Prevalence and risk factors for postpartum depression among women with preterm and low-birth-weight infants: a systematic review


Dr SN Vigod, Reproductive Life Stages Program, Women’s College Hospital, 76 Grenville St., 9th Floor, Toronto, ON, Canada M5S 1B2. Email


Please cite this paper as: Vigod S, Villegas L, Dennis C-L, Ross L. Prevalence and risk factors for postpartum depression among women with preterm and low-birth-weight infants: a systematic review. BJOG 2010;117:540–550.

Background  Although much is known about the risk factors for postpartum depression (PPD), the role of giving birth to a preterm or low-birth-weight infant has not been reviewed systematically.

Objective  To review systematically the prevalence and risk factors for PPD among women with preterm infants.

Search strategy  Medline, CINAHL, EMBASE, PsycINFO and the Cochrane Library were searched from their start dates to August 2008 using keywords relevant to depression and prematurity.

Selection criteria  Peer-reviewed articles were eligible for inclusion if a standardised assessment of depression was administered between delivery and 52 weeks postpartum to mothers of preterm infants.

Data collection and analysis  Data on either the prevalence of PPD or mean depression score in the target population and available comparison groups were extracted from the 26 articles included in the review. Risk factors for PPD were also extracted where reported.

Main results  The rates of PPD were as high as 40% in the early postpartum period among women with premature infants. Sustained depression was associated with earlier gestational age, lower birth weight, ongoing infant illness/disability and perceived lack of social support. The main limitation was that most studies failed to consider depression in pregnancy as a confounding variable.

Author’s conclusions  Mothers of preterm infants are at higher risk of depression than mothers of term infants in the immediate postpartum period, with continued risk throughout the first postpartum year for mothers of very-low-birth-weight infants. Targeted clinical interventions to identify and prevent PPD in this vulnerable obstetric population are warranted.


Postpartum depression (PPD) affects approximately 10–15% of women and is one of the most common complications of child-bearing.1 The consequences of PPD for both mother and infant are well established: women who suffer from PPD are twice as likely to experience future episodes of depression over a 5-year period.1 PPD can also impair maternal–infant interactions, leading to attachment insecurity, developmental delay and social interaction difficulties in affected children.2–6

Risk factors for PPD have been examined in three meta-analyses, indicating that the strongest predictors of PPD include depression or anxiety during pregnancy, personal and family history of depression, lack of social support and stressful life events.1,7,8 However, as most women who become depressed do not necessarily have personal and family histories of depression, it is important to focus on variables that occur more commonly, such as stressful events. Few infant-related variables have been considered in the above-mentioned meta-analyses, yet childcare stress has been identified as having a moderate impact on PPD in one meta-analysis, based on data from seven studies.7 Despite the widely held belief that preterm birth is a stressful experience,9–11 neither preterm birth nor low birth weight (LBW) have been examined in any of the published meta-analyses.

There is significant variation in reports of the prevalence of prematurity internationally, partly as a result of varying definitions across countries. However, both preterm birth [<37 weeks’ gestational age (GA)] and LBW (<2500 g at birth) are common. For example, the prevalence of preterm birth approaches 13% in the USA,12 with lower rates reported in Europe and other developed countries (5–9%).13 Yet, few population-based studies of PPD have included an adequate number of cases to provide a precise estimate of PPD risk among mothers with premature and/or LBW infants. Still fewer studies have examined factors (maternal, infant or obstetric) that may modify the relationship between preterm birth and PPD. These are important questions for obstetric, neonatal and general practitioners who may be the first point of contact for the identification of PPD in this context. A systematic review of population-specific studies attending to the complexity of the relationship between preterm birth and PPD is therefore warranted.

In the light of these research gaps, this systematic review has two major objectives: (i) to estimate the prevalence of PPD among women who have preterm and/or LBW infants (with comparison, when possible, with mothers of term infants); and (ii) to determine what factors are associated with PPD in this population. The results of this systematic review will provide clinicians with a summary of the current knowledge on whether giving birth to a preterm or LBW infant is an independent risk factor for PPD.


