Home > Results - Efficacy and Safety of...

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Myers ER, Aubuchon-Endsley N, Bastian LA, et al. Efficacy and Safety of Screening for Postpartum Depression [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Apr. (Comparative Effectiveness Reviews, No. 106.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Results

Introduction

We begin by describing the results of our literature searches. We then provide a brief description of the included studies. The remainder of the chapter is organized by Key Question (KQ). Under each of the six KQs, we begin by listing the key points of the findings, followed by a brief description of included studies and a detailed synthesis of the evidence. We conducted quantitative syntheses where possible, as described in the Methods chapter.

A list of abbreviations and acronyms used in this chapter is provided at the end of the report.

Results of Literature Searches

Figure 2 depicts the flow of articles through the literature search and screening process. Searches of PubMed®, Embase®, PsycINFO®, and CDSR yielded 5,059 citations, 1,528 of which were duplicate citations. Manual searching identified 154 additional citations, for a total of 3,685 citations to be screened. After applying inclusion/exclusion criteria at the title-and-abstract level, 1,293 full-text articles were retrieved and screened. Of these, 1,248 were excluded at the full-text screening stage, leaving 45 articles for data abstraction. These 45 articles described 40 unique studies. The relationship of studies to the review questions is as follows: 18 studies relevant to KQ 1, 15 studies relevant to KQ 2, 2 studies relevant to KQ 3, 5 studies relevant to KQ 4, 1 study relevant to KQ 5, and 6 studies relevant to KQ6 (some studies were relevant to more than one KQ).

This figure depicts the flow of articles through the literature search and screening process. Searches of PubMed®, Embase®, PsycINFO®, and CDSR yielded 5,059 citations, 1,528 of which were duplicate citations. Manual searching identified 154 additional citations, for a total of 3,685 citations to be screened. After applying inclusion/exclusion criteria at the title-and-abstract level, 1,293 full-text articles were retrieved and screened. Of these, 1,248 were excluded at the full-text screening stage, leaving 45 articles for data abstraction. These 45 articles described 40 unique studies. The relationship of studies to the review questions is as follows: 18 studies relevant to KQ 1, 15 studies relevant to KQ 2, 2 studies relevant to KQ 3, 5 studies relevant to KQ 4, 1 study relevant to KQ 5, and 6 studies relevant to KQ6 (some studies were relevant to more than one KQ).

Figure 2

Literature flow diagram. a Some studies were relevant to more than one KQ.

Appendix C provides a detailed listing of included articles and associated publications that were used during abstraction to provide additional details on study methods. Appendix D provides a complete list of articles excluded at the full-text screening stage, with reasons for exclusion.

Description of Included Studies

Overall, we included 40 studies represented by 45 publications: 18 studies were relevant to KQ 1, 15 studies to KQ 2, 2 studies to KQ 3, 5 studies to KQ 4, 1 study to KQ 5, and 6 studies to KQ 6. Studies were conducted in Europe (33%), the United States or Canada (33%), the UK (15%), Asia (10%), Australia or New Zealand (7%), and other locations (2%). Further details on the studies included for each KQ are provided in the relevant results sections, below, and in Appendix E. Forty of these studies reported results for women, while one reported on the test characteristics of the Edinburgh Postnatal Depression Scale (EPDS) in fathers.

As described in the Methods chapter, we searched ClinicalTrials.gov to identify completed but unpublished studies as a mechanism for ascertaining publication bias. Our search yielded 117 trial records. A single reviewer identified six of these records as potentially relevant. Of note, no other potentially relevant records beyond these six were identified from a parallel search of the WHO ICTRP registry platform. Three of the six identified studies of interest are complete; we identified and screened publications for all three of these studies.

Two of the six identified studies are currently recruiting participants. Of these, 1 study with a target enrollment of 650 women was considered potentially applicable to KQs 4, 5, and 6. The other, with a targeted enrollment of 30 women, was considered potentially relevant to KQ 6. Upon completion, these 2 studies may provide additional evidence on the comparative benefits and/or harms of screening, or on how various factors affect the likelihood of an appropriate action after a positive screening result. One additional study with a targeted enrollment of 170 women was identified that may be relevant to KQ 6; however, that study record has not been updated since its initial entry into ClinicalTrials.gov in 2008. At that time, the study was not yet open for participant recruitment. We did not find any further information suggesting that this study has since progressed to begin enrolling. In summary, our search of ClinicalTrials.gov did not find evidence for completed but unpublished studies relevant to our KQs.

Key Question 1. Performance Characteristics of Screening Instruments

KQ 1.

This question has two parts:

  1. What are the sensitivity and specificity of currently available screening instruments for detecting postpartum depression, and how do these translate into the likelihood of false-negative and false-positive results in different populations and settings?
  2. Are there clinically relevant differences in the ability of currently available screening instruments to correctly identify specific signs or symptoms of depression (e.g., suicidal ideation)?

Key Points

  • Studies of individual screening tests rarely used the same threshold, preventing meaningful quantitative comparison or synthesis.
  • For any given screening test, sensitivity was generally higher and specificity lower as the threshold for postpartum depression on the screening test was lowered.
  • Precision was better for specificity estimates than for sensitivity estimates.
  • Both sensitivity and specificity generally were in the 80 to 90% range for most screening tests across studies.
  • In two studies at low risk of bias, a two-question screen had a sensitivity of 100 percent (95% confidence intervals [CIs], 93.3 to 100% and 83.3 to 100%), with specificities of 44.0 percent (95% CI, 39.5 to 48.8%) and 64.5 percent (53.7 to 75.2%, suggesting that it is possible to use an initial simple step for selecting patients for more specific screening instruments.
  • In one study at low risk of bias, the 24-item Leverton Questionnaire had higher sensitivity (95.2%; 95% CI, 90.4 to 98.1%) and specificity (91.3%; 95% CI, 88.4 to 93.7%) than was generally observed for other screening instruments, but we did not identify any confirmatory studies in a U.S. population.
  • We did not identify any studies comparing the ability of screening instruments to correctly identify specific signs or symptoms of depression. One study found moderate agreement between suicidal ideation on the Edinburgh Postnatal Depression Scale (EPDS) and suicidal ideation on the diagnostic instrument, but suicidal ideation was not predictive of response to treatment. In another study comparing the EPDS to the Mood Spectrum Self-Report (MOODS-SR), suicidal ideation was more common in the EPDS than the MOODS-SR, although formal tests of agreement were not performed.
  • In one large study at high risk of bias, performance characteristics for self-administered instruments designed for screening (the EPDS and Beck Depression Inventory [BDI] were similar to those of an interview-based instrument typically used in diagnostic settings (the Hamilton Rating Scale for Depression [HRSD]).
  • For the EPDS and Postpartum Depression Screening Scale (PDSS), the two most commonly studied screening tools, the overall strength of evidence was moderate that both sensitivity and specificity for major depression are generally in the 80–90 percent range at commonly used thresholds; as sensitivity increases with choice of thresholds, specificity decreases and vice versa. The evidence was insufficient for other screening tests. The evidence was also insufficient to determine if there are any clinically meaningful differences in test characteristics between individual screening tests.
  • One study judged at high risk of bias reported test characteristics for the EPDS in fathers tested 7 weeks after birth (sensitivity 89.5%; 95% CI, 66.9 to 98.7%; specificity 78.2%; 95% CI, 71.3 to 84.2% at a threshold of 10) that are similar to those reported for mothers.

