| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (D.T.S.L., S.S.M.C., W.S.W.) and Department of Obstetrics and Gynaecology (A.S.K.Y., M.H.Y.T., W.S.W., T.K.H.C.), Chinese University of Hong Kong, Sha Tin, Hong Kong; and Department of Social Medicine (D.T.S.L.), Harvard Medical School, Boston, Massachusetts.
Address reprint requests to: Dr. Dominic TS Lee, Department of Psychiatry, Chinese University of Hong Kong, Sha Tin, Hong Kong. Email: tak_lee{at}hms.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: A prospective cohort of 145 women completed the Edinburgh Postnatal Depression Scale (EPDS), Beck Depression Inventory (BDI), and General Health Questionnaire (GHQ) within 2 days of delivery. Six weeks after delivery, the participants were interviewed by a psychiatrist, who used the Structured Clinical Interview for DSM-III-R (SCID, nonpatient version) to establish the diagnosis. The psychometric performance of the EPDS, BDI, and GHQ in detecting PPD was assessed using the SCID diagnosis as the gold standard.
RESULTS: When the cutoffs of the EPDS, BDI, and GHQ were lowered to achieve a sensitivity of 80%, the positive predictive values of these scales were to 13%, 18%, and 21%, respectively. When the cutoffs were raised to achieve a positive predictive value of 50%, the sensitivity rates were 6% (EPDS), 14% (GHQ), and 36% (BDI).
CONCLUSIONS: When commonly used depression rating scales were administered to identify PPD immediately after delivery, their psychometric properties were unsatisfactory. Healthcare providers should not screen for PPD in the first few days after delivery.
Key Words: postnatal depression, • screening, • rating scales.
Abbreviations: AUC = area under the curve;; BDI = Beck Depression Inventory;; EPDS = Edinburgh Postnatal Depression Scale;; GHQ = General Health Questionnaire;; NPV = negative predictive value;; PPD = postpartum depression;; PPV = positive predictive value;; ROC = receiver operating characteristics;; SCID-NP = nonpatient version of the Structured Clinical Interview for DSM-III-R;; SENS = sensitivity;; SPEC = specificity.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Identification of PPD can be improved by sharpening the awareness and skills of healthcare professionals in eliciting depressive symptoms. Another approach, which is adopted by an increasing number of healthcare providers, is to systematically screen for PPD using self-report questionnaires (8). It has been shown that a systematic screening program effectively improves the identification of PPD, which in turn expedites treatment (9, 10) .
Most PPD research has used the second and third postpartum months as the timeframe for screening (8, 10, 11). Few studies have examined whether PPD can be identified in the immediate few days after delivery. Although it may be argued that some cases of PPD are not yet fully developed in the first postpartum week, there is growing data to suggest that antepartum or peridelivery depression scores are among the most powerful predictors of PPD (12–14). Elevated peridelivery depression scores may signify developing PPD.
It is thus legitimate to ask whether PPD screening can be conducted immediately after delivery as opposed to waiting until the second and third postpartum month. Administering PPD screening in this manner can potentially reduce the direct staff cost of the program (because screening can be done in the hospital instead of the community). In addition, the diagnosis and treatment can be brought forward by a few weeks.
We decided to examine the psychometric performance of three commonly used depression rating scales in screening for PPD within a few days after delivery.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Design
The initial interview was conducted in the ward by a research assistant within 2 days of delivery (ie, before "postpartum blues" set in). Sociodemographic, medical, and psychiatric data were collected when written informed consent was obtained. After the interview, participants completed the Chinese version of the Edinburgh Postnatal Depression Scale (EPDS) (8), General Health Questionnaire (GHQ) (15), and Beck Depression Inventory (BDI) (16).
At 6 weeks postpartum, the participants were assessed by one of the authors (D.T.S.L.), who was unaware of the results of prior assessments, using the Chinese nonpatient version of the Structured Clinical Interview for DSM-III-R (SCID-NP). The SCID-NP is a semistructured diagnostic interview used to reliably establish DSM-III-R diagnoses (17). Because the participants were assessed at 6 weeks postpartum, the SCID-NP was modified to make 6-week diagnoses instead of 1-week diagnoses. Modification was also made to the SCID-NP to allow diagnosis of DSM-IV minor depressive disorder (2-week period of at least two, but less than five, symptoms of depression, depressed mood or anhedonia being mandatory) (18). The study protocol was approved by the Institutional Review Board of the Chinese University of Hong Kong.
Rating Instruments
The EPDS, one of the best known screening tools for postnatal depression, is a 10-item self-report questionnaire that has high levels of reliability and validity (8). The instrument has been widely evaluated in English and translated versions (19). We translated and validated the Chinese version of the EPDS, which was shown to be useful in screening for postpartum and postmiscarriage depression in the Chinese population (20, 21) . The psychometric properties of the Chinese version are as good as those of the original English version (22).
