This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lee, D. T. S. Articles by Chung, T. K. H. Search for Related Content PubMed PubMed Citation Articles by Lee, D. T. S. Articles by Chung, T. K. H. Related Collections Depression Sexual Medicine: Female

Screening for Postnatal Depression Using the Double-Test Strategy

From the Departments of Psychiatry (D.T.S.L., H.F.K.C.) and Obstetrics and Gynaecology (A.S.K.Y., T.K.H.C.), Chinese University of Hong Kong, Shatin, Hong Kong.

Address reprint requests to: Dr. Dominic T. S. Lee, Department of Social Medicine, Harvard Medical School, 641 Huntington Ave., Cambridge, MA 02215. Email: tak_lee{at}hms.harvard.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
OBJECTIVE: Postnatal depression affects 10% to 15% of women after childbirth. Self-report rating instruments, such as the Edinburgh Postnatal Depression Scale (EPDS), have been developed and administered to postpartum women to facilitate early detection. Most postnatal depression screening scales, however, focus solely on depressive symptomatology. We hypothesized that applying two complementary rating scales of symptoms and functioning as a double test would significantly enhance the positive predictive value of screening.

METHODS: A prospective cohort study was conducted at the postnatal clinic of a university teaching hospital. One hundred forty-five Chinese women completed the EPDS and 12-item General Health Questionnaire (GHQ) 6 weeks after delivery. They were then interviewed by a psychiatrist, who used the Structured Clinical Interview for third revised edition of the Diagnostic and Statistical Manual of Mental Disorders, nonpatient version (SCID-NP), to validate the diagnoses.

RESULTS: The positive predictive value of the EPDS and GHQ, when administered independently, was 44% and 52%, respectively, at their respective optimal cutoff scores. When the EPDS-GHQ double test was administered, the positive predictive value was significantly increased to 78%.

CONCLUSIONS: Simultaneous administration of the EPDS and GHQ can substantially improve identification of women with postnatal depression. This can potentially reduce unnecessary referrals to general practitioners and psychiatrists and may enhance the overall cost-effectiveness of population-wide screening.

Key Words: postnatal depression • screening • Edinburgh Postnatal Depression Scale • General HealthQuestionnaire

Abbreviations: CI = confidence interval; DSM-III-R = Diagnosticand Statistical Manual of Mental Disorders, third revisededition; EPDS = Edinburgh Postnatal Depression Scale; GHQ =General Health Questionnaire; SCID-NP = Structured ClinicalInterview for DSM-III-R, nonpatient version.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Postnatal depression is the most common disorder following childbirth, affecting 10% to 15% of mothers (1–4). Apart from inflicting distress on the mother, postnatal depression undermines the marital relationship, impairs emotional and cognitive development of the newborn child (5, 6), and may even lead to infanticide and maternal suicide (7). Left untreated, one-third and one-tenth of mothers continue to be depressed by the end of the first and the second postnatal years, respectively (8). Although postnatal depression is common and potentially serious, only a minority of the cases are identified by primary care health workers during routine care (9–13). The majority of depressed mothers in the community are unrecognized and therefore untreated.

Better detection of postnatal depression can be achieved by improving the skills of general practitioners, obstetricians, and midwives in recognizing and eliciting signs and symptoms of depression. Alternatively, paper-and-pencil self-report rating scales can be administered routinely to identify mothers who have substantial depressive symptomatology and hence require further assessment. The EPDS is the most widely used screening scale for postnatal depression (14). This 10-item self-report questionnaire inquires about affective symptoms and suicidal ideas and has been shown to be both reliable and valid in detecting depressive illness among postpartum women. In some countries, like the United Kingdom, the EPDS is administered systematically to postnatal women to ensure early detection of an eminently treatable condition.