Description of Included Studies

We identified 18 studies that met the inclusion criteria for KQ 1.82101 Two of these studies are each represented by two included publications. The 200490 and 200689 publications by Felice et al. describe results for the same study population, as do the 200991 and 201192 publications by Gjerdingen et al. All 18 studies confirmed the diagnosis of depression using a validated clinical interview or diagnostic instrument in screen positives and all or a sample of screen negatives. Four studies were performed in the United States,84,85,91,92,96 six in Europe,87,89,90,95,9799 four in the UK,83,88,94,100 and one each in Australia,82 New Zealand,101 Asia,93 and Canada.86 Ten studies were judged to have a high risk of biased results;8284,86,88,9597,99,101 the remainder were judged to be at low risk. One of the 18 studies focused on fathers88 and the rest on mothers.

Because no more than two studies provided results for the same test at the same threshold, we did not perform meta-analyses. Below, we present and discuss the results of the studies for each screening test qualitatively, then present the results for the three studies where two or more screening tests were directly compared. Only two studies94,99 were relevant to KQ 1b.

Eleven studies provided sensitivity and specificity data on the Edinburgh Postnatal Depression Scale (EPDS), four on the Postpartum Depression Screening Scale (PDSS), four on various versions of the Beck Depression Inventory (BDI), two on various versions of the Patient Health Questionnaire (PHQ), and one each on the Antenatal Risk Questionnaire, the 17- and 21-Item Hamilton Rating Scale for Depression (HRSD-17 and HRSD-21), and the Leverton Questionnaire.

Detailed Synthesis

Edinburgh Postnatal Depression Scale (EPDS)

Eleven studies provided sensitivity and specificity data for major depression for the EPDS.83,85,86,8890,9397,101 Studies varied in the threshold used to define a positive screening test. There was a clear trend toward increasing sensitivity and decreasing specificity as the threshold value decreased (Figures 3 and 4). For sensitivity, confidence intervals were wide and overlapped, except for the studies that used thresholds of 895 and 13.85 Even though, as expected, confidence intervals were considerably narrower for specificity, there was again considerable overlap across thresholds from 10 through 12.

There was a clear trend toward increasing sensitivity and decreasing specificity as the threshold value decreased. Confidence intervals were wide and overlapped, except for the studies that used thresholds of 8 and 13.

Figure 3

Sensitivity of the EPDS at various thresholds. CI = confidence interval; EPDS = Edinburgh Postnatal Depression Scale Note: Data from Ekeroma 2012 are for the Samoan subgroup.

There was a clear trend toward increasing sensitivity and decreasing specificity as the threshold value decreased. Even through confidence intervals were considerably narrower for specificity, there was considerable overlap across thresholds from 10 through 12.

Figure 4

Specificity of the EPDS at various thresholds. CI = confidence interval; EPDS = Edinburgh Postnatal Depression Scale Note: Data from Ekeroma 2012 are for the Samoan subgroup.

Of note, one of these studies88 was performed in 189 fathers, with test characteristics (sensitivity 89.5%; 95% CI, 66.9 to 98.7%; specificity 78.2 %; 95% CI, 71.3 to 84.2%) quite similar to those observed in women at the same threshold as seen in Figures 3 and 4. The study was judged to be at high risk of bias because of relatively low participation, and the reference standard was preferentially applied to men with high scores on the screening test, which creates potential for ascertainment bias (overestimation of sensitivity).

Postpartum Depression Screening Scale (PDSS)

Four studies provided sensitivity and specificity data for the PDSS across a range of thresholds.8486,98 Of note, Beck et al.84 was a validation study of a long- and short-form Spanish version of the PDSS in a U.S. Latina population and presented results primarily for combined major and minor depression. Figures 5 and 6 depict results for the Beck study84 (major and minor depression combined), and for five studies where the outcome was major depression alone; they also indicate whether the long- or short-form PDSS was used. As with EPDS, confidence limits were wider for sensitivity than for specificity, and there was a clear trend toward increasing sensitivity and decreasing specificity as thresholds decreased. Qualitatively, the values for sensitivity and specificity at a given threshold were similar, with sensitivities between 80 and 90 percent associated with specificities in the same range.

Confidence limits are wider for sensitivity than for specificity, and there is a clear trend toward increasing sensitivity and decreasing specificity as thresholds decreased. Qualitatively, the values for sensitivity and specificity at a given threshold were similar, with sensitivities between 80 and 90 percent associated with specificities in the same range.

Figure 5

Sensitivity of the PDSS at various thresholds. CI = confidence interval; PDSS = Postpartum Depression Screening Scale

Confidence limits are wider for sensitivity than for specificity, and there is a clear trend toward increasing sensitivity and decreasing specificity as thresholds decreased. Qualitatively, the values for sensitivity and specificity at a given threshold were similar, with sensitivities between 80 and 90 percent associated with specificities in the same range.

Figure 6

Specificity of the PDSS at various thresholds. CI = confidence interval; PDSS = Postpartum Depression Screening Scale

Other Tests

There were only one to two studies providing data for each of the other screening tests of interest in this review. Table 6 summarizes sensitivity/specificity results for these studies.

Table 6. Test characteristics for postpartum depression screening instruments other than EPDS or PDSS.

Table 6

Test characteristics for postpartum depression screening instruments other than EPDS or PDSS.

Results for these other tests were generally consistent with those for the EPDS and PDSS: precision of the estimate was greater for specificity than for sensitivity, and both sensitivity and specificity point estimates were generally in the 80–90 percent range. There were several exceptions to these general observations. The Antenatal Risk Questionnaire had a sensitivity at the low range of those of the other tests but a lower specificity. The BDI and the 25-item Leverton Questionnaire had both sensitivity and specificity above 90 percent in a Hungarian validation study.87 This study was rated as having a low risk of bias, and included 1,552 subjects.