Search strategy

There was no pre-existing review protocol for this review. Therefore, articles included in this review were identified through a broad electronic search of the following databases: Medline, CINAHL, EMBASE, PsycINFO and the Cochrane Library. Each database was searched from its start date to August 2008 using the following keywords designed to capture any studies related to the stated objectives: postpartum, postnatal, pregnan*, perinatal, childbirth, obstetr*, labor, puerperal, parturition, parity, or maternal LBW, premature birth, premature labor, high-risk pregnancy, GA, preterm infant, preterm birth, neonatal intensive care unit, NICU, special care nursery, or ill infant in combination with depress*, mental health, mental illness, mood disorder, affective disorder, emotion, or baby blues. The search was conducted by a trained research assistant under the supervision of one of the co-investigators who has substantial experience in conducting systematic reviews and meta-analyses in the field of reproductive health (Cindy-Lee Dennis). Secondary references and review articles were scanned to identify primary articles that may have been missed in the initial search. Only published, peer-reviewed articles available in English or French were considered for this review, as resources were not available to assess the quality of studies published in other languages.

Peer-reviewed publications were identified in the initial stage of the search process, and potentially relevant abstracts meeting the predetermined eligibility criteria were subsequently obtained for further examination (n = 2243 records). The reference lists of the included studies and relevant reviews were reviewed in an attempt to identify any additional articles (n = 0). Research articles were then retrieved and assessed more rigorously to determine inclusion suitability (n = 98). In total, 26 articles met the inclusion and exclusion criteria and were included in the review.

Criteria for study selection

Studies that reported either the prevalence of PPD or a mean score on a measure of depressive symptoms in mothers of preterm infants were considered for eligibility in this review.

Assessment of exposure  For the purpose of this review, we included studies that defined preterm birth as <38 weeks’ GA or that defined an LBW infant as a birth weight of <2500 g. Studies without such definitions were included when the maximum GA at birth was reported to be <38 weeks and/or the maximum birth weight was reported to be <2500 g in the target group.

Assessment of outcome  Only research studies that included a standardised measure of depression (either self-report or observer-rated), performed during the first postpartum year, were included. Studies were required to report either the prevalence of PPD (as variously defined by the authors) in the population of interest or mean depression scores using scales with well-validated cut-off scores that indicate high risk for a major depressive episode [e.g. Edinburgh Postpartum Depression Scale (EPDS); Centre for Epidemiological Studies-Depression Scale (CES-D); Beck Depression Inventory (BDI)]. The cut-off scores used in the studies varied depending on the primary objective of the studies. For example, a cut-off score of EPDS > 9 has a greater sensitivity, but poorer specificity, than a cut-off score of EPDS > 12. The study authors therefore chose their cut-off score according to whether their primary objective would be best achieved by maximising the sensitivity or specificity. When the mean scores are not presented, this limits our capacity to compare (and combine) outcomes of different studies. However, using these standard measures of depression, results of individual studies can still be meaningfully interpreted in groups of women with preterm or LBW infants by comparison with validated norms or cut-off scores. Therefore, studies with and without comparison groups of term infants were included where an appropriate standardised depression instrument was administered. Only studies in which subjects were assessed after ascertainment that the infants were sufficiently well to survive to discharge home from hospital were included. To avoid increased complexity related to the effect of multiple births, studies that limited their samples to mothers with multiple births were excluded.

Assessment of validity

The methodological quality of each study was assessed by first examining the primary research question of the study, the stated target population and the ultimate study sample for external validity. To assess the internal validity of the studies, each study was examined individually for risk of bias. Specifically, risk of selection bias in each of the individual studies was assessed by reviewing participant demographic characteristics, and noting any known or potentially unknown differences between the groups. These characteristics were considered to be important to evaluate for this review, as certain variables are known to be associated with both preterm/LBW and PPD, introducing the potential to confound any interpretation of the effect of preterm/LBW infants on PPD. These variables include sociodemographic, medical and psychiatric characteristics, such as low income, poor social support, other stressful life events (e.g. deaths, domestic violence, loss of employment), high-risk pregnancy, depression in pregnancy and/or anti-depressant use in pregnancy.14,15 The included articles were evaluated systematically for whether they reported on these potential confounding factors. This information is cited in the text when relevant throughout the review.

