Dagher, Rada K. PhD, MPH; McGovern, Patricia M. PhD, MPH, RN; Dowd, Bryan E. PhD, MS; Gjerdingen, Dwenda K. MD, MS
Women who have just given birth often perceive a diminished quality of life1 and are at risk for psychiatric disorders such as postpartum depression, posttraumatic stress disorder, anxiety disorders, obsessions of child harm, and postpartum psychosis.2 Postpartum depression is the most common serious mental disorder after childbirth, affecting on average 13% of women3 or approximately 500,000 women per year in the United States.4 It is characterized by debilitating symptoms such as diminished interest/pleasure, depressed mood, fatigue, increased or decreased sleep and appetite, poor self-esteem, difficulty concentrating, psychomotor agitation, and thoughts of suicide.5 Postpartum depression often begins within 4 weeks after childbirth; symptoms can be severe and persistent.6 If untreated, it may last beyond the first postpartum year.7 The postpartum period poses a higher risk of new onset of major depressive disorder for women than do nonpostpartum periods8,9 and may increase health services utilization and health care expenditures.
Given the potential serious outcomes of postpartum depression and the high labor force participation rate among mothers of infants in the United States (56% in 201010), coupled with a relatively quick return to work after childbirth,11 the issue of postpartum depression is important for expectant mothers, their families, and employers. Generally, depression is a prevalent disease that causes substantial disability. The total US economic burden of depression in 2000 has been estimated to be $83.1 billion.12 The costs of direct depression treatment (This study did not examine the costs of treatment of postpartum depression in particular.) accounted for 31% of the total, 7% was attributable to depression-related suicide costs, and 62%, or $51.5 billion, was workplace costs, including wage-related costs of days missed from work and reduced productivity.12 In general, depression among US workers costs employers $44 billion per year in lost productivity and about $12.4 billion in health care expenditures.13,14
Given the important role of employers as health care purchasers15 and the rapidly increasing costs of health insurance,16 there is a need for more information about the cost of illnesses and effective treatments that could guide employers' decisions regarding choice of health plans and benefit packages. Such information would be particularly timely for small employers who may want to purchase insurance through the American Health Benefit Exchanges that will be created by the states under the Patient Protection and Affordable Care Act (H.R. 3590).
Few studies specifically examine the health services used by women experiencing postpartum depression and none have examined the associated health care costs. A survey of 574 women who delivered at an Australian hospital found that at 4 months postpartum, depressed women had a higher likelihood of visiting a psychiatrist, social worker, postnatal depression group, pediatrician, or a general practitioner than nondepressed women.17 However, the investigators only conducted bivariate analyses of these relationships. Another study of 594 women from British Columbia found that postpartum depressed women were more likely to visit their family physician at least once compared with nondepressed women; however, there were no differences between the two groups in utilization of clinic visits or the emergency department.18 Again, these bivariate analyses did not control for potential confounders. The general literature on nonpuerperal depression shows that depressed patients have higher rates of office visits, take more prescription drugs, and have higher hospitalization rates than nondepressed patients.19–21 Studies that control for preexisting medical conditions show that depressed patients utilize two to four times as much nonpsychiatric medical care as nondepressed patients.22–24 This study examines the associations between postpartum depression and health care expenditures 11 weeks after childbirth in a sample of employed women.
The theory guiding this study is Andersen's Behavioral Model of health services use.25 It suggests that health services utilization is a function of predisposing, enabling, and need factors. Predisposing variables represent the natural tendency of individuals to use health services and include demographic variables such as age and gender and social structural variables such as education, occupation, ethnicity, and social networks.25 Enabling components represent the resources (eg, income and insurance coverage) available to individuals for the use of services.25 Need variables describe health status or illness and are the most direct causes of health service use.25 In this study, we examine the association of the primary need variable—postpartum depression at 5 weeks after childbirth—with health care costs 11 weeks postpartum, independent of the effects of predisposing factors (age, marital status, education, race, and parity); enabling factors (poverty level, health insurance, leave status, and social support); and other need factors (number of labor and delivery complications, number of chronic health conditions, delivery type, prenatal moods, and physical health and maternal symptoms at 5 weeks postpartum).