Notably, a screen consisting of two questions (“During the past month, have you often been bothered by feeling down, depressed, or hopeless?” and “During the past month, have you often been bothered by having little interest or pleasure in doing things?”) had a sensitivity of 100 percent if the response to either question was yes, but low specificity.91,92,100

Within-Study Comparisons

Three studies compared different instruments in the same population (Figures 7 and 8).85,86,96 The largest96 compared the EPDS, BDI, HRSD-17, and HRSD-21 at multiple time points across pregnancy. Performance characteristics were similar for all the tests when performed from 8 weeks to 6 months postpartum, which are the time points depicted in Figures 7 and 8 (the similarity in characteristics across tests was also seen in the early postpartum period—differences within tests by timing are discussed under KQ 6). The HRSD is an interview-based instrument and is generally not considered a screening test—providing some evidence that self-administered instruments offer comparable performance to interview-based instruments. The two studies directly comparing the EPDS and PDSS85,86 found slightly lower sensitivity but higher specificity for the EPDS, depending on the threshold, but there was considerable overlap in confidence limits, especially for sensitivity.

Results for these other tests were generally consistent with those for the EPDS and PDSS: precision of the estimate was greater for specificity than for sensitivity, and both sensitivity and specificity point estimates were generally in the 80–90 percent range.

Figure 7

Comparative sensitivity of various screening instruments. BDI = Beck Depression Inventory; BDI-II = Beck Depression Inventory-II; EPDS = Edinburgh Postnatal Depression Scale HRSD-17 = 17-Item Hamilton Rating Scale for Depression; HRSD-21 = 21-Item Hamilton (more...)

Results for these other tests were generally consistent with those for the EPDS and PDSS: precision of the estimate was greater for specificity than for sensitivity, and both sensitivity and specificity point estimates were generally in the 80–90 percent range.

Figure 8

Comparative specificity of various screening instruments. BDI = Beck Depression Inventory; BDI-II = Beck Depression Inventory-II; EPDS = Edinburgh Postnatal Depression Scale HRSD-17 = 17-Item Hamilton Rating Scale for Depression; HRSD-21 = 21-Item Hamilton (more...)

Specific Signs and Symptoms

We identified two studies, one performed in Scotland94 and one in Italy,99 that evaluated the suicidal ideation component of the EPDS.

Howard et al.94 performed a nested cohort study within an RCT for postpartum depression. The prevalence of suicidal ideation (Item 10) on the EPDS at 6 to 8 weeks was 9.0 percent (95% CI, 8.3 to 10.1%), with 4 percent (95% CI, 3.2 to 4.4%) reporting having thoughts of harming themselves sometimes or quite often. Agreement between suicidality in the EPDS versus the diagnostic instrument used in this study (CIS-R) was only moderate, with a kappa of 0.66; 68 percent of those with some suicidal ideation on the EPDS would have been defined as suicidal using CIS-R criteria. Suicidal ideation at baseline did not correlate with any outcome in the trial at 18 weeks postpartum, including SF-12 physical and mental scores and repeat EPDS score, even after adjustment for differences between those with and without suicidal ideation (younger, unmarried, unemployed or with an unemployed partner, and worse quality of relationship).

Mauri et al.,99 as part of a larger cohort study of perinatal depression, compared the prevalence of suicidal ideation on the EPDS to that on the Mood Spectrum Self-Report (MOODS-SR) at 1, 3, 6, 9, and 12 months postpartum. Period prevalence for suicidal ideation during the postpartum period was 8.6 percent (95% CI, 7.4 to 9.8%) for the EPDS and 4.3 percent (95% CI, 3.4 to 5.2%) for the MOODS-SR. Point prevalence generally declined over time, with estimates based on the EPDS consistently higher than for the MOODS-SR at every time point (prevalence at 1 month postpartum for EPDS was 2.7%, 95% CI, 2.1 to 3.3%; for MOODS-SR it was 1.2%, 95% CI, 0.8 to 1.6%). Formal statistics on agreement between the two instruments were not provided, and insufficient data were provided to allow calculation of kappa statistics. Twenty-five percent of those with suicidal ideation on the EPDS also would have also met criteria on the MOODS-SR. In multivariate analysis, suicidal ideation on the EPDS was associated with major depression during pregnancy or the postpartum period.

Key Question 2. Effect of Individual Subject Factors on Screening Performance

KQ 2.

This question has two parts:

  1. Are there individual factors (age, race, parity [number of live births], history of mood disorders, history of intimate partner violence, perinatal outcomes, cultural factors) that affect the baseline risk of postpartum depression and, therefore, the subsequent positive and negative predictive values of screening instruments?
  2. Are there validated predictive models or algorithms based on such factors that would improve the performance of screening instruments?

Key Points

  • The positive and negative predictive values of screening for postpartum depression are affected by the prevalence of depression; screening women who are at higher risk would improve the positive predictive value.
  • We did not identify any studies that explicitly and directly compared the predictive values of screening instruments in different populations; only one study reported on potential differences in test sensitivity and specificity based on the presence of a specific characteristic or risk factor.
  • Maternal age and socioeconomic status were generally not associated with risk of postpartum depression; maternal unemployment increased the risk of postpartum depression in one study. The overall strength of evidence was low.
  • Complications of pregnancy, including preterm birth, low birthweight, and fetal abnormalities are associated with an increased risk of postpartum depression. Parity was not consistently associated with postpartum depression. One study found no significant effect of parity on test characteristics for screening tests. The overall strength of evidence was moderate.
  • Chronic medical conditions predating pregnancy may increase the risk of postpartum depression. The strength of evidence was low.
  • Past history of depression or anxiety, whether or not associated with a previous pregnancy, consistently increases the risk of postpartum depression. The strength of evidence was moderate.
  • Poor relationship quality and poor social support consistently increase the risk of postpartum depression. The strength of evidence was moderate.
  • All of the associations noted above are consistent with findings of studies published prior to our search dates and noted in recent reviews.
  • We did not identify any studies of clinical predictive models or algorithms (comparable to the Gail model for breast cancer risk) for improving the performance of screening instruments.

Description of Included Studies

We identified 16 articles describing 15 unique studies that met the inclusion criteria for KQ 2.95,96,102115 (The 2005109 and 2008108 publications by Chee et al. described results for the same study population.) Three were from the United States,96,102,105 seven were from Europe,95,103,110114 two (three publications) were Asian,108,109,115 and there was one study each from the UK,104 Australia,107 and Israel.106 Applying QUADAS-2 criteria across the studies applicable to KQ 2, 2 studies were judged to be low risk of bias,104,115 10 high risk of bias,95,96,102,105,106,108112,114 and 3 unclear risk of bias.103,107,113 We did not identify any studies relevant to KQ 2b. One study judged to be at high risk of bias96 did not provide an estimate of the association between parity and postpartum depression, but did provide separate estimates of screening test sensitivity and specificity stratified by parity.

