Abstract
Objective: To describe possible correlations between incidence of postpartum depression and the following patient characteristics: age, breastfeeding status, tobacco use, marital status, history of depression, and method of delivery.
Study Design: Data gathered at routine 4-week postnatal visits were obtained from the patient records of 209 women who gave birth between June 1, 2001, and June 1, 2003, at three university medical clinics in Tulsa, Okla. Inclusion criteria required that the records of potential study subjects contain data on the characteristics noted as well as patient-completed Edinburgh Postnatal Depression Scale forms.
Results: Formula feeding in place of breastfeeding, a history of depression, and cigarette smoking were all significant risk factors for an Edinburgh Postnatal Depression Scale score of 13 or higher, indicating probable postpartum depression.
Conclusion: The authors' findings corroborate the results of previous investigators. To facilitate prophylactic patient education and intervention strategies, a larger study is recommended to determine risk factors for postpartum depression.
Postpartum depression (PPD) is a serious public health concern
1 that affects approximately 13% of women who give birth.
2 As routine screening for this condition has become more common in routine obstetric care,
3,4 data are now more readily available to assist researchers in identifying patient characteristics that may be associated with increased risk of PPD.
The self-administered, 10-question Edinburgh Postnatal Depression Scale (EPDS)
5 is an effective screening tool for PPD because it is reliable, easy for clinicians to score, and predictive of a clinical diagnosis of PPD.
3–8 In addition, other patient characteristics routinely recorded in patient records may be underutilized by physicians when screening patients for PPD: age, breastfeeding status, tobacco use, marital status, history of depression, and method of delivery.
Identification of clear correlations between certain risk factors and a diagnosis of PPD could lead to earlier intervention for these patients.
We sought to analyze patient records for the following patient characteristics at 4-weeks postnatal: EPDS numerical score, breastfeeding status, method of delivery (ie, vaginal birth vs cesarean section), history of depression, marital status, tobacco use, and patient age (< 21 years vs ≥21 years). With regard to dividing subjects into two groups by patient age, we used the same protocol used in a 1996 study by Chen.
9
Potential subjects for our study were women who gave birth between June 1, 2001, and June 1, 2003, at three sites in the Oklahoma State University (OSU) Physician clinic system in Tulsa. The urban population served by the OSU Physician clinic system is largely uninsured.
Our study was conducted prior to the implementation of universal screening for PPD in the OSU Physician clinic system. However, screening for PPD was often conducted at clinic sites during the standard 4-week postnatal visit.
Study inclusion was dependent on the presence of complete patient data for the characteristics under investigation as well as a patient-completed EPDS from the 4-weeks postnatal visit. In addition, to distinguish potential PPD from a previous diagnosis of major depression, the records of patients whose list of current medications included an antidepressant were excluded from study. Finally, any patients whose infants were stillborn or died before the 4-week postnatal follow-up were excluded from the study because we sought to examine risk factors that correlate with typical PPD, rather than the grief process that accompanies the loss of an infant.
Because other relevant and potentially interesting data, such as race and family history of depression, were not available in many patient records, these data were excluded from analysis in the present study.
The study protocol was developed in accordance with the Health Insurance Portability and Accountability Act guidelines enacted in 2003, and patient confidentiality was protected through all phases of this investigation. In addition, the protocol described was reviewed and approved by the institutional review board at the OSU Center for Health Sciences.
Patient data were systematically recorded by one investigator (S.B.M.S.) and a medical student for analysis on a spreadsheet (Microsoft Excel 2003; Microsoft Corp, Redmond, Wash).
After appropriate study subjects were identified, investigators masked the identity of the patients, concealing patient names and assigning a random number that was used as an identifier. The master list that contained the name-number conversions was destroyed at the project's close. All data provided in the present study are reported as mean scores of all study participants.
Patient characteristics were analyzed individually against EPDS scores with χ
2 tests. Possible patient scores for the 10-question EPDS can range from 0 to 30. Test sensitivity and specificity for the EPDS are reported at 86% and 78%, respectively.
5 Cox et al,
5 the instrument's developers, recommend that patients with EPDS scores higher than 12 receive a clinical evaluation for diagnosis of PPD.
Australian investigators in a 1993 study
8 (N=103) reported even better results for the sensitivity (100%) and specificity (95.7%) of the EPDS when using the developers' recommended 12- or 13-point cutoff for a diagnosis of probable PPD. Although the 1993 article by Boyce and colleagues
8 had a sample population from Australia and New Zealand, and their results, therefore, perhaps cannot be applied to the present study, we believe that the sensitivity and specificity values reported by the instrument's developers
5 may be conservative.
In light of the long-term reliable performance of the EPDS,
4,5,8,10,11 we decided to label subjects as either “depressed” or “not depressed” based on EPDS scores alone, with a score of 13 on the EPDS indicating probable PPD.