We used data from a prospective cohort study of employed mothers, the Maternal Postpartum Health Study.26 Women, aged 18 years and older, were recruited from three community hospitals in Minneapolis and St. Paul, MN, while hospitalized for childbirth in 2001. Vital statistics revealed that recruited mothers were comparable on demographics and birth characteristics to mothers who delivered at 41 other hospitals in the seven-county metropolitan area. Eligibility criteria were that women spoke English, resided in the seven-county area, had been employed for at least 20 hours per week, planned to continue employment after childbirth, and had a healthy singleton infant. Institutional review boards at the participating hospitals and the University of Minnesota approved this study. Of an eligible population of 1157, 817 women were interviewed by perinatal nurses while in the hospital for childbirth (71% response rate). Statistical comparisons between participants and nonparticipants showed no differences in infant birth weight, gestation duration, maternal age, marital status, and duration of employment for primary employer.
Enrolled subjects were interviewed by telephone at 5 weeks and 11 weeks after delivery, with 88% (n = 716) and 81% (n = 661) response rates, respectively. We utilized data from the 638 women who completed the questionnaires at enrollment, 5 weeks, and 11 weeks postpartum. The time periods related to clinical changes in mothers' postpartum health and return-to-work patterns. New mothers typically visited their health care provider between 4 and 6 weeks after childbirth, and an earlier study in Minnesota revealed that most women returned to work at approximately 5 to 6 weeks postpartum if they lacked paid-leave benefits and 10 to 11 weeks postpartum if they had them.27
The dependent variable is expenditure on health care services from discharge until 11 weeks postpartum. We computed this variable by first collecting self-reported data on the use of health care services, including outpatient surgeries/procedures, hospitalizations, doctor office/urgent care visits, mental health counseling visits, and emergency department care. The recall period for each service was approximately 3 months (ie, since childbirth), consistent with periods identified as most accurate for recall.28 We used the unit price of services in 2001 from claims data for women of reproductive years provided by Blue Cross Blue Shield of Minnesota and self-reported utilization to create a price-weighted volume of services for each subject. Because the distribution of health services expenditures was skewed, we normalized the errors in our regression by taking the natural log of the observed expenditures plus $1 to account for the small number of women (4.1%) who had no expenditures during the observation period.
Postpartum depression, the primary explanatory variable, was measured at 5 weeks postpartum using the Edinburgh Postnatal Depression Scale (EPDS).29 The EPDS has satisfactory validity in identifying mothers with postpartum depression, where a threshold score of 12 to 13 identified women with definite major depressive illness.29 In this study, women who scored 13 or higher on the EPDS were coded as depressed. To verify the convergent validity of the categorical depression score, t tests showed that depressed women had significantly lower mental health scores on the mental component summary scale30 (37.8 vs 50.1; P < 0.001) than nondepressed women.
Physical and mental health were measured at 5 weeks after childbirth by using the 12-Item Short-Form Health Survey (version 2).30 This is a 12-item measure of general health, which includes a physical component summary and a mental component summary. When participants' 5-week scores were compared with national norms for women 25 to 34 years of age,30 our participants scored slightly worse on physical health (mean: 51.4 vs 52.7 [SD: 7.2, 9.13]; z = −3.9; P < 0.001) and slightly better on mental health (mean: 49.6 vs 47.2 [SD: 7.9, 12.1]; z = 4.9; P < 0.001).
Maternal symptoms were measured at 5 weeks after childbirth using a physical problem checklist adapted from Gjerdingen et al.31 It asks women to report the presence of 28 symptoms frequently experienced during the postpartum period. These include acne, rash, hair loss, headaches, runny or stuffy nose, sore throat, cough or cold, bronchitis or pneumonia, asthma, sinus trouble, hypertension, irregular heartbeats, decreased appetite, diarrhea or stomach flu, abdominal pain, hemorrhoids, constipation, back or neck pain, numbness, breast discomfort, nipple irritation, breast infection or mastitis, decreased sexual desire, uterine infection, fever greater than 100°F, hot flashes, fatigue, and dizziness. Maternal symptoms are scored as a summation of symptoms experienced in the preceding 4 weeks.
Labor and delivery complications were abstracted from medical records that had information about the presence or absence of anesthetic complications, excessive bleeding, lacerations, seizures, eclampsia, abruptio placenta, or infections. This variable was scored as the number of complications experienced.
Chronic health problems were abstracted from medical records that had information about the presence or absence of hypertension, diabetes, cardiac disease, renal disease, and asthma. This variable was scored as the number of chronic health problems experienced.