Data extraction

The first author extracted the data from all studies and the co-authors each independently extracted data from a subset of the articles into data summary tables. Data on the population(s) studied, study design, assessment of exposure, measure of depression and timing of measurement(s), prevalence (or mean score) of depression and potential confounding factors (as outlined above) were extracted from each included study. In studies in which there was an intervention designed to treat or prevent depression, only the results from the control group were considered when estimating the risk for PPD.

The included studies were also reviewed systematically for attention to the major risk factors for PPD: personal history of depression, depression in pregnancy, adequacy of social support and financial support, and any other stressful life events. In addition, studies were examined for possible mediating variables in the setting of having a preterm/LBW infant. These included both the level of prematurity and level of illness of the neonate. When authors reported on additional risk factors, these were also extracted into the data summary tables.

Quantitative data synthesis

Our intention was to perform a quantitative synthesis of the available data. Specifically, fixed-effect meta-analyses were to be used to combine data in the absence of significant heterogeneity. If there was heterogeneity, where results were being pooled from studies examining different exposures, or where it was unclear whether the same outcome was being measured in all studies, random-effects meta-analyses were to be used.


The article selection process is presented in Figure 1. There were 26 articles included in this review for a total of 2392 mothers of preterm infants. A description of these studies, including an assessment of validity, is presented in Table S1. As a result of the substantial diversity of the data with respect to sample characteristics, methodology, definition of exposure and outcome reporting (assessment measures, cut-off scores and timing), we were unable to conduct the planned quantitative data synthesis. Therefore, we performed a qualitative synthesis, presenting the results of each study individually with narrative comment in the text regarding the risk of bias for individual studies.

Figure 1.

 Study selection flow diagram.

During the data extraction process, it became clear that a subset of the included studies was focused solely on very preterm or very LBW infants. Clinically, this is an important distinction, as the medical prognosis for this latter group of infants differs from that for preterm infants as a whole, and there is likely to be increased care-giving burden and stress for mothers. Therefore, the results of studies that included mothers of preterm infants of any GA or birth weight (i.e. <38 weeks’ GA or birth weight of <2500 g) are presented in Table 1. Studies that included primarily mothers of very preterm or very LBW infants (as defined by GA at birth of <33 weeks and/or birth weight of <1500 g) are presented as a subgroup analysis (Table 2).

Table 1.   Postpartum depression in mothers of preterm and/or low birth weight infants (broadly defined)
StudyAssessment measureTiming (weeks postpartum)Target group (mean, SD) or %Comparison groups*Statistical testing (as reported by authors)
  1. BW, birthweight; CES-D, Centre for Epidemiological Studies-Depression Scale; EPDS, Edinburgh Postpartum Depression Scale; GA, gestational age at birth; NS, not statistically significant.

  2. *See text for descriptions of comparison groups. **At hospital discharge, no exact timing reported. ***Adjusted for prematurity.

Adewuya et al. (2005)18EPDS > 9633%10.8%Unadjusted OR 4.21 (95% CI 2.78–6.39)
Aradine et al. (1990)25CES-D4
14.6 (11.7)
11.0 (8.8)
(i) 14.9 (8.70),
(ii) 14.2 (10.4)
(i) 13.5 (10.9),
(ii) 10.2 (7.60)
No significant differences in CES-D scores between any groups.
Drewett et al. (2004)16EPDS > 128
= 19.34, < 0.001
= 3.43, = 0.064
Logsdon et al. (1997)21CES-D**
** + 4
24.88 (12.37)
19.36 (12.82)
= 2.68, = 0.006 over time
Logsdon et al. (2001)17CES-D6–820.7 (13.6)(i) 14.5 (10.2)
(ii) 20.5 (10.9)
(i) = 0.020
(ii) > 0.05
Nielson Forman et al. (2000)24EPDS > 12168.5%5.4%No significant difference
O’Brien et al. (1999)23CES-DIn hospital
6 weeks post**
21.1 (11.1)
12.9 (10.1)
16.1 (11.0)
Over time: F(2,43) = 13.10, < 0.001
Tamaki et al. (1997)19EPDS > 94
No test statistics reported, > 0.1
No test statistics reported, < 0.1
No test statistics reported, > 0.1
Veddovi et al. (2001)26EPDS77.13 (4.87)
Veddovi et al. (2004)20EPDS > 12∼8
Table 2.   Postpartum depression in mothers of very preterm and/or very low birth weight infants
StudyAssessment measureTiming (weeks postpartum)Target group(s) mean (SD/95% CI) or %Comparison group(s)*Statistical testing (as reported by authors)
  1. BDI, beck depression inventory; BSI, brief symptom inventory; BPD, bronchopulmonary dysplasia; BW, birthweight; CES-D, Centre for Epidemiological Studies-Depression Scale; EPDS, Edinburgh Postpartum Depression Scale; GA, gestational age; HADS, Hospital Anxiety and Depression Scale (T, total; D, Depression subscale); LBW, low birthweight; MADRS, Montgomery Aspers Depression Rating Scale; NICU, Neonatal Intensive Care Unit; PPD, postpartum depression; RDS, respiratory distress syndrome; SADS, schedule for affective disorders and schizophrenia; SCID, Structured Clinical Interview for DSM-IV; VLBW, very low birthweight.