The GHQ is a self-report questionnaire designed to detect psychiatric morbidity in general practice and medical outpatient settings (15). The scale has been shown to have good reliability and validity (15). It has been widely used in research and clinical settings, including screening for postnatal depression (23, 24) . The validated 12-item Chinese version was used in this study (25).
The BDI is a 21-item rating scale established to measure severity of depression (16). Previous studies have shown that the instrument has good internal consistency and convergent validity with psychiatrists’ ratings of depression severity or with other self-report measures of depression (26). The Chinese BDI has been validated, showing good validity and reliability (27).
Statistical Analysis
Participants were classified as cases or noncases of postnatal depression at 6 weeks postpartum on the basis of their SCID diagnosis. The methodology of the validation study by Cox (8) was adopted, and both major and minor depression were used to define caseness. Previous data have shown that more than 50% of cases of first-onset major depression are associated with earlier minor depression; screening for minor depression would therefore identify individuals who are suffering from subthreshold illness as well as those who are progressing to major depression. The psychometric performance of the GHQ and BDI was assessed by the sensitivity (the percentage of true "cases" identified by the instrument), specificity (the percentage of true "noncases" identified by the instrument), positive predictive value (PPV; the proportion of all those tested as positive who were correctly identified as such), and negative predictive value (NPV; the proportion of all those tested as negative who were correctly identified as such) using the SCID diagnosis as the gold standard criterion. The receiver operating characteristic (ROC) curve, a plot of sensitivity against 1-specificity at different cutoffs, was used to help identify the optimal cutoff score, which is usually the point when an increase in sensitivity is associated with a sharp drop in specificity. The area under the curve (AUC) was also calculated. In general, the higher the AUC, the better the balance between sensitivity and specificity.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
One hundred forty-five participants (66%) returned for follow-up assessment at 6 weeks. Seventy-five participants (34%) did not return for the follow-up appointment but completed the GHQ and BDI by phone. Participants who could not attend the face-to-face interview had lower BDI and GHQ scores at 6 weeks compared with those who returned for follow-up (Mann-Whitney U tests, p < .005). Those who did not attend the follow-up were otherwise not different from those with face-to-face assessment in terms of baseline EPDS, GHQ, and BDI scores and demographic and psychosocial characteristics.
A total of 145 participants were assessed using the SCID, and 17 women (12%) met the criteria for postnatal depression. The sensitivity, specificity, PPV, and NPV for the EPDS, GHQ, and BDI at various cutoff scores are listed in Table 1. The corresponding ROC curves are shown in Figure 1. The AUC for the EPDS, GHQ, and BDI was 0.63, 0.75, and 0.80 respectively. The ROC curve for the BDI when applied at 6 weeks postpartum to the same study population was included for comparison (23).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The unsatisfactory psychometric properties are unlikely be due to the rating scales chosen. The EPDS, GHQ, and BDI have all been shown to be useful in screening for PPD when applied around 6 weeks postpartum (8, 24). In an earlier study conducted in the same study population at 6 weeks postpartum, we found that using a cutoff of 9.5, the EPDS can detect 82% of the PPD cases with a 44% PPV (23). Comparable performance was found for the GHQ at 6 weeks (sensitivity of 88% and PPV of 52% with a cutoff of 4.5) and the BDI at 6 weeks (sensitivity of 82% and PPV of 50% with a cutoff of 10.5) in the same study. Hence, it is more likely that the timing of assessment, rather than the scales themselves, was responsible for the unsatisfactory psychometric performance observed.
Therefore, even though elevated peridelivery depression scores are associated with an increased risk of PPD, they cannot be used to accurately identify PPD. The level of depression for early PPD may not be substantial enough to allow differentiation from ordinary adjustment. Furthermore, some PPD cases begin only when the mothers return home to face the burden of childcare. Some new mothers do not develop PPD until they return to work. These cases will be difficult to anticipate if screening is conducted immediately after delivery.
Our study has a number of limitations. First, although we used a representative and community-based sampling frame, about 30% of the participants did not complete the study. Second, we did not study the blues symptoms and hence could not exclude the possibility of postpartum blues confounding the analysis. However, we deliberately administered the screening scales within 48 hours of delivery, a period generally regarded as unlikely for postpartum blues. Last, the present study was conducted in ethnic Chinese; replications in other populations are therefore essential.
We hope this study will generate more research on this important issue. As PPD gains growing recognition, healthcare providers will face growing pressure to offer proactive and responsive service. Even more, as the empirical evidence on early intervention accumulates, the demands for systematic postnatal depression screening programs will increase. In many places where suitable infrastructure (eg, maternal and child health center and program) and personnel (eg, community midwives and health visitors) are unavailable, there is a temptation to screen for PPD immediately after delivery instead of later on in the puerperium. Our data show that such screening is unsatisfactory (high false-positive) and unsafe (low sensitivity). It seems that the first few days after delivery is not an appropriate time to effectively screen for PPD.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication January 28, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  OTHER ARTICLES NOTED (25 Apr 2003 to 18 Jul 2003) Evid. Based Nurs., October 1, 2003; 6(4): e1 - 12. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