Health insurance coverage was measured at 5 weeks after childbirth by asking about the woman's primary source of health insurance. Response categories were coded as two dummy variables: (1) insurance from own employer/family member and (2) public assistance. The reference category was “purchased plan/currently uninsured.”
The measure for social support was taken from Sherbourne and Stewart.32 It is a five-item scale that assesses how often various kinds of functional support are available for the woman if she needs them, including positive social interaction, emotional, informational, affectionate, and tangible support. Response options varied from one (none of the time) to five (all of the time) and were summed (scoring of 5–25).
Leave status (1 = working; 0 = on leave) was assessed at 5 weeks postpartum. Since leave status is an endogenous variable, we estimated the model with two-stage least squares (2SLS) analyses, using the maximum available duration of all paid leave according to employer policy (vacation, sick, maternity, or disability leave) as an instrument for the woman's leave status. The F statistic was 11.41 (P = 0.0008) for the significance test predicting leave status with the instrument.
Age, race, education, marital status, parity status, delivery type, and poverty status were collected in person at enrollment. The poverty status measure was constructed by investigators using income guidelines of the Minnesota At-Home Infant Child Care Program.33 The prenatal mood disturbances measure was taken from McGovern et al34: “During this pregnancy did you ever have a problem with your mood, such as feeling depressed or anxious?” and was assessed at enrollment.
The women averaged 30 years of age; 87% were white, 76% were married, 49% were college educated, and 99.4% had health insurance. Ninety percent of the women had employment-based health insurance, 7.5% had insurance through public assistance, and 1.6% purchased their own plan. Twenty-eight women (4.1%) incurred zero health care expenditures; of these women, only one met the threshold for depression. Table 1 compares depressed and nondepressed women by personal and delivery characteristics. Depressed women were more likely to be single, have no college education, meet the poverty threshold, experience prenatal moods, be back to work at 5 weeks postpartum, have lower social support, and have more maternal symptoms than nondepressed women. Chi-square comparisons between participants who completed the questionnaire at 11 weeks and those who did not showed that nonparticipants were significantly more likely to meet the depression threshold on the EPDS (11.5%; n = 11) at 5 weeks postpartum than participants (4.9%; n = 31).
Five percent (n = 31) of the women met the depression threshold on the EPDS at 5 weeks postpartum. Among these 31 women, 96.8% (n = 30) had doctor office visits, 19.4% (n = 6) went to the emergency department (reasons include uterine infection, mastitis, uterine fibroid, and hemorrhoids), one was admitted to the hospital (for mastitis), and one had outpatient surgery (fibroid tumor). Only 7 of these depressed women (22.6%) had mental health counseling visits despite 93.5% (n = 29) having health insurance.
In two-stage least squares analyses, variables that significantly predicted health services expenditures 11 weeks after discharge were: older age, poverty, nonpublic assistance insurance status, more maternal symptoms, and postpartum depression (Table 2). In these analyses, the dependent variable is in its log-transformed state; thus, coefficients on continuous explanatory variables give the percentage change in expenditures as the explanatory variable changes one unit, while exp(b) −1 gives the percentage change for discrete variables. Controlling for demographics, health status, and other characteristics, depressed women incurred 90% more health services expenditures than nondepressed women. To explain this finding further, we conducted bivariate analyses, comparing the frequency and average number of health services used by depression status. Depressed (vs nondepressed) women were significantly more likely to attend the emergency department (19.4% vs 4.4%) and to seek mental health counseling (22.6% vs 3.8%); the two groups did not differ on hospitalizations, outpatient surgeries, or office visits (Table 3). Moreover, Mann-Whitney tests (Mann-Whitney tests are used instead of t tests because they are non-parametric tests that accommodate the skewed nature of the expenditures data.) showed that depressed women had significantly higher counseling and emergency department expenditures, as well as total expenditures than nondepressed women (Table 4).
Little information exists on the prevalence and health care costs of postpartum depression in the workforce. Here, we found a 5% point prevalence of depression at 5 weeks postpartum, comparable with the 5.3% 2-week prevalence rate of major depression among working women in the US workforce.14 Depressed women incurred considerably higher health care expenditures than nondepressed women. This finding is consistent with the increased health care utilization seen among people with general (nonpostpartum) depression21,23,24 and supports limited previous research findings on postpartum depression and health care services utilization.17,18
Only 7 of the 31 depressed women (23%) had seen a mental health professional; yet this represented a six-fold (95% confidence interval, 2.77–12.61) increase in mental health counseling visits compared with nondepressed women. This finding, together with the four-fold (95% confidence interval, 1.94–9.98) increase in emergency department visits, appeared to be driving the increased costs among depressed women. Applying Andersen's model, postpartum depression is a significant need factor associated with health care utilization and expenditures. Other variables that significantly predicted higher health care expenditures at 11 weeks postpartum were older age, low income, purchased plan or uninsured (vs public assistance), and maternal symptoms. With every additional symptom, there was a 6.1% increase in health care costs.