We are aware that several other instruments may be used by clinicians to diagnose PPD more definitively than the EPDS
10 (eg, Beck Depression Inventory,
12–14 Center for Epidemiologic Studies Depression Scale,
15 Zung's Self-rating Depression Scale,
16 Hamilton Rating Scale for Depression
17). Physicians in the OSU Physician clinic system use the EPDS for convenience (ie, it is reliable, easy to score, and highly predictive). Therefore, the EPDS provided the only standardized data related to PPD available in patient records for our retrospective investigation.
We calculated the relative risk for each statistically significant risk factor. Significant risk factors were then analyzed using log-linear models, or multidimensional χ2 tests. We determined that P values at the level of ≤.05 would be considered statistically significant. Small P values (ie, <.05) were considered to indicate a lack of additivity—or to indicate a lack of independence of these factors as they affect PPD. Conversely, nonsignificant P values indicated some degree of additivity.
During the study period, 1072 women delivered infants at the three study sites. Of these 1072 potential subjects, 217 (20%) women had patient records in which none of the variables noted was missing from their patient records.
Of the 217 women whose records contained a patient-completed EPDS, 7 (3.2%) were eliminated from the study because their lists of current medications included an antidepressant. One potential subject was excluded because her infant was stillborn.
Of the 209 women remaining in the study population after inclusion and exclusion criteria were applied, 128 (61%) had an EPDS score of 12 or below (ie, not depressed), and 81 (39%) had a score of 13 or higher (ie, depressed).
Among the 81 women whose EPDS results indicated a possible diagnosis of PPD, there was no significant difference by patient age or marital status at 4-weeks postnatal (Table).
Breastfeeding was associated with a significantly lower occurrence of PPD than formula feeding only (P<.001). The relative risk factor for formula feeding only was 2.04 (P<.05).
There was a significant difference in the occurrence of PPD between women who had a history of depression noted in their records and those without a history of depression (P=.003). The relative risk factor for women with a history of depression was 1.87 (P<.05).
Cigarette smoking was associated with a significantly higher occurrence of PPD than was not smoking (P=.01). The relative risk factor for cigarette smoking was 1.58 (P<.05).
Vaginal birth was not associated with a significantly lower occurrence of a positive screen for PPD than cesarean section (P=.09).
Three combinations of patient characteristics were found to have additive effects on the likelihood of subjects receiving an EPDS score that was predictive of PPD:
not breastfeeding and a history of depression (P=.12),
cigarette smoking and a history of depression (P=.29), and
cigarette smoking and not breastfeeding (P=.96).
To determine whether multiple significant risk factors enhanced clinicians' abilities to identify women at increased risk of PPD, we assessed additivity of patient characteristics.
As noted elsewhere, we identified three significant risk factors: formula feeding in place of breastfeeding, history of depression, and cigarette smoking. None of the three possible combinations of these risk factors were significantly nonadditive. In other words, all three possible combinations may be somewhat additive.
However, the association between increased incidence of PPD and the combination of formula feeding in place of breastfeeding and prior history of depression was weakest (ie, closest to significantly nonadditive). Therefore, based on the data reported, additivity between formula feeding only and prior history of depression is least convincing.
Formula feeding and cigarette smoking were the most additive, but cigarette smoking and history of depression were also additive. In other words, a patient with two of these risk factors is even more likely to suffer from PPD than ifshe had any one of these three risk factors alone. Confirmation of these relationships with further research could supply physicians with valuable information for educating and evaluating their patients for PPD.
It is our hope that researchers will assist clinicians in identifying the conditions and patient characteristics of postpartum women that are associated with an increased risk of PPD. Researchers may then develop reliable screening tests that predate the onset of postpartum dysphoria.
We recommend that physicians inquire about breast-feeding status, history of depression, and tobacco use among their postpartum patients. Three risk factors for PPD may serve as valuable “red flags” to assist physicians in diagnosing patients with PPD: formula feeding in place of breastfeeding, history of depression, and cigarette smoking. The presence of more than one of these risk factors in postpartum patients should serve as an even greater warning to physicians that these patients should be observed, evaluated for, and educated about PPD. As more women with PPD are diagnosed earlier as a result of such precautions, it is our hope that the suffering of both mother and infant can be reduced.
Editor's message: In the original print publication, the Yes and No column headings under “EPDS Score, ≥13” were accidentally reversed. The error has been corrected here.
Kumar RC. “Anybody's child”: severe disorders of mother-to-infant bonding. Br J Psychiatry. 1997;171:175 –181.
O'Hara MW, Swain AM. Rates and risk of postpartum depression—a metaanalysis. Int Rev Psychiatry. 1996;8:37 –54.
Georgiopoulos AM, Bryan TL, Wollan P, Yawn BP. Routine screening for postpartum depression [published correction appears in J Fam Pract. 2001;50:470]. J Fam Pract. 2001;50:117 –122.