Because of the inconsistency in how specific risk factors were described in the studies, we were unable to perform quantitative synthesis of the results.

Detailed Synthesis

Because we were unable to perform meta-analyses for any of the risk factors, we summarize the results in a series of tables below. Unless otherwise noted, results in the tables are presented for the final multivariate analysis (usually logistic regression) presented in each paper and represent the results for an outcome of major depression; in the list of variables, the bold text refers to the specific predictor for which measures of association were presented (e.g., maternal age). Some reported associations were not amenable to display in the table and are discussed separately in the text.

Among potential maternal demographic risk factors (Table 7), no statistically significant association was found between maternal age, education, income, or type of employment. One study95 did, however, find a significant association between maternal unemployment and postpartum depression (OR 2.8; 95% CI, 1.1 to 4.9).

Table 7. Maternal demographic risk factors for postpartum major depression.

Table 7

Maternal demographic risk factors for postpartum major depression.

Table 8 summarizes the available data on potential risk factors relating to obstetric history. As shown there, having a preterm95 or very low birthweight102 baby were both significantly associated with postpartum depression. In another study not shown in Table 8,114 having a termination of pregnancy for a severe fetal malformation or chromosomal abnormality in the second or third trimester was associated with a significant risk of depression 14 months after the event compared with women with healthy children, with better social support reducing the risk of depression; there was no control group of women with similar fetal abnormalities who did not undergo termination. As shown in Table 8, higher parity was significantly associated with increased risk of depression in one study,102 with a positive but nonsignificant association in another.103 One study102 reported a significant increase in risk of depression with prenatal smoking, while another103 reported a nonsignificant decrease in risk among smokers.

Table 8. Obstetric history risk factors for postpartum depression.

Table 8

Obstetric history risk factors for postpartum depression.

One study96 estimated sensitivity and specificity for the BDI, EPDS, HRSD-17, and HRSD-21 separately based on timing of screening (discharge to 8 weeks postpartum vs. 8 weeks to 6 months postpartum) and by gravidity (primigravid vs. multigravid) (Table 9). Although both the BDI and EPDS had higher sensitivity in primigravidas during both time periods, confidence intervals were wide and overlapped.

Table 9. Sensitivity and specificity of screening instruments by timing of screening and gravidity.

Table 9

Sensitivity and specificity of screening instruments by timing of screening and gravidity.

Among potential general medical history risk factors (Table 10), fair/poor self-reported health status102 and a history of chronic illness outside of pregnancy103 both increased the risk of postpartum depression over two-fold. One small study not shown in Table 10 found a significant association between maternal epilepsy and postpartum depressive symptoms;112 however, the study was underpowered to detect the potential impact of antiepileptic drugs on affective symptoms.

Table 10. General medical history risk factors for postpartum depression.

Table 10

General medical history risk factors for postpartum depression.

Past history of depression or anxiety, including both postpartum and before pregnancy, were consistently associated with an increased risk of postpartum depression, with odds ratios well above 2.0 (Table 11). Two studies also found that certain personality traits (neuroticism, vulnerability, low organization) were risk factors for depression.107,113

Table 11. Psychiatric history risk factors for postpartum major depression.

Table 11

Psychiatric history risk factors for postpartum major depression.

Finally, although studies used a variety of different scales to measure the effect of relationship quality and social support on risk of depression, and were conducted in a wide range of settings ranging from urban United States to Singapore, the qualitative results were consistent: postpartum depression was significantly more common among women in poorer quality relationships (or no relationship), and among women with poor social support (Table 12).

Table 12. Relationship and social support risk factors for postpartum major depression.

Table 12

Relationship and social support risk factors for postpartum major depression.

Key Question 3. Effect of Testing Variables on Screening Performance

Are the performance characteristics (sensitivity, specificity, predictive values) of screening instruments affected by:

  1. Timing (prenatal, peripartum, or at various times in the first postpartum year) and frequency of screening?
  2. Setting (prenatal visit, hospital/birthing center/home, postpartum maternal visit, or well-child visit)?
  3. Provider (obstetrician, midwife, pediatrician, family practitioner, other health provider)?

Key Points

  • Screening instrument performance characteristics vary by timing of administration, but the absolute difference in sensitivity and specificity across different time points is relatively small.
  • Screening for postpartum depression in the immediate postpartum period, e.g. within the first week postpartum, likely identifies only those women at highest risk of developing depression and misses those with a slower onset of symptomatology.
  • When screening for depression within the first 6 weeks postpartum, the Edinburgh Postnatal Depression Scale (EPDS), 17-Item Hamilton Rating Scale for Depression (HRSD-17), 21-Item Hamilton Rating Scale for Depression (HRSD-21), and Beck Depression Inventory (BDI) appear to have equivalent performance when using optimal instrument-dependent cutoffs.

Description of Included Studies

Two studies met the inclusion criteria for KQ 3a.96,116 No studies were identified that met the inclusion criteria for KQ 3b or KQ 3c.

The first study, a prospective investigation of maternal mental illness conducted at a single academic center in the United States,96 enrolled women prior to 28 weeks gestation and followed them through 6 months postpartum. Participants completed the EPDS, BDI, HRSD-17, and HRSD-21 during six perinatal windows: preconception, first trimester, second trimester, third trimester, early postpartum (0–6 weeks), and later postpartum (7–26 weeks). The diagnosis of depression was confirmed by the Mood Module of the Structured Clinical Interview for Depression (SCID).

The second study116 was a single-center prospective investigation conducted in Dublin, Ireland. This study enrolled women during the immediate postpartum period to determine if the EPDS, administered prior to hospital discharge, was predictive of depression at 6 weeks postpartum. Nine hundred fifty-one enrolled women completed the EPDS at 3–5 days postpartum with planned followup at 6 weeks postpartum for repeat EPDS and diagnostic interview using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) if screen-positive.

Detailed Synthesis

In the study by Ji et al.,96 results from 534 of 708 enrolled women were analyzed to construct receiver operating characteristic (ROC) curves to determine optimal cutoffs for the four screening instruments used. For diagnosis of postpartum depression, the HRSD-17, HRSD-21, and BDI had an optimal cutoff of 14 in both the early (PP-E) and late (PP-L) postpartum periods (see Table 13). The HRSD-17 was more sensitive but less specific at the PP-E time point compared with the PP-L time point, while the HRSD-21 did not differ in sensitivity by time point. Performance of the BDI was essentially the same at both time points. However, the optimal cutoff for the EPDS increased from 11 at PP-E to 12 at PP-L. The EPDS was more sensitive but less specific at the PP-E time point compared with the PP-L time point.