Although paper-and-pencil self-report rating scales are generally simple, inexpensive, and easy to administer, they are not without limitations. To achieve high sensitivity (ie, a low false-negative rate), a lower cutoff score is generally chosen, which compromises the positive predictive value (false-positive rate) of the screening scale. For the EPDS, at a cutoff value of 12.5, the sensitivity is 86%, and the corresponding positive predictive value is 73%. This means that for every 100 high scorers, 27 of them do not meet the criteria of major or minor depression. Most of these false-positive cases, however, have a significant level of depressive symptomatology but do not meet the criteria of postnatal depression because there is no significant impairment of their daily functioning.

Theoretically, it may be possible to reduce these false-positive results and to improve the positive predictive value by adding items of functional impairment to the EPDS. This can be accomplished by constructing a modified EPDS or by using the EPDS in combination with a rating scale that assesses functional disability. The strategy of simultaneous administration of two or more screening measures as a double or triple test has been widely adopted in various branches of medicine. This strategy, however, has not been readily taken up in psychiatry.

In this study, we combined the EPDS with the GHQ to screen for postnatal depression. The GHQ was chosen because it is short, easy to administer, well established, and contains items on functional impairment. By defining a high scorer as an individual who scored above the cutoff points of both the EPDS and GHQ, we hoped to limit the high scorers to women who are not only depressed but also functionally impaired. We hypothesized that simultaneous administration of these two complementary psychometric scales would improve the positive predictive value and reduce the number of false-positive results in screening for postnatal depression.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Subjects
A prospective cohort study was conducted in the Prince of Wales Hospital, a university teaching hospital in Hong Kong, which has a catchment population of 1 million people with diverse socioeconomic backgrounds. Subjects comprised all Chinese women who were admitted to the postnatal ward of the Department of Obstetrics and Gynaecology over a 3-month period (November 1996 through January 1997). Non-Chinese women and those who did not have permanent residency rights in Hong Kong (eg, illegal immigrants) were excluded from the study. People who were illiterate were assisted by a research assistant in completing the questionnaires.

Rating Instruments
The EPDS is one of the best known screening tools for postnatal depression (14). Research evidence suggests that this 10-item self-report questionnaire has a high level of reliability and validity and is useful in helping health professionals detect depression. The instrument has been widely evaluated in white populations, in English and translated versions. We translated and validated the Chinese version of the EPDS, which was shown to be useful in screening for postnatal and postmiscarriage depression in the Chinese population (15, 16). The psychometric properties of the Chinese version are as good as those of the original English version.

The GHQ is a self-report questionnaire designed to detect psychiatric morbidity in general practice and medical outpatient settings. The scale has been shown to have good reliability and validity (17). It has been widely used in research and clinical settings, including screening for postnatal depression (18, 19). The 12-item version was used in this study (20).

Assessment
A research assistant recruited subjects within 2 days after admission to the ward. Written, informed consent was obtained before sociodemographic, medical, and psychiatric data were collected. Six weeks after delivery, subjects were interviewed again and asked to complete the EPDS and GHQ. They were then assessed with the Chinese version of the SCID-NP (21) by one of the authors (D.T.S.L.), who was unaware of the results of the prior assessment. The SCID-NP was used to establish the DSM-III-R psychiatric diagnosis (22). Because subjects were assessed 6 weeks after delivery, the SCID-NP was modified to make 6-week instead of 1-month diagnoses.

Although the SCID-NP is a semistructured interview, it allows the interviewer to use additional questions to inquire about idioms of distress that are specific to the local context. This ensures that the diagnostic interview is culturally informed and sensitive. Traditional Chinese culture stipulates that women should adhere to a variety of customs in the first month after childbirth. For instance, a recently parturient woman should avoid going out, being exposed to drafts, and contact with cold water. It is believed that failure to comply with these ethnomedical restrictions would cause "wind" to enter the body, leading to chronic poor health, headache, and rheumatism. Such cultural patterning of the postpartum period affects the clinical presentation of postnatal depression. Hence, instead of saying "I am depressed," some Chinese women may present with somatic complaints or physical idioms of distress, such as headache, head numbness, "wind inside the head," diffuse joint pain, or "wind illness" (23, 24). If an interviewer fails to elicit or acknowledge these emotional cues, the women may feel that their complaints are not validated and would be less ready to report their depressive symptoms. Eventually, this would lead to underdiagnosis of postpartum depression.