  2. *See text for descriptions of comparison groups. **timing not reported. ***adjusted for prematurity.

Allen et al. (2004)32BDI∼59.7 (7.8)
Carter et al. (2005)28HADS-T
312.6 (6.4)
5.4 (3.3)
8.5 (5.1)
33–35 weeks36–42 weeks (NICU)No NICU= 0.031
= 0.017
= 0.285
11.7 (6.6)
4.9 (3.5)
7.7 (5.3)
9.4 (6.8)
3.6 (3.6)
6.8 (5.4)
9.7 (6.4)
4.2 (3.3)
6.8 (5.1)
Davis et al. (2003)33EPDS > 12440.3%
Feldman et al.(2002)34BDIInfant 37 weeks GA9.05 (4.27)
Finello et al. (1998)39CES-D10
17.9 (12.9)
18.6 (9.2)
Over time: No change
Hagan et al. (2004)35SADS8, 24 and 5235% by 52 weeks
Kersting et al. (2004)29BDI2
10.6 (7.9)
6.3 (5.9)
8.5 (8.7)
4.7 (4.3)
4.1 (4.0)
4.0 (3.9)
< 0.001
= NS
= 0.037
= 0.129 over time
12.6 (8.2)
8.6 (7.5)
11.4 (7.0)
3.6 (3.7)
2.8 (2.6)
0.9 (1.2)
< 0.001
< 0.001
< 0.001
= 0.480 over time
No statistics reported.
Lambrenos et al. (1996)40DSM-III-R interview52***CP: 28%
No CP: 26%
26%F(2) = 0.09, = 0.916
Meyer et al. (1994)36BDI > 9Baseline**
Preyde et al. (2003)43BDI1
4.53 (3.81)
4.88 (3.51–6.27)
Pridham et al. (2001)41CES-D
9.90 (7.13)
9.00 (8.82)
9.14 (7.40)
8.65 (7.63)
All between and within group comparisons ‘non-significant’
8.82 (7.48)
9.41 (12.4)
9.00 (8.62)
10.68 (14.1)
10.43 (7.92)
10.09 (7.91)
10.77 (8.39)
7.78 (7.37)
Pridham et al. (2005)37CES-D4***
16.50 (10.26)
13.12 (8.57)
13.14 (7.63)
11.50 (5.20)
Over time: Trend NS
Saigal et al. (2003)38CES-D1
19.7 (9.8)
10.2 (8.6)
Over time: < 0.001
Segal et al. (1995)42CES-D > 16
13.89 (7.60)
10.69 (6.28)
= 1.34, NS
Singer et al. (1996)30BSI4***BPD: 0.7 (0.9)
No BPD: 0.7 (0.9)
0.3 (0.4)‘Significant’ difference
Singer et al. (1999)31BSI > 84th Percentile4***BPD: 32%
No BPD: 29%
17%Not reported

For dichotomous data, results are presented as prevalence (%) with statistical comparisons when available. For continuous data, means are reported with statistical comparisons when available.

Mothers of preterm or LBW infants

The results of the ten studies that included mothers of infants born up to GA < 38 weeks at birth and/or birth weight <2500 g are summarised in Table 1. The studies are arranged alphabetically for ease of reference.