Table 13. Sensitivity and specificity of screening instruments in the early and late postpartum periods.

Table 13

Sensitivity and specificity of screening instruments in the early and late postpartum periods.

This study had a high risk of bias. There were multiple screening tests, and the order of administration was not described. The order of administration and the potential for repetitive questions may influence response. Finally, the timing of the followup diagnostic evaluation was not specified.

In the study by Crotty et al.,116 625 of 951 women completed a second EPDS at 6 weeks postpartum (66% response rate). Ninety of the 170 women who scored above the cutoff of 12 agreed to diagnostic testing using the SCAN interview. Twenty-three women scoring below 12 also completed the SCAN interview. While the early EPDS identified 58 percent (28 of 48) of women with a confirmed diagnosis of depression, 20 women who subsequently had a high EPDS score at 6 weeks and confirmed depression would have been missed (false negatives). Therefore, while EPDS screening in the immediate postpartum period may identify women at high risk for early development of postpartum depression, sole reliance on screening at this time point is inadequate.

The risk of bias was deemed to be high for this study based on the convenience sampling (recruitment 3 days per week up to a maximum of 10 women per day) and on enrollment up to 3 to 5 days postpartum, as women who remain hospitalized longer may have more medical complications and thus be at higher risk of postpartum depression.

Key Question 4. Comparative Benefits of Screening

What are the comparative benefits of screening for postpartum depression when compared with no screening, or between different screening strategies (based on choice of screening instrument, timing, setting, etc.)?

Key Points

  • Change in a screening instrument depression score was the most common outcome used to assess the comparative benefits of screening for postpartum depression.
  • Across a variety of low-intensity interventions, screening was associated with modest improvements in depression.
  • One good-quality U.K.–based RCT found that there is benefit to the overall mental health in mothers in screening, with initial treatment in screen positives, by health visitors at 6 weeks postnatally.
  • One fair-quality U.S.–based RCT found that a practice-level intervention to improve screening between 5 and 12 weeks postpartum compared with usual care in family medicine clinics led to lower levels of depressive symptoms at 6 and 12 months. A smaller, fair-quality Hong Kong–based RCT also found improvement in depressive symptoms, but no difference in overall mental health scores.
  • None of the three RCTs which included a measure of parental stress (the Parental Stress Inventory [PSI]) showed improvement in this measure with screening and treatment of depressive symptoms. The included studies do not allow an assessment of the comparative benefits of screening for postpartum depression by screening instrument, timing, or setting.
  • None of the studies included outcomes for fathers.

Description of Included Studies

Five studies met our inclusion criteria and evaluated the comparative benefits of screening for postpartum depression. Four were RCTs, and one was a quasi-experimental study. Among the four RCTs, one was rated as poor quality,117 two as fair,118,119 and one as good.120 The quasi-experimental study was rated as poor in quality.121 The most common relevant outcome was change in a screening instrument depression score. Sample sizes ranged from 99 recruited at a single site117 to 4,084 enrolled from 101 practices.120 Two studies were conducted in the United States.117,119 The remaining studies were conducted in the UK,120 Europe,121 and Asia.118

Detailed Synthesis

A good-quality cluster-randomized trial evaluated screening in 101 general practices in Trent, England.120 In the intervention practices (n=63), health visitors assessed the mother’s mood, administered the EPDS, and provided treatment based on either cognitive-behavioral principles or on person-centered principles. Women were sent a questionnaire by mail at 6 weeks postpartum that included the EPDS and an assessment of other factors potentially associated with depression including social support, stressful life events, threatening experiences, and previous depression. The threshold used for a positive EPDS was 12. Repeat mailings were made to women whose infants were 6, 12, and 18 months old. No specific intervention was given in the control practices (n=38). The EPDS was commonly used in these practices for women whose infants were 6 weeks old. However, unlike the health visitors in the intervention practices, those in the control practices did not provide treatment but instead referred women who screened positive for depression. The SF-12 mental component summary (SF-MCS) and physical component summary (SF-PCS) were used to assess the impact of the intervention at 6 and 12 months. Parenting stress was also measured at 6, 12, and 18 months using the PSI-Short Form (PSI-SF).

Among those who scored 12 or greater on the EPDS at 6 weeks (n=418), the unadjusted differences in the SF-12 MCS and SF-12 PCS between the control and intervention groups were 4.7 (95% CI, 1.8 to 7.6) and −1.4 (95% CI, −3.5 to 0.7). After adjusting for the 6-week EPDS score, living alone, history of postpartum depression, and any life events, the differences in the SF-12 MCS continued to be statistically significant (5.2 [95% CI, 2.5 to 7.8]) and the difference in the SF-12 PCS remained statistically not significant (−1.7 [95% CI, −3.6 to 0.1]). Across all women in the trial (n=2,659), differences between the control and intervention groups in SF-12 MCS and SF-12 PCS did not reach statistical significance (p=0.1 and p=0.469, respectively). However, the SF-12 MCS was statistically significantly different in the whole population of women after adjustment for the previously described factors (1.4 [95% CI, 0.5 to 2.3]). Similar differences persisted across the 12- and 18-month assessments.

Among those who scored 12 or greater on the EPDS at 6 weeks, the unadjusted score on the PSI-SF was statistically significantly higher (p=0.001) in the intervention (n=211) compared with the control group (n=106), with a difference of 9.2 (95% CI, 4.8 to 13.7). At 12 months, the difference between the intervention (n=156) and the control groups (n=90) was 8.0 (95% CI, 3.1 to 13.0), and at 18 months, the difference between the intervention (n=82) and the control groups (n=46) was 9.1 (95% CI, 1.1 to 17.4).

The main potential source of bias for this study was the dropout rate. However, among women in the control group, 87.8 percent had followup at 6 weeks and 74.5 percent had followup at 6 months; in the intervention group, 82.8 percent had followup at 6 weeks and 68.4 percent had followup at 6 months. The study was not designed to determine which component of the intervention was effective, but instead looked at whether the entire bundle of interventions could improve outcomes.