The study protocols were approved by the Research Ethics Committee of the Faculty of Medicine at the Chinese University of Hong Kong.

Statistics
The SCID diagnosis was used to categorize each subject as a case or noncase of postnatal depression at 6 weeks postpartum. We applied the methodology described by Cox et al. (14) and used both major and minor depression to define cases. It has been shown that more than 50% of cases of first-onset major depression are associated with an earlier presence of minor depression (25). Screening for minor depression would thus identify individuals who are suffering from subthreshold illness as well as those who are progressing to major depression. The psychometric properties (sensitivity, specificity, positive predictive value, and negative predictive value) of the EPDS and GHQ were first calculated independently, assuming that the scales were separately administered. The optimal cutoff score of each scale was then identified by construction of receiver operating characteristic curves.

Subsequently, the psychometric properties of a combined EPDS-GHQ screening test was calculated, and a high scorer was defined as an individual who scored above the optimal cutoff points of both the EPDS and GHQ.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Both the EPDS and the GHQ are brief, simple, and easy to complete. In general, women completed both questionnaires in less than 10 minutes. The scales can be scored in less than 2 minutes, even by persons unfamiliar with scoring the tests.

During the study period, 220 women agreed to participate in the study, and 145 (66%) were assessed 6 weeks after delivery. Women who did not return for the 6-week follow-up assessment were contacted by telephone. These women had lower scores on the EPDS and GHQ at 6 weeks than those who returned for the follow-up assessment (Mann-Whitney U test, p < .005). Nonattendees were not otherwise different from those who completed the face-to-face assessment in terms of baseline EPDS and GHQ scores or demographic and psychosocial characteristics.

The mean age of participating subjects was 29 years (range, 16–42 years). Most subjects (97%) were married, and only 3% cohabited. The median number of previous children was 1 (range, 0–6). Two percent of subjects had no formal education, 45% had primary education only, 46% had completed secondary education, and 7% had received a university education. Half of the subjects were housewives, 44% were employed full time, 1% worked part time, and 5% were unemployed. Socioeconomic status of the group, as rated by Registrar General Classification, was as follows: class 1 (highest socioeconomic group; upper), 1%; class 2 (upper middle), 15%; class 3 (middle), 78%; class 4 (lower middle), 5%; and class 5 (lowest socioeconomic group; lower), 1%. The characteristics of our subjects are similar to those of the local population; hence, the sample is representative of the community population. Five subjects (3.4%) were illiterate, and another 14 subjects (9.7%) required assistance from the research assistant to complete the questionnaires. The EPDS and GHQ scores of those who completed the questionnaires with assistance were not different from the scores of the rest of the sample (t test, p > .05).

Seventeen subjects (11.7%) met DSM-III-R criteria for postnatal depression. Figure 1 shows the receiver operating characteristic curves of the GHQ and EPDS. The sensitivity, specificity, positive predictive value, and negative predictive value of the scales, at their optimal cutoff scores (4.5 for the GHQ and 9.5 for the EPDS), are summarized in Table 1. The GHQ and EPDS have a positive predictive value of 0.52 (95% CI, 0.43–0.61) and 0.44 (95% CI, 0.35–0.53), respectively.

View larger version (11K): [in this window] [in a new window]   Fig. 1. Receiver operating characteristic curves of the EPDS and GHQ.