Studies supporting an increased risk of PPD  Four of the six studies with comparison groups reported either significantly increased depression prevalence or mean depression scores in mothers of preterm infants at some point in the first postpartum year.16–19 Drewett et al.,16 using EPDS (cut-off of >12), found that the risk of depression at 8 weeks postpartum in mothers of preterm infants was 1.6 times the risk in mothers of term infants after adjusting for potential confounding factors [adjusted relative risk, 1.6 (95% CI, 1.2–2.1)]. This was the only study to account for depression during pregnancy in multivariable modelling. However, the results of the three other studies were similar despite this limitation. Using EPDS (cut-off of ≥9), Adewuya et al.18 found that GA at delivery remained a significant predictor of PPD at 6 weeks postpartum after accounting for admission to hospital during pregnancy (which is a possible reflection of severe psychiatric as well as medical illness during pregnancy). Similarly, Logsdon et al.17 found significantly increased CES-D scores at 6–8 weeks postpartum in Caucasian mothers of preterm infants compared with demographically similar mothers of term infants. Finally, Tamaki et al.19 found increased prevalence of PPD (EPDS ≥ 9) in mothers of preterm infants at 12 weeks postpartum. Taken together, these four studies suggest an increased risk of depressive symptoms in mothers of LBW infants, at least up to 12 weeks postpartum.

Three of four studies without comparison groups similarly revealed high levels of depressive symptomatology in mothers of preterm infants.20,21 Specifically, another study by Logsdon et al.21 reported a CES-D mean score of 24.88 (SD, 12.87) at discharge from hospital in a convenience sample of Caucasian mothers (with a mean education level of 13 years) of preterm infants recruited from a neonatal intensive care unit (NICU). These results indicate that approximately 75% of the sample scored above the CES-D score of ≥16, which is considered as high risk for depression.22 O’Brien et al.23 similarly reported high CES-D scores both in hospital and 6 weeks post-hospital discharge (mean, 21.1; SD, 11.1; mean, 16.1; SD, 11.0) in an American NICU sample in which at least 50% of the infants were born at >1500 g. Finally, Veddovi et al.20 reported a prevalence of PPD (EPDS > 12) of 18.9% in a sample of women with preterm infants at 52 weeks postpartum (corrected for GA). Despite the benefit of comparison groups of mothers of term infants, these studies are consistent with those described above in supporting an increased risk of PPD in this population.

Studies not supporting an increased risk of PPD  In total, three of the ten studies did not support an increased risk for PPD at any point in the postpartum period for mothers of preterm/LBW infants. The most methodologically rigorous study was a birth cohort study,24 in which depression was assessed at 16 weeks postpartum using EPDS. The authors found no difference in the prevalence of PPD (EPDS cut-off of >12) between mothers of preterm and term infants, nor did ‘pregnancy or delivery complications’ predict PPD after accounting for prepregnancy and antenatal psychiatric disturbance in multivariable models. Similarly, a multiple cohort study25 compared three groups of women: women with high-risk pregnancies who delivered preterm infants; women with high-risk pregnancies who delivered term infants; and women with low-risk pregnancies who delivered term infants. The comparison group of mothers with high-risk pregnancies (but term infants) helps to account for the possible confounder of illness during pregnancy. In this study, no differences in mean depression scores (using CES-D) were found at either 4 or 16 weeks postpartum. However, it is notable that the birth weights of the infants in the preterm group significantly overlapped with the birth weights of the infants in the target groups of term infants, making the classification of exposure status (i.e. prematurity) questionable. Finally, another study by Veddovi et al.26 did not include a comparison group of mothers with term infants; however, they found a relatively low mean EDPS score (7.13; SD, 4.87) in an NICU sample at 7 weeks postpartum (a score of >12 on EPDS is generally regarded as high risk for depression27). The mothers in this Australian study were primarily Caucasian, married, highly educated and over three-quarters were employed. It is possible that these characteristics may have mitigated depressive symptoms in this population. In addition, mothers of infants who suffered any ‘major’ neonatal complications (as defined by the authors) were excluded from the study.