The earliest of the five studies117 identified 201 women receiving public assistance and at high risk for postpartum depression based on a 17-item depression risk survey.122 Of these high-risk women, 32 were either already receiving treatment or met criteria for current depression or substance use disorder. Of the remaining 131, 70 (53.4%) were unable to be assessed because of disconnected phones, relocation, or refusal to return calls. The remaining 99 subjects were randomized to either standard antenatal care or to an intervention consisting of four 60-minute group sessions over a 4-week period and a 50-minute individual booster session after delivery. Depressive symptoms were assessed at baseline and at 3 months after delivery using the Beck Depression Inventory (BDI). Eighty-six (87%) of the enrollees had followup 3 months after delivery. Two subjects (4%) in the intervention group and eight (20%) in the standard care group had depression at 3 months based on the depression module of the Longitudinal Interval Follow-up Evaluation. However, there were no differences in depression severity based on the BDI or in social impairment based on the Range of Impaired Functioning Tool. This study was considered to be of poor quality because of the small sample size, differential dropout, and the lack of an intention-to-treat analysis.

A quasi-experimental posttest study121 compared two areas in Norway, one of which used public health nurses to evaluate women for postpartum depression using the EPDS and clinical assessment and, when necessary, to provide supportive counseling. Screening with the EPDS was offered at 6 weeks and at 3, 6, and 12 months postpartum. The main outcome was the Parenting Stress Index at 12 months postpartum. There was no difference in the overall Parenting Stress Index between the two groups, although there was greater improvement in the EPDS score over time among depressed women in the intervention group (−6.9 in the intervention group versus −4.4 in the control group; p=0.01). However, insufficient data were provided to assess the degree to which this change over time might have been related to the dropout rate. Across all subjects, the dropout rate was 14.5 percent at 3 months, 33.7 percent at 6 months, and 45.8 percent at 12 months. Insufficient data were presented to determine whether the dropout rate differed between groups. This study was rated as poor in quality because of the study design (post-evaluation of a natural experiment) and because of incomplete data about the effect of differential dropout.

In a fair-quality RCT, 28 of 33 primary care practices completed a study in which they were randomized into two arms: usual care with training about postpartum depression and an active arm with more extensive training and implementation of the EPDS for screening of women between 5 and 12 weeks postpartum.119 Initial followup within the practices for those with an elevated EPDS was with a practice-administered PHQ-9. Women in the usual care practices completed the EPDS and PHQ-9, which were submitted to a central study site instead of to the clinicians in the practice. Overall, there were 990 women in the usual care group and 1,353 in the intervention group. Among these, 255 (26%) in the usual care group and 399 (29%) in the intervention group had an EPDS ≥10 or PHQ-9 ≥10. Overall, women in the intervention group with elevated EPDS scores were more likely to be diagnosed with depression than those in the usual care group (66% vs. 41%; p=0.001). Similarly, those in the intervention group with elevated EPDS scores were more likely to receive medication (56% vs. 35%; p<0.001), counseling (20% vs. 11%; p=0.02), or both (60% vs. 37%; p<0.0001). At 12 months, the adjusted odds ratio for a 5-point or greater decrease in the PHQ-9 in the intervention group compared with the control group was 1.82 versus 1.74 (p<0.001). Interestingly, in the same multivariate analysis, higher scores on the Parenting Stress Index (PSI) at baseline were associated with significantly lower chance of a 5-point or greater decrease in the PHQ-9 in both the intervention and control groups. There was no significant difference in changes in scores at 12 months for either the PSI or the Dyad Adult Satisfaction (DAS-6) scale (a measure of relationship satisfaction). This study was considered to be fair quality primarily because the main outcome of depression was based on self-reported symptoms and a chart audit rather than a standardized clinical assessment.

Another fair-quality RCT evaluated the effectiveness of screening for postpartum depression with the EPDS compared with no screening.118 Participants (n=462) were mothers of 2-month-old babies attending maternal and child health centers in Hong Kong for routine care. In the intervention group (n=231), women attending the centers were screened using the EPDS. Those who scored above the cutoff of 9/10 or answered affirmatively to the suicidal ideation question were then offered nondirective counseling by a maternal and child health (MCH) nurse or by a member of a community psychiatric team. Women randomized to the control group (n=231) received care as usual, which consisted of clinical assessment by an MCH nurse. If this clinical assessment suggested further management, women in the control group were offered the same services as those in the intervention group, namely, nondirective counseling by an MCH nurse or by a member of a community psychiatric team. The same MCH nurse provided counseling to both groups. Participants in both groups completed a set of questionnaires (including the EPDS) at 6 and 18 months postpartum.

A total of 67 women in the intervention arm screened positive for postpartum depression; of these, 51 (76.1%) received treatment. In the control group, 14 women were assessed as having postpartum depression, and 10 (71.4%) received treatment. Based on an intention-to-treat analysis, fewer women in the intervention group than in the usual care group had EPDS scores above the designed cutoff at 6 months postpartum (13% vs. 22.1%; RR 0.59; 95% CI, 0.39 to 0.89). This difference in EPDS scores remained statistically significant after adjusting for marital relationship at 2 months, history of psychiatric illness, depression during pregnancy, and relationship with mother-in-law (analysis not reported). There were no statistically significant differences between groups on measures of maternal well-being (General Health Questionnaire-12 ([GHQ-12]) or parenting (Parenting Stress Index [PSI]). However, children of screened women had more visits to the doctors than did children of women in the usual care group (p=0.039), even after adjusting for possible differences in baseline health status. This study was considered to be of fair quality because there appear to be baseline differences in the groups with lack of clarity on how adjustment was performed; these differences also suggest potential problems with the randomization process. In addition, missing data were imputed by means of group substitution at followup, which may bias the findings.

These five studies do not allow an assessment of the comparative benefits of screening for postpartum depression by screening instrument, timing, or setting. However, the good-quality RCT120 suggests that screening provides a benefit to overall mental health in mothers based on the SF-12 MCS. Similarly, two fair-quality RCTs118,119 found a benefit from screening with the EPDS, with reduced levels of depressive symptoms. Although there is no direct evidence of differences in setting, it is notable that the two studies with the greatest effect sizes119,120 tested strategies where treatment was provided within the same setting as screening (home visitation or family practice clinic), rather than a setting where further management of women with positive screening results required referral to a different provider. It is unclear how these observed benefits translate into improved quality of life, family functioning, or other health outcomes, especially since the three studies that collected data on a measure of parental functioning found no difference between groups. Interpreting changes in depression screening scores is challenging because of the fluctuations in these scores over time.

Key Question 5. Comparative Harms of Screening

What are the comparative harms of screening for postpartum depression when compared with no screening, or between different screening strategies (based on choice of screening instrument, timing, setting, etc.)?

Key Points

  • Only one study reported potential harms of screening for PPD. Children of women randomized to screening had more doctor visits, even after adjustment for baseline health, than did women in a control group. It is unclear whether this difference represents overutilization on the part of the screened group, or underutilization by the unscreened group.

Description of Included Study

Only one study met the inclusion criteria for KQ 5; this was a fair-quality RCT conducted in Asia.118 Most women in the study were married (95.5%) and had no past history of psychiatric illnesses (98.4%).