Participants were further divided into four groups on the basis of their EPDS and GHQ scores: 1) high scores on both the EPDS and GHQ, 2) high score on the EPDS and low score on the GHQ, 3) low score on the EPDS and high score on the GHQ, and 4) low scores on both the EPDS and GHQ. Table 2 summarizes the distribution of postnatal depression among these four groups. Compared with the group that scored high on both the EPDS and GHQ (ie, high score on the double test), the other three groups had a low to very low rate of postnatal depression (0–9%).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

Examination of the score distribution explains the underlying mechanism of the double test. The EPDS-GHQ double test requires a high scorer to score high on both the EPDS and GHQ. As a result, those who score high on the EPDS-GHQ double test have both depressive symptoms and impaired psychosocial functioning and are very likely to meet the diagnostic criteria of postnatal depression. Not surprisingly, the positive predictive value in this group of women is the highest (78%).

Most women who scored high on the GHQ but low on the EPDS had difficulties in coping with childbearing, but no functional impairments were evident in results of the SCID interview, and depressive symptoms were not prominent. Individuals who scored high on the EPDS but low on the GHQ were depressed but did not have significant functional impairment. As shown in Table 2, none of the subjects in this category met the criteria for postnatal depression, and the rate of depression in this group was 0%. Hence, by removing this category of individuals from the high scorer group, we eliminated 78% of the false-positive cases (14 of 18) and substantially improved the overall predictive performance of the screening measure.

A breakdown of the distribution of depressed cases among the four categories of participants showed that the subgroup of women scoring high on the EPDS and low on the GHQ is very different from the subgroup scoring high on both tests (Table 2). Although individuals in both subgroups scored above the EPDS cutoff point, the rate of depression in the former category is 0%, whereas that in the latter is 78%. It would hence be wrong to classify both subgroups together as high scorers, as is done when the EPDS alone is used. The EPDS-GHQ double test provides a more valid reflection of the distribution of depression among postpartum women by distinguishing between the subgroup of women scoring high on both the EPDS and GHQ and the other subgroups, which have substantially lower rates of depression.

Our data show that the improvement in predictive value is not accompanied by a reduction of sensitivity, which would normally be expected. This is because none of the individuals in the subgroup scoring high on the EPDS and low on the GHQ are clinically depressed. Thus, classifying this subgroup as low scorers on the double test did not lead to any false-negative results. This observation, however, is likely to be a chance finding because the rate of depression in the subgroup scoring low on both the EPDS and GHQ stratum was 2%. Hence, the rate of depression in the subgroup with high EPDS scores and low GHQ scores should be at least 2%.

Although the EPDS was used to assess depressive symptomatology in this study, other rating scales may perform similarly. In a separate study, we compared the performance of the Beck Depression Inventory (21 item), GHQ (12 item), and EPDS (10 item) in screening for postnatal depression and found that they were similar (26). Because the EPDS is shorter than the Beck Depression Inventory, the former was chosen for the double test.

Our study has several limitations. First, this study was conducted using the Chinese version of the EPDS, which has a positive predictive value (44%) lower than that reported for the English version (73%). We believe the lower positive value of the Chinese EPDS is due to a lower prevalence of postnatal depression in our study population (12%). The sensitivity and specificity of the Chinese EPDS (82% and 86%, respectively) are very close to those of the original English version (86% and 78%, respectively). In fact, the positive predictive value of the Chinese EPDS would be similar to that of the English version if the prevalence of depression were taken into account. Nonetheless, previous studies have shown that the optimal cutoff level of the EPDS varies with different populations (15, 27, 28). Hence, if the double test is to be used in other populations, validation studies are needed to identify the appropriate EPDS and GHQ cutoff scores for the population concerned.

In the present study, a 3-month psychiatric diagnosis was used as the gold standard for validation because it was assumed that treatment of postnatal depression would be based on a psychiatric diagnosis. This assumption, however, is contrary to clinical practice in some settings, in which treatment of postnatal depression is based on symptom level rather than diagnosis. To address this issue, a follow-up study is currently under way to examine outcomes in participants who had high EPDS scores but no disorders according to DSM-III-R criteria. This follow-up study will also provide information on the outcomes of participants who scored high on the GHQ and low on the EPDS.

