidepressants: 6.2% (5.1%-7.3%); PHQ-9 and antidepressants: 9.6% (8.2%-11.0%)] After adjusting for demographic characteristics, chronic medical conditions, and pregnancy variables, pre-existing DM was not associated with major or any antenatal depression (P value not reported)
Katon 2014 (PPD)[42] Retrospective, hospital database	PHQ-9	2nd or 3rd trimester and 6 wk after delivery	n = 1423	Prevalence of pre-existing DM was higher in women with PPD (14.5%) than in women without PPD (6.9%) (P = 0.02) Of 104 women with pre-existing DM, 12 (11.5%) had PPD Pre-existing DM was a risk factor for postpartum depression [OR (95%CI): 1.98 (1.12-3.52)] (P = 0.02) Prevalence of concurrent pre-existing DM and depression was 0.84%
Kozhimannil[46] Retrospective cohort	ICD9 codes 296.2, 296.3, 300.4, 301.12, 309.1, and 311	During the 6 mo before and up to 1 yr after delivery	n = 11024 With pre-existing DM (type NR), n = 311 (taking insulin, n = 57); no DM, n = 10367	Prevalence of depression in women with pre-existing DM taking insulin was 14.0% vs 16.1% among women with pre-existing DM not taking insulin (P value not reported)
Levy-Shiff[66] Prospective	BDI	2nd trimester	n = 153 Pre-existing DM, n = 53 (type NR) No DM, n = 49	No significant difference in depression during 2nd trimester between pre-existing DM [mean (SD) BDI score 6.17 (5.16)] and controls [6.59 (5.88)] (P≥ 0.05) For sample as a whole, higher levels of cognitive assessment of pregnancy as a challenge was associated with lower depression (P < 0.05) Among women with pre-existing DM, higher levels of medical support were associated with lower levels of depression (P < 0.01)
Mei-Dan[50] Retrospective, health administration database	ICD-9, ICD-10CA, and/or DSM-IV (ICD codes NR)	Within 5 yr before pregnancy	n = 437941 With pre-pregnancy depression, n = 3724 No known mental illness, n = 432358	Prevalence of DM (type NR) within 1 year before the index pregnancy was significantly higher in women with pre-pregnancy depression (3.4%) than in women with no known mental illness (1.2%) (P value NR) Prevalence of pre-existing DM and pre-pregnancy depression was 0.029%
Moore[67] Prospective	Depression Adjective Checklist; Perceived Stress Scale	3rd trimester	n = 131 Pre-existing insulin-dependent DM, n = 73 High risk of preterm birth, n = 48 Low risk of preterm birth, n = 25	White women with DM who were tested at a private clinic had higher Depression Adjective Checklist and Perceived Stress Scale scores than any other group (variables of white vs black, private vs public medical centre, DM vs low or high risk of preterm birth) (P value not reported)
Ragland[54] Prospective, cross-sectional	BDI > 13	During pregnancy	n = 50 Type 1 DM, n = 8 Type 2 DM, n = 20	Mean BDI score was 10.0 among women with type 1 DM and 17.1 among women with type 2 DM No women with type 1 DM and 12 (60%) women with type 2 DM had BDI > 13
1Räisänen 2013[56] Retrospective, registry review	ICD10 codes F31.3, F31.5, F32-34	Up to 6 wk postpartum or a history of depression	n = 511422	Prevalence of pre-existing DM: 8.4% of women without any depression (n = 492103), 11.1% of women with history of depression but not PPD (n = 17881), 14.6% of women with PPD but no history of depression (n = 431), and 13.3% of women with both history of depression and PPD (n = 1007) (P ≤ 0.001)