Risk of PPD over time  Of the six studies that examined PPD over time, five revealed decreasing prevalence or mean scores over time.16,19,21,23,25 Using the study with the strongest methodology as an example, in a birth cohort of 12 391 mothers in the UK, where 673 (5.4%) had given birth to preterm infants (<38 weeks’ gestation at birth), Drewett et al.16 found that, at 8 weeks postpartum, the prevalence of depression (EPDS > 12) was significantly higher in the mothers of preterm infants than in the mothers of term infants, even after controlling for depression in pregnancy [15.4% versus 9.4%; P < 0.001; relative risk, 1.6 (95% CI, 1.2–2.1)]. By 32 weeks postpartum, however, this difference was no longer significant (10.7% versus 8.4%; P = 0.64).

Veddovi et al.20 was the only study in which the prevalence of PPD (EPDS > 12) was higher at 52 weeks postpartum (18.9%) than at discharge from hospital at approximately 8 weeks postpartum (12.5%). However, it should be noted that, in this study, mothers of infants with severe neonatal complications, such as disorders of the central nervous system, intraventricular haemorrhage and congenital anomalies, were excluded. It is possible that this contributes to the fact that the rate of PPD on discharge from hospital is consistent with general population levels.

Risk factors for PPD  Three studies reported the risk factors for PPD in mothers of preterm infants. Factors associated with either PPD or poor maternal adjustment were as follows: first-born infants, low self-esteem, reduced levels of social support and infant sleep difficulties (for details, see Appendix S1).

Mothers of very preterm or very LBW infants

Sixteen studies examined the risk for PPD in samples composed primarily of mothers of very preterm or very LBW infants (Table 2). The studies are arranged alphabetically in Table 2 for ease of reference.

Studies supporting an increased risk for PPD  Four of seven studies that compared mothers of very preterm or very LBW infants with mothers of term infants found a significant difference between the groups during at least one assessment in the first year postpartum.28–31 Of these studies, only Carter et al.28 assessed maternal psychiatric history, identifying no differences in the rates of past depression between mothers of preterm and term infants. This study reported a dose–response relationship between the level of prematurity and depressive symptoms, with higher depression scores in mothers of infants born at <33 weeks’ gestation than in mothers of infants born at 33–35 weeks’ gestation and mothers of term infants (including term infants who required NICU admission). Although the other three studies did not assess for maternal psychiatric history, the risk for selection bias in these studies was low, as mothers in both the preterm and term groups were similar in age, education level, marital status and parity in each case. Using a structured clinical interview for depression, Kersting et al.29 found a slightly increased prevalence of PPD at 2 and 24 weeks’ postpartum in mothers of very preterm infants compared with matched controls. They also found significantly higher scores in mothers of very preterm infants at 2 and 60 weeks postpartum on BDI and at 2, 24 and 60 weeks postpartum (all measured time points) on the Montgomery–Asberg Depression Rating Scale (MADRS). In two separate studies by Singer et al.,30,31 scores on the brief symptom inventory were higher for mothers of preterm infants (with and without bronchopulmonary dysplasia) than for comparison groups of mothers of term infants.

Eight of nine studies without comparison groups reported either high mean depression scores or prevalence rates, relative to that expected in the general population of postpartum women.23,32–39 The highest prevalence rate was reported by Davis et al.33 in an Australian sample of highly educated mothers (55% primiparous), 16% of whom reported a previous history of depression. In this study, 40.3% of women had an EPDS score of >12 at 4 weeks postpartum. The remaining studies were primarily composed of samples of well-educated Caucasian women who had given birth to preterm singleton infants, evenly split in terms of primiparity versus multiparity, with good levels of social support.23,33,34,36,37 It should be noted, however, that not all studies included these types of sample, and participants in some studies may have had risk factors for elevated depression scores in addition to very preterm births. Specifically, two studies included mothers with multiple births in their sample (in Saigal et al.,38 21% were multiple births; in Hagan et al.,35 16% were multiple births). Further, the mothers in the study by Finello et al.39 were, on average, relatively new immigrants to the USA (3.8 ± 1.1 years in the USA), and 31% of the mothers in the study by Allen et al.32 were not Caucasian.