Our findings may have substantial impact on service provision and future research. Identification of hidden psychiatric morbidity is an important public health issue. Screening questionnaires, among other strategies, have been shown to be useful in identifying postnatal depression. Lowering the false-positive rate in population screening for postnatal depression will reduce unnecessary alarm and referrals to general practitioners and psychiatrists. Combined administration of the EPDS and GHQ will improve correct identification of depressed women and may improve overall cost-effectiveness of postnatal depression screening programs, although this will need to be tested separately.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
This study was supported by Grant 621019 from the Health Services Research Fund. We thank W. Lam, S. Yip, K. P. M. Chan, I. O. L. Chau, and T. Leung for assistance. Thanks are also extended to A. Z. M. Chang and W. K. Wing for their comments and suggestions.

Received for publication May 13, 1999.

Revision received September 9, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 

1. O’Hara MW, Swain AM. Rates and risk of postpartum depression—a meta-analysis. Int Rev Psychiatry 1996; 8: 37–54.
2. Georgiopoulos AM, Bryan TL, Houston MS, Yawn BP, Rummans TA, Evans MP, McKeon KK, Therneau TM. Screening for postpartum depression in Olmsted County, Minnesota: a population-based study using the Edinburgh Postnatal Depression Scale [abstract]. Psychosomatics 1999; 40: 135.[Free Full Text]
3. Nonacs R, Cohen LS. Postpartum mood disorders: diagnosis and treatment guidelines. J Clin Psychiatry 1998; 59 (Suppl 2): 34–40.
4. Stowe ZN, Nemeroff CB. Women at risk for postpartum-onset major depression. Am J Obstet Gynecol 1995; 173: 639–45.[Medline]
5. Murray L, Cooper PJ, Stein A. Postnatal depression and infant development. BMJ 1991; 302: 978–9.
6. Hay DF, Kumar R. Interpreting the effects of mothers’ postnatal depression on children’s intelligence: a critique and re-analysis. Child Psychiatry Hum Dev 1995; 25: 165–81.[Medline]
7. Yip SK, Chung TKH, Lee TS. Suicide and maternal mortality in Hong Kong [letter]. Lancet 1997; 350: 1103.
8. Oates M. Psychiatric disorder and childbirth. Curr Obstet Gynaecol 1995; 5: 64–9.
9. Kendell RF, Chalmers I, Platz C. The epidemiology of puerperal psychoses. Br J Psychiatry 1987; 150: 662–73.[Abstract/Free Full Text]
10. Oates MR. Postnatal mental illness: organisation and function of a psychiatric service. In: Cox J, Holden J, editors. Aspects of perinatal psychiatry: use and misuses of the EPDS. London: Gaskell; 1994. p. 109–21.
11. Holden J. The role of health visitors in postnatal depression. Int Rev Psychiatry 1996; 8: 79–86.
12. Beck CT. Screening methods for postpartum depression. J Obstet Gynecol Neonatal Nurs 1995; 24: 308–12.[Medline]
13. Whitton A, Warner R, Appleby L. The pathway to care in post-natal depression: women’s attitudes to post-natal depression and its treatment. Br J Gen Pract 1996; 46: 427–8.[Medline]
14. Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression Scale. Br J Psychiatry 1987; 150: 782–6.[Abstract/Free Full Text]
15. Lee DTS, Yip SK, Chiu HFK, Leung TYS, Chan KPM, Chau IOL, Leung HCM, Chung TKH. Detecting postnatal depression in Chinese women: validation of the Chinese version of the Edinburgh Postnatal Depression Scale. Br J Psychiatry 1998; 172: 433–7.[Abstract/Free Full Text]
16. Lee DT, Wong CK, Ungvari GS, Cheung LP, Haines CJ, Chung TK. Screening psychiatric morbidity after miscarriage: application of the 30-item General Health Questionnaire and the Edinburgh Postnatal Depression Scale. Psychosom Med 1997; 59: 207–10.[Abstract/Free Full Text]
17. Goldberg D. Manual of the General Health Questionnaire. Windsor, United Kingdom: NFER-Nelson; 1978.
18. Kitamura T, Shima S, Sugawara M, Toda MA. Temporal variation of validity of self-rating questionnaires: repeated use of the General Health Questionnaire and Zung’s Self-Rating Depression Scale among women during antenatal and postnatal periods. Acta Psychiatr Scand 1994; 90: 446–50.[Medline]
19. Nott PN, Cutts S. Validation of the 30-item General Health Questionnaire in postpartum women. Psychol Med 1982; 12: 409–13.[Medline]
20. Pan PC, Goldberg DP. A comparison of the validity of GHQ-12 and CHQ-12 in Chinese primary care patients in Manchester. Psychol Med 1990; 20: 931–40.[Medline]
21. Spitzer RL, Williams JB, Gibbon M, First MB. The Structured Clinical Interview for DSM-III-R (SCID). I. History, rationale, and description. Arch Gen Psychiatry 1992; 49: 624–9.[Abstract]
22. DSM-III-R. Diagnostic and statistical manual of mental disorders. 3rd ed revised. Washington DC: American Psychiatric Association; 1987.
23. Pillsbury BLK. "Doing the month": confinement and convalescence of Chinese women after childbirth. Soc Sci Med 1978; 12: 11–22.
24. Eisenbruch M. ‘Wind illness’ or somatic depression? A case study in psychiatric anthropology. Br J Psychiatry 1983; 143: 323–6.[Abstract/Free Full Text]
25. Horwath E, Johnson J, Klerman GL, Weissman MM. Depressive symptoms as relative and attributable risk factors for first-onset major depression. Arch Gen Psychiatry 1992; 49: 817–23.[Abstract]
26. Lee DTS, Yip ASK, Chiu HFK, Leung TYS, Chung TKH. Screening for postnatal depression: are specific instruments mandatory? J Affect Disord. In press 2000.
27. Bergant AM, Nguyen T, Heim K, Ulmer H, Dapunt O. [German version and validation of the Edinburgh Postnatal Depression Scale]. Dtsch Med Wochenschr 1998;123:35–40. German.
28. Ghubash R, Abou-Saleh MT, Daradkeh TK. The validity of the Arabic Edinburgh Postnatal Depression Scale. Soc Psychiatry Psychiatr Epidemiol 1997; 32: 474–6.[Medline]