1Räisänen 2014[55] Retrospective, registry review	ICD10 codes F31.3, F31.5, F32-34	At hospital discharge after delivery	n = 511938	Prevalence of pre-existing DM (type NR): 8.4% of women without any depression (n = 493037), 10.9% of women with history of depression but not during pregnancy (n = 14781), 11.6% of women with depression during pregnancy but no history of depression (n = 2189), and 13.6% of women with both depression during pregnancy and history of depression (n = 1931) (P ≤ 0.001) Depression during pregnancy was not associated with pre-existing DM [adjusted OR (95%CI): = 1.10 (0.93-1.31)]
Singh[59] Retrospective	BDI ≥ 10; self-reported medical history	During pregnancy	n = 152 History of depression, n = 39 No history of depression, n = 113	Type 2 DM was significantly more common in women with history of depression than in women with no history of depression (P < 0.05) Of 39 women with history of depression, 5 (13%) had type 1 DM, and 19 (49%) had type 2 DM Of 113 women with no history of depression, 18 (16%) had type 1 DM, and 28 (25%) had type 2 DM
Sundaram[62] Prospective, exploratory	Survey of PPD diagnosis; survey of symptoms based on PHQ-2	Postpartum	Up to 61733 pregnancies	In analysis of data from 22 states, pre-existing DM was not a significant predictor of PPD symptoms [OR (95%CI): 1.16 (0.78-1.59), n = 45669, P = 0.39) or diagnosis [OR (95%CI): 1.31 (0.45-3.06), n = 5924, P = 0.56)] In analysis of data from 2 states that included both PPD symptoms and diagnosis on the survey, pre-existing DM was a significant predictor of PPD diagnosis [OR (95%CI): 5.65 (1.72-15.37), n = 2136, P < 0.01)]

¹The Räisänen et al 2013 and 2014 publications use the same database within the same time period (2002-2010) and, therefore, the study populations are almost identical. BDI: Beck Depression Inventory; CES-D: Center for Epidemiologic Studies Depression scale; CI: Confidence interval; DM: Diabetes mellitus; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; EPDS: Edinburgh Postnatal Depression Scale; HADS: Hospital Anxiety and Depression Scale; ICD: International Classification of Disease; ICD-10CA: enhanced version of International Classification of Diseases, 10^th revision, for use in Canada; NR: Not reported; PHQ: Patient Health Questionnaire; PPD: Postpartum depression; SD: Standard deviation.

Table 3 Outcomes of included studies involving women with any type of diabetes (not specified or data grouped)

First authorstudy design	Definition/measures of depression	Timing of depression measures	Overall nSubgroups, n	Main outcomes/findings
Ahmed[21] Prospective, cross-sectional	EPDS ≥ 10	6-8 wk postpartum	n = 1000 With DM (type NR), n = 31 No DM, n = 969	The proportion of women with DM who had PPD (51.6%) was significantly higher than the proportion of women without DM who had PPD (27.7%) (P = 0.004) Calculated prevalence of women with both DM and PPD was 1.6% (16 of 1000)
Bansil[22] Retrospective	ICD9 codes 296.2, 296.3, 300.4, 311, 298.0, 309.0, 309.1	At the time of delivery	n = 32156438 With depression, n = 244939; With DM (type 1, type 2, or GDM), n = 1536514 With DM and depression, n = 18245	Rate of concurrent DM at the time of delivery higher in women with depression (74.5 per 1000 deliveries) vs women without depression (47.6 per 1000 deliveries; OR (95%CI): 1.52 (1.47-1.58)] Calculated prevalence of DM and depression = 0.06% (18245 of 32156438 deliveries)
Benute[24] Prospective	PRIME-MD	During prenatal outpatient visits/hospital-isation	n = 326 With DM, n = 84 With MDD, n = 29	Prevalence of DM in women with MDD was 7.1% Calculated prevalence of DM and MDD = 0.61% (7.1% of 29 = 2; 2/326 = 0.61%)
Berger[25] Retrospective	EPDS ≥ 13 or did not answer “No” to self-harm question	Within 4 d after delivery	Unselected, n = 322 History of mental illness, n = 215	Prevalence of any DM did not differ between women with or without postpartum depression in either the unselected group or the group with history of mental illness