Studies not supporting an increased risk of PPD  Three of seven studies with comparison groups did not find a statistically significant increased risk for depression in mothers of very preterm or very LBW infants when compared with mothers of preterm infants.40–42 However, in all of these studies, the prevalence of depression or mean scores on depression rating scales were relatively high in the comparison groups. For example, the two studies reporting the prevalence of depression found rates of 25% (at 28 weeks postpartum, corrected for GA) and 26% (at 52 weeks postpartum, corrected for GA) in the respective control groups40,42 (see Table 2). Only one of nine studies without a comparison group of mothers of term infants did not report high mean depression scores.43 This randomised controlled trial evaluated an NICU intervention to reduce distress in mothers of preterm infants, in which mothers were assessed for depression (before and after the intervention) using BDI whilst their infants were still in hospital. Although the results reported here are for the ‘control’ group, it is possible that there was a selection bias in terms of desire to enroll (e.g. mothers who were more well were more likely to enroll) and/or that the effect of being enrolled in a trial may have been therapeutic in and of itself. Interestingly, all measurements of depression were made whilst the infants remained hospitalised.

Risk of PPD over time  Six of the seven studies in which depression was measured at more than one time point reported sustained depression scores and PPD prevalence rates throughout the first postpartum year.29,32,33,36,37,39 For example, Finello et al.39 measured depression in an NICU cohort within 4 weeks of hospital discharge (approximately 10 weeks postpartum) and then again at 52 weeks postpartum. There were high mean CES-D scores at both time points [17.9 (SD, 12.9) and 18.6 (SD, 9.2), respectively; P > 0.05]. Only one study reported statistically significant decreases in depression scores. However, in this study,38 the first measurement was taken only 1 week after birth (when high distress would be expected).23

Risk factors for PPD  Only five of these studies reported risk factors for PPD in the subgroup of mothers with very preterm or very LBW infants. When the level of illness of the preterm neonate was examined as a potential risk factor for PPD,30,31,34,38 the results were conflicting and did not clearly support the level of illness as a major risk factor for PPD in mothers of very preterm infants. The other risk factors reported for PPD in mothers of very preterm or very LBW infants were similar to those found overall in the PPD literature1,7,8 (for details, see Appendix S1).


Although the limitations of the included literature necessitated a qualitative rather than quantitative analysis of the results, several patterns have emerged that may aid clinicians to identify women at risk for PPD after giving birth to preterm and/or LBW infants. Overall, the included studies indicated that mothers of preterm LBW infants have higher depression scores than mothers of term infants, at least up to 12 weeks postpartum. Further, studies have even more consistently demonstrated that mothers of very preterm or very LBW infants have higher levels of depressive symptoms throughout the first postpartum year, with a limited reduction in symptoms after hospital discharge.

One potential hypothesis to explain the ‘dose–response’ relationship between prematurity and PPD is that there is enhanced stress with increasing prematurity. The distinction between preterm/LBW and very preterm/very LBW infants may be important because of the increased amount of medical comorbidity, length of hospitalisation and long-term disability in very preterm/very LBW infants. However, the studies in our review did not find medical risk to be a specific risk factor for PPD in the subpopulation of mothers of very preterm or very LBW infants. It is possible that there is a certain threshold for medical comorbidity, above which additional problems do not result in additional distress. More research is needed in this area to guide our understanding of how the degree of neonatal illness might mediate the risk for PPD in mothers of preterm infants. A cohort study with clear definitions for the level of prematurity and valid measures of neonatal and infant morbidity would serve this purpose.

Another implication of this difference between mothers of preterm/LBW and very preterm/very LBW infants is that it may be necessary to determine whether high scores on depression scales within the first 12 weeks postpartum are reflective of an adjustment disorder with depressed mood, as defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR), rather than a diagnosis of major depression.44 By definition, although symptoms of an adjustment disorder are greater than would normally be expected from a stressor, they do not continue for longer than 6 months after the termination of a stressor and do not interfere markedly with an individual’s level of functioning. It is possible that mothers of preterm (but not very preterm) infants show increased distress early on, but do not have ongoing symptoms that develop into a major depressive disorder. A future study in which depression is assessed during pregnancy and postpartum by clinical interview would help to clarify this point. In the meantime, as the treatments for major depressive disorder and adjustment disorder can differ, it is important to consider both in the differential diagnosis, with referral to psychiatric specialists for clarification as needed.