Detailed Synthesis

In the Hong Kong RCT described above,118 children of screened women had more doctor visits (mean 2.39; 95% CI, 2.07 to 2.7) compared with children of women in the usual care group (mean 1.97; 95% CI, 1.72 to 2.21; p=0.039) at 3 months, without any evidence of differences in child health status. This difference was no longer statistically significant at 18 months (mean visits in screened group 5.14 [95% CI, 4.57 to 5.71]; mean visits in control group 4.97 [95% CI, 4.58 to 5.36]). This study was considered to be of fair quality because there appear to be baseline differences in the groups with lack of adjustment. In addition, missing data were imputed by means of group substitution at followup, which may bias the findings. Although adjustment of baseline health status suggests that these differences in visit utilization may reflect differences in the appropriateness of the visits, there is no evidence to suggest whether any differences were due to overutilization among the screened or underutilization among the unscreened.

Key Question 6. Factors Affecting the Likelihood of an Appropriate Action After a Positive Screening Result

Is the likelihood of an appropriate action (referral, diagnosis, treatment, etc.) after a positive screening result affected by timing, setting, patient characteristics, or other factors?

Key Points

  • The EPDS was the most common screening tool used across studies.
  • Overall rates of referral and treatment for women who screened positive for postpartum depression were low, ranging from 0–30 percent, except for one trial where screening, diagnosis, and treatment were all conducted within a primary care setting, and where 60 percent rates of treatment were achieved.
  • Evidence on the effect of timing of screening on referral rates was mixed:
    • One good-quality cross-sectional study found that women who screened positive for depression at delivery had a higher proportion of psychiatric followup than those who screened positive prenatally or at 6 weeks postpartum.
    • A fair-quality prospective cohort study reported that women who screened positive at 6 weeks postpartum had lower rates of referral and treatment for symptoms of anxiety and depression than women who screened positive during the third trimester.
    • Conversely, a poor-quality prospective cohort study conducted in a private-practice setting found that, while all women who screened positive for depression received a referral for followup care, no women who screened positive during pregnancy sought care, and only 18 percent of those who screened positive at 6 weeks postpartum sought care.
  • A fair quality RCT demonstrated high levels of receipt of appropriate services among primary care practices where screening and treatment occurred within the same setting. These levels were substantially higher than those reported in other settings and were associated with significant improvement in depressive symptoms, but the study design precludes drawing inferences about the comparative effectiveness of screening in this type of setting compared with other settings.

Description of Included Studies

Six studies met the inclusion criteria for KQ 6. Two were prospective cohort studies,123,124 one was a cross-sectional study,125 one was a pre-post intervention study,126 one was a quasi-experimental design.127, and one was an RCT where practices were randomized to usual care or study intervention.119 One of the cohort studies were rated as fair quality,123 and one was rated as poor.124 The cross-sectional study was rated as good quality,125 the pre-post intervention study126 and quasi-experimental study127 were rated as poor quality, and the RCT as fair quality.119 All six studies were conducted in the United States.119,123127 All studies provided some measure of appropriate diagnosis and treatment of depression. Screening most commonly occurred in the first 8 weeks postpartum. Five of the six studies used the EPDS as the screening tool; the sixth study used the PRIME-MD PHQ.127

Detailed Synthesis

Timing

A good-quality cross-sectional study125 assessed 293 U.S. women at 36 weeks gestation, delivery, or the 6-week postpartum visit with the self-completed EPDS. The stated goal of the study was to assess the most advantageous timing for postpartum depression screening that optimized access to care. A cutoff of 10 was used to signal probable postpartum depression, and if a woman screened positive she was offered followup psychiatric services. The study assessed rates of psychiatric followup care for all women who screened positive at each time point. Overall, 12.6 percent of women screened positive for postpartum depression. However, prevalence varied across time: 5 percent screened positive at 36 weeks, 16 percent at delivery, and 14 percent at 6 weeks postpartum. Among those with positive screens, the proportion receiving psychiatric evaluation varied significantly with timing: 33 percent completed evaluations at 36 weeks, 100 percent at delivery, and 15 percent at 6 weeks postpartum (p<0.001). Prenatal and 6-week postpartum evaluation took place in outpatient settings, while the delivery assessment took place prior to discharge from the hospital, which likely contributed to the rates of completed evaluations. Of the 37 women who screened positive, 20 (54%) were subsequently diagnosed with depression, and 19 percent of these started treatment for depression.

A fair-quality prospective cohort study sought to examine detection, treatment, and referral of both postpartum depression and anxiety by obstetrical providers during pregnancy and at 6 weeks postpartum among 491 U.S. women.123 Postpartum depression was assessed with the EPDS, and a cutoff of 10 was used to indicate a positive screening result. Anxiety was assessed using the anxiety portions of the Patient Health Questionnaire (PHQ). Obstetric medical record reviews were used to assess documentation of mental health diagnosis, referral, and treatment. A total of 22.2 percent of women screened positive for postpartum depression, and 4.3 percent were positive for an anxiety disorder during the prenatal assessment in the third trimester. Only 46 of 113 women (41%) who screened positive during the third trimester had documentation of psychiatric symptoms or diagnosis by a provider in the medical record. Of those with medical records documentation, only 37 percent had further documentation of mental health treatment, and 43 percent (n=20) had documentation of a referral. Only 10 of the referred women (50%) accessed the referral. Thus, only 15 percent of women who screened positive for postpartum depression or anxiety had documentation of treatment during pregnancy, and an additional 18 percent had documentation of a referral for treatment. At 6 weeks postpartum, 17 percent (51 of 299) screened positive for postpartum depression and anxiety. Of this 17 percent, only 29.4 percent had documentation of psychiatric symptoms or diagnosis in the medical records, but nearly all (93%) had subsequent documentation of treatment or referrals for mental health as assessed by medical record review. Overall, only 27.5 percent (14 of 51) of women who screened positive for anxiety or depression at 6 weeks postpartum received any treatment or referral for mental health services. Thus, documented rates of referral and treatment were low overall. Women who screened positive at 6 weeks postpartum had slightly lower rates of referral and treatment for symptoms of anxiety and depression compared with women who screened positive during the third trimester (27.5% vs. 33%; p value NR). Again, no multivariate analysis was performed to assess predictors of referral, diagnosis, or treatment for depression.