Chen[32] Retrospective	ICD9 codes 296.2, 296.3, 300.4, and 311	History of depression within 2 years before delivery	n = 5283 With DM (type NR), n = 319	Calculated prevalence of DM among women with depression was 6.0%
Kozhimannil[46] Retrospective cohort	ICD9 codes 296.2, 296.3, 300.4, 301.12, 309.1, and 311	During the 6 mo before and up to 1 year after delivery	n = 11024 With DM (pre-existing or GDM), n = 657;	Overall calculated prevalence of women with both DM (any type) and depression was 1.1% Prevalence of depression among women with any DM was 15.2% vs 8.5% among women without DM (P value not reported)
Ragland[54] Prospective, cross-sectional	BDI > 13	During pregnancy	No DM, n = 10367 n = 50 Type 1 DM, n = 8 Type 2 DM, n = 20 GDM, n = 22	Women with any DM had an increased odds of experiencing depression during or after pregnancy [OR (95%CI): 1.85 (1.45-2.36)] vs women without DM Women with any DM and no prenatal depression (9.6%) had increased odds of experiencing PPD or taking an antidepressant in the year after delivery [OR (95%CI): 1.69 (1.27-2.23)] vs women without DM Mean (SD) BDI score was 14.1 (9.9), range 3-43 Number (%) women with DM and severe (BDI ≥ 29), moderate (BDI 20-28), mild (BDI 14-19), and minimal (BDI 0-13) depression was 5 (10%), 8 (16%), 8 (16%), and 29 (58%) 42% of women with DM had BDI scores > 13, indicating clinical depression Among patients with clinical depression, only 19% were receiving treatment for depression Number of pregnancies showed a positive correlation with BDI score (P = 0.0078) Least mean squares of HbA1c level was higher, but not significantly, in women with depression [7.3% (56 mmol/mol)] than in those without [6.9% (52 mmol/mol)] (P≥ 0.05)
Räisänen 2013[56] Retrospective, registry review	ICD10 codes F31.3, F31.5, and F32-34	Up to 6 wk postpartum or a history of depression	n = 511422	Calculated prevalence of DM (any type) and depression in pregnant women = 0.06%
Singh[59] Retrospective	BDI ≥ 10; self-reported medical history	During pregnancy	n = 152 History of depression, n = 39 No history of depression, n = 113	Current BDI scores were higher in women with DM and history of depression [mean (SD) 17.2 (11.5)] than in women with DM and no history of depression [7.8 (7.4), P < 0.0001] Percentage of women with BDI ≥ 10 significantly greater in women with DM and history of depression (72%) than in women with DM and no history of depression (28%, P < 0.0001)
York[65] Prospective	Multiple Adjective Check List	36 wk gestation, and 2 d, 1 wk, 4 wk, and 8 wk postpartum	n = 36 Pre-existing DM, n = 6 GDM, n = 30	Most women did not report high levels of depression Among all women with DM, depression scores decreased significantly (P < 0.001) over time [mean (SD) scores of 9.2 (6.6), 10.1 (8.3), 6.7 (8.2), 5.6 (7.0), and 3.8 (4.2) at 36 wk gestation, 2 d postpartum, 1 wk postpartum, 4 wk postpartum, and 8 wk postpartum, respectively] There were no differences between women with GDM and women with pre-existing DM in depression scores during pregnancy (P = 0.17) or postpartum (P value not reported)

BDI: Beck Depression Inventory; CI: Confidence interval; DM: Diabetes mellitus; EPDS: Edinburgh Postnatal Depression Scale; GDM: Gestational diabetes mellitus; ICD: International Classification of Disease; MDD: Major depressive disorder; NR: Not reported; PPD: Postpartum depression; PRIME-MD: Primary Care Evaluation of Mental Disorders classification system; SD: Standard deviation.