Morris-Rush JK, Freda MC, Bernstein PS. Screening for postpartum depression in an inner-city population. Am J Obstet Gynecol. 2003;188:1217 –1219.
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 –786.
Fergerson SS, Jamieson DJ, Lindsay M. Diagnosing postpartum depression: can we do better? Am J Obstet Gynecol. 2002;186:899 –902.
Boyce P, Stubbs J, Todd A. The Edinburgh Postnatal Depression Scale: validation for an Australian sample. Aust N Z J Psychiatry. 1993;27:472 –476.
Chen CH. Postpartum depression among adolescent mothers and adult mothers. Kaohsiung J Med Sci. 1996;12:104 –113.
Beck CT. Predictors of postpartum depression: an update. Nurs Res. 2001;50:275 –285.
Georgiopoulos AM, Bryan TL, Yawn BP, Houston MS, Rummans TA, Therneau TM. Population-based screening for postpartum depression. Obstet Gynecol. 1999;93(5 Pt 1):653 –657.
Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:561 –571.
Beck AT, Rial WY, Rickels K. Short form of depression inventory: cross-validation. Psychol Rep. 1974;34:1184 –1186.
Beck AT, Steer RA. Internal consistencies of the original and revised Beck Depression Inventory. J Clin Psychol. 1984;40:1365 –1367.
Zung WW. A self-rating depression scale. Arch Gen Psychiatry. 1965;12:63 –70.
Hamilton M. A rating scale for depression. J Neurol Neursurg Psychiatry. 1960;23:56 –62.
Misri S, Sinclair D, Kuan A. Breast-feeding and postpartum depression: is there a relationship? Can J Psychiatry. 1997;42:1061 –1065.
Abou-Saleh MT, Ghubash R, Karim L, Krymski M, Bhai I. Hormonal aspects of postpartum depression. Psychoneuroendocrinology. 1998;23:465 –475.
Labbok MH. Effects of breastfeeding on the mother [review]. Pediatr Clin North Am. 2001;48:143 –158.
Pfost KS, Stevens MJ, Lum CU. The relationship of demographic variables, antepartum depression, and stress to postpartum depression. J Clin Psychol. 1990;46:588 –592.
Foster HW Jr, Bond T, Ivery DG, Treasure OA, Smith D, Sarma RP, et al. Threatened pregnancy: Environment and reproduction at risk. Teen pregnancy-problems and approaches: panel presentations. Am J Obstet Gynecol. 1999;181:S32 –S36.
Anda R, Williamson D, Jones D, Macera C, Eaker E, Glassman A, et al. Depressed affect, hopelessness, and the risk of ischemic heart disease in a cohort of US adults. Epidemiology. 1993;4:285 –294.
Cutrona CE, Troutman BR. Social support, infant temperament, and parenting self-efficacy: a meditational model of postpartum depression. Child Dev. 1986;57:1507 –1518.
O'Hara MW, Neunaber DJ, Zekoski EM. Prospective study of postpartum depression: prevalence, course, and predictive factors. J Abnorm Psychol. 1984;93:158 –171.
O'Hara MW, Rehm LP, Campbell SB. Predicting depressive symptomatology: cognitive-behavioral models and postpartum depression. J Abnorm Psychol. 1982;91:457 –461.
Chasnoff IJ, Neuman K, Thornton C, Callaghan MA. Screening for substance use in pregnancy: a practical approach for the primary care physician. Am J Obstet Gynecol. 2001;184:752 –758.
Bergant A, Nguyen T, Moser R, Ulmer H. Prevalence of depressive disorders in early puerperium [in German]. Gynakol Geburtshilfliche Rundsch. 1998;38:232 –237.
Jadresic E, Araya R. Prevalence of postpartum depression and associated factors in Santiago, Chile [in Spanish]. Rev Med Chil. 1995;123:694 –699.
Hobfoll SE, Ritter C, Lavin J, Hulsizer MR, Cameron RP. Depression prevalence and incidence among inner-city pregnant and postpartum women. J Consult Clin Psychol. 1995;63:445 –453.
Wing JK, Cooper JE, Sartorius N. Measurement and Classification of Psychiatric Symptoms: An Instruction Manual for the PSE and Catego Program. Cambridge, England: Cambridge University Press;1974 .
Pfost KS, Lum CU, Stevens MJ. Femininity and work plans protect women against postpartum dysphorias. Sex Roles. 1989;21:423 –431.
Whiffen VE. Vulnerability to postpartum depression: a prospective multivariate study. J Abnorm Psychol. 1988;97:467 –474.
Barnet B, Duggan AK, Wilson MD, Joffe A. Association between postpartum substance use and depressive symptoms, stress, and social support in adolescent mothers. Pediatrics. 1995;96(4 Pt 1):659 –666.
Fisher J, Astbury J, Smith A. Adverse psychological impact of operative obstetric interventions: a prospective longitudinal study. Aust N Z J Psychiatry. 1997;31:728 –738.