A poor-quality quasi-experimental study127 sought to examine the impact of a Healthy Start depression treatment initiative in New Haven, Connecticut. The Healthy Start depression initiative consisted of mental health assessment with the PRIME-MD PHQ and referrals to services. Women who had depression could also attend weekly drop-in services that provided behavioral and pharmacological treatment. The study constructed three cohorts to assess the impact of Healthy Start on depression detection, referral, and treatment: a pre-Healthy Start depression initiative cohort, a post-Healthy Start cohort that was enrolled in the depression initiative, and a post-Healthy Start cohort that was not enrolled in the depression initiative. Propensity scoring was used to control for imbalance of baseline covariates. Rates of depression detection (p=0.003) and referral (p<0.001) were significantly different among the three groups, with the pre-Healthy Start group demonstrating the highest rates. The proportion of women in treatment for depression was not significantly different across groups (p=0.077); only 0.3 percent of women in the pre-Healthy Start group, 2 percent of women in enrolled Healthy Start, and 1 percent of women not enrolled in the Healthy Start program were in treatment for depression. The quasi-experimental study design created potential for selection, detection, and performance biases.

A poor-quality pre-post intervention study evaluated a brief obstetric clinic-based intervention on perinatal depression treatment in the context of a newly implemented policy of routine screening at a university-affiliated obstetric clinic in the United States.126 In accordance with the new policy, all women were screened at their first prenatal visit with the EPDS, and a score of 10 was used to signal probable depression and to prompt referral for further evaluation and treatment. A total of 1,298 new obstetric patients were screened for depression in accordance with this policy from November 2002 to January 2004. A total of 207 women (16%) scored above 10 on the EPDS, and 73 of these (35%) consented to be in the study and completed baseline interviews, which occurred 2 weeks after the second prenatal visit. The baseline survey included the Mood Disorders Module of the SCID for DSM-IV to obtain diagnosis of current or past depression. Depression treatment was assessed by self-report. Women were interviewed again 1 month after baseline and 6 weeks postpartum. The SCID was repeated at the 6-week postpartum interviews. The intervention consisted of notification to the treating physician of an elevated EPDS score via a flag in the medical record and nurse-delivered feedback to the patient on depression score, education about depression, and a referral for the patient occurring before the second prenatal visit. Based on medical record review and study interviews, authors constructed four time points for the assessment of depression treatment: Time 1, 3 months prior to the first prenatal visit; Time 2, time between first prenatal care visit and baseline prenatal interview; Time 3, time between the baseline prenatal interview and 1-month prenatal interview; and Time 4, time between 1-month prenatal interview and 6-week postpartum interview.

At baseline, 40 percent of the women in the study who screened positive for postpartum depression with the EPDS met diagnostic criteria for depression. At Time 1 (3 months prior to first prenatal visit), 16 percent of women with an EPDS of 10 or more were receiving some form of depression-related treatment as assessed by medical records review. At Time 2 (after EPDS screening and intervention), 21 percent of EPDS screened positive women self-reported that they were receiving treatment for depression, and this proportion remained constant through Time 3 (one month after baseline interviews). By 6 weeks postpartum (Time 4), 18 percent of EPDS screened positive women reported receiving treatment for depression. As part of the baseline survey (which occurred after routine EPDS screening and second prenatal visit), women were also asked if their physicians had discussed their elevated EPDS scores with them. The majority (67%) reported that their physician did discuss depression during their prenatal visit. Assessing a limited number of covariates, study investigators modeled the likelihood of depression treatment throughout the study using multivariable logistic regression. The only significant predictors of depression treatment were treatment prior to EPDS screening and greater depression severity as measured by the BDI-II. This study had a small, highly selected sample of women; only 35 percent (72 of 207) of women with an elevated EPDS consented to participate in the study and were followed over time.

A poor-quality prospective cohort study sought to determine whether universal depression screening during pregnancy and at 6 weeks postpartum affected rates of seeking treatment after recommendations for followup behavioral health assessments.124 The cohort consisted of 2,199 pregnant women who received obstetric care in a large multispecialty group practice in the United States. Postpartum depression was assessed using the EPDS. Patients scoring 9 or higher were alerted to their elevated score and encouraged by the obstetrician to seek a behavioral health provider. For patients who scored 14 or greater, the patient’s obstetrician or nurse assisted the patient during the visit to schedule a behavioral health appointment (unless the patient declined the referral). Of the 2,199 new obstetric patients screened during the universal screening program, 412 (18.7%) scored 9 or higher, and 102 (4.6%) scored 14 or higher. Of the 102 patients who scored 14 or higher, none followed the recommendations to be assessed by a behavioral health provider. Of the original cohort, 569 had progressed to the 6-week postpartum visit and had screening data available via a chart review. Of these 569 women, 28 (4.9%) has an EPDS of 14 or higher, and 5 (17.9%) had followed recommendations to seek care. There was no systematic analysis of factors affecting the probability of seeking additional care.

Setting

In a fair-quality RCT, 28 of 33 primary care practices completed a study in which they were randomized into two arms: usual care with training about postpartum depression and an active arm with more extensive training and implementation of the EPDS for screening of women between 5 and 12 weeks postpartum.119 Subjects in both arms completed the EPDS, but scores were not provided to the usual care sites. Initial followup within the intervention practices for those with an elevated EPDS was with a practice-administered PHQ-9. Women in the usual-care practices completed the EPDS and PHQ-9, which were submitted to a central study site instead of to the clinicians in the practice. Overall, there were 990 women in the usual care group and 1,353 in the intervention group. Among these, 255 (26%) in the usual care group and 399 (29%) in the intervention group had an EPDS ≥10 or PHQ-9 ≥10. Overall, women in the intervention group with elevated EPDS scores were more likely to be diagnosed with depression than those in the usual care group (66% vs. 41%; p=0.001). Similarly, those in the intervention group with elevated EPDS scores were more likely to receive medication (56% vs. 35%; p<0.001), counseling (20% vs. 11%; p=0.02), or both (60% vs. 37%; p<0.0001). These differences in treatment rates appear to be almost entirely due to differences in the initial detection of depression—rates of treatment were almost identical in the two groups among those women who did receive a diagnosis of depression (89.7% of women with a diagnosis of depression received medication and counseling in the usual care group vs. 90.7% in the intervention group). Although these rates of treatment are substantially higher than those reported in other settings, the study design, where usual care practices were blinded to EPDS scores, precludes drawing any direct inferences about whether provision of screening, diagnosis, and treatment within the same practice setting improves the likelihood of an appropriate response to an abnormal screening result.

Cover of Efficacy and Safety of Screening for Postpartum Depression
Efficacy and Safety of Screening for Postpartum Depression [Internet].
Comparative Effectiveness Reviews, No. 106.
Myers ER, Aubuchon-Endsley N, Bastian LA, et al.

Download

AHRQ (US Agency for Healthcare Research and Quality)

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...