The results of this review highlight the variability within and methodological limitations of research examining the important question of whether prematurity and LBW are risk factors for PPD. Many included studies had relatively small samples (n < 50 in the target group), limiting the power to detect significant differences between groups. Further, some of the included studies had no comparison groups of mothers of term infants. In addition, multiple assessment tools were administered: a broad definition of PPD was used in this review to ensure that all potentially relevant studies were examined; however, only three of the included studies employed a diagnostic assessment of depression.29,35,40 with the remainder relying on self-report assessments of depressive symptoms (with varying cut-off scores to delineate PPD). Many of these measures identify women at high risk for depression and are not diagnostic for major depression specifically.

The most concerning limitation is that only two studies accounted for depression in pregnancy or a personal history of depression as a risk factor for PPD, despite strong evidence that these are two of the strongest risk factors for PPD in the general population.14 Further, there is some evidence that depression and, possibly, antidepressant use are associated with risk for prematurity.15 Therefore, without controlling for antenatal depression, it cannot be ruled out that increased rates of PPD among women with preterm birth and/or LBW infants are simply a manifestation of the high rates of prenatal depression among this group. The two studies that controlled for antenatal depression showed conflicting results.16,24 The fact that only two studies included depression history as a potential covariate could be because many of the studies included in this review are more than 5 years old, and the importance of depression in pregnancy as a risk factor for prematurity has only been reported more recently.14,15 However, other key variables that may confound the relationship between prematurity and PPD (e.g. socioeconomic status and marital stress, among others) were similarly only accounted for in a minority of the studies of this review. Future studies should include these key variables in their analyses.


Although there is a need for more rigorous research to confirm the findings of this review, there is converging evidence that PPD is a problem in this population. This critical assessment of the literature to date has revealed that having a premature or LBW infant may be conceptualised as a stressful experience leading to depressive symptoms. These symptoms among mothers of preterm and/or LBW infants may resolve with time. However, having a very preterm or very LBW infant is associated with a sustained increased risk for PPD throughout the first postpartum year.

The major clinical implication of this review for healthcare providers treating women with preterm (and particularly very preterm) infants is that it highlights the need to identify symptoms of depression in this vulnerable obstetric population. As discussed, PPD can result in poor bonding and attachment between mother and infant, with long-term consequences. As premature infants spend so much time in intensive care units, there is already a risk that mothers will have difficulty with bonding and attachment to their infants. In fact, most of the intervention studies in the preterm infant literature have been designed to decrease distress in mothers and increase bonding with their infants. Therefore, although such interventions may actually be most important for mothers who are depressed, women with PPD may not be able to benefit optimally from such interventions until their depression is identified and treated. Therefore, it is important that women who have depressive symptoms are further assessed by mental health professionals for diagnostic clarification and treatment, with the goal of optimal psychological outcomes for mothers and infants.

Given the important limitations to the available evidence, a study with prospective measurement of depression in pregnancy (and information about treatment of antenatal depression) that aims to ascertain whether preterm delivery is an independent risk factor for PPD would be helpful to confirm our conclusions. Secondary aims should be focused on a further understanding of how the level of prematurity and other risk factors might modify the risk for PPD in mothers of preterm infants.

Disclosure of interests

There are no financial disclosures or conflicts of interest for any authors.

Contribution to authorship

SNV extracted the data from all the included studies, synthesised the results and drafted the manuscript. LV executed the search strategy, reviewed the articles for inclusion criteria and extracted the data from some included studies. C.-L.D planned the study and search strategy, and edited the manuscript. LER planned the study and search strategy, reviewed the articles for inclusion criteria, extracted the data from some included studies and edited the manuscript.

Details of ethics approval

Ethics approval was not obtained as this was a secondary analysis of published studies.


SNV is supported by a Research Training Fellowship from the Ontario Mental Health Foundation and by the Department of Psychiatry, Women’s College Hospital. LER is supported as a New Investigator by the Canadian Institutes of Health Research and Ontario Women’s Health Council, Award NOW-84656. In addition, support to the Centre for Addiction and Mental Health (CAMH) for the salary of scientists and infrastructure has been provided by the Ontario Ministry of Health and Long Term Care. The views expressed here do not necessarily reflect those of the Ministry of Health and Long Term Care.


We wish to thank Ms Corrie Goldfinger and Ms Vashti Campbell for their assistance in the preparation of this article.