This article has been cited by other articles:

				P. K. Mallikarjun and F. Oyebode Prevention of postnatal depression The Journal of the Royal Society for the Promotion of Health, September 1, 2005; 125(5): 221 - 226. [Abstract] [PDF]

				K. K. Hood, S. Bennett Johnson, S. K. Carmichael, L. M.B. Laffel, J.-X. She, and D. A. Schatz Depressive Symptoms in Mothers of Infants Identified as Genetically at Risk for Type 1 Diabetes Diabetes Care, August 1, 2005; 28(8): 1898 - 1903. [Abstract] [Full Text] [PDF]

				D. T. S. Lee, A. S. K. Yip, S. S. M. Chan, M. H. Y. Tsui, W. S. Wong, and T. K. H. Chung Postdelivery Screening for Postpartum Depression Psychosom Med, May 1, 2003; 65(3): 357 - 361. [Abstract] [Full Text] [PDF]

				S. L. KYE Pregnancy in Women With Eating Disorders Am J Psychiatry, July 1, 2002; 159(7): 1249 - 1250. [Full Text]

				Other articles noted Evid. Based Ment. Health, November 1, 2000; 3(4): 104 - 104. [Full Text]

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lee, D. T. S. Articles by Chung, T. K. H. Search for Related Content PubMed PubMed Citation Articles by Lee, D. T. S. Articles by Chung, T. K. H. Related Collections Depression Sexual Medicine: Female