The health care costs of postpartum depression have policy implications in terms of the importance of addressing mental health issues in the workplace. Under the Patient Protection and Affordable Care Act, employers have the option of purchasing insurance from the state-based insurance exchanges, which offer different tiers of benefit packages, all of which must have a base-level package that includes mental health coverage (H.R. 3590). Employers with a high proportion of female employees of reproductive age may want to choose the plans that have more generous coverage of mental health services. Wang et al35 conducted a cost-effectiveness analysis, which showed that introducing a program of enhanced depression care in the workplace resulted in significant gains in quality-adjusted life years for workers at a reasonable cost in comparison with usual care. The incremental cost-effectiveness ratio was $19,976 per quality-adjusted life years, which falls within the cost range of interventions covered by employer-sponsored plans.
Our study had limitations pertaining to participants, measures, and methods. Results can mainly be generalized to employed women with similar demographic (87% white, 76% married, and 49% college educated) characteristics as those studied. However, the Institute for Women's Policy Research reports that national estimates of women's occupation, education, and earnings reveal similarities between Minnesota women and women nationwide.36 For example, the proportion of employed Minnesota women in managerial and professional occupations at 34.2% is comparable with that for the nation at 33.2%.36 Similarly, median annual earnings of $31,900 per year for Minnesota women are comparable with women's median annual earnings at the national level ($30,100).36 In this study, postpartum depression was measured by self-report; thus, interpretation of the findings should be specific to postpartum depressive symptoms. Moreover, the relatively low frequency of labor and delivery complications and chronic health problems in this employed sample and our measure of frequency data may explain why these two variables were marginally rather than statistically significant. Despite the high retention rate of the study, women who were lost to follow up at 11 weeks postpartum were more likely to be depressed than study participants. Thus, the association between postpartum depression and health services expenditures may have been underestimated. Finally, the cross-sectional nature of this study prohibited causal inferences.
Depressed women had considerably higher health care expenditures than nondepressed women. The disproportionate increase in counseling and emergency department visits may be driving the increased costs among depressed women. Additional research should compare estimates of self-reported depression symptoms with medical record data to identify the variation across postpartum depression estimates. Moreover, researchers should explore whether earlier identification and treatment of postpartum depression may decrease health care expenditures after childbirth.
This study was not supported by any of the following institutions: National Institutes of Health, Wellcome Trust, Howard Hughes Medical Institute, or other institutions. This research was supported by grant No. 5 R18 OH003605-05 from the National Institute for Occupational Safety and Health (NIOSH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIOSH. The authors have no conflicts of interest. This article was presented at The American Public Health Association 135th Annual Meeting. Washington, DC, November 3–7, 2007.
1. Gjerdingen DK, Center B. First-time parents' prenatal to postpartum changes in health. J Am Board Fam Pract. 2003;16:304–311.
2. Brockington I. Postpartum psychiatric disorders. Lancet. 2004;363:303–310.
3. O'Hara MW, Swain AM. Rates and risk of postpartum depression: a meta-analysis. Int Rev Psychiatry. 1996;8:37–54.
4. Wisner KL, Parry BL, Piontek CM. Postpartum depression. N Engl J Med. 2002;347:194–199.
5. First MG, Spitzer RL, Gibbon M, Williams JB. Structured Clinical Interview for DSM-IV Axis I Disorder—Clinical Version Administration Booklet. New York, NY: Biometrics Research Department; 1997.
6. Cooper P, Murray L. Fortnightly review: postnatal depression. Br Med J. 1998;316:1884–1886.
7. Gale S, Harlow BL. Postpartum mood disorders: a review of clinical and epidemiological factors. J Psychosom Obstet Gynaecol. 2003;24:257–266.
8. Eberhard-Gran M, Tambs K, Opjordsmoen S, et al. A comparison of anxiety and depressive symptomatology in mothers. Soc Psychiatry Psychiatr Epidemiol. 2003;38:551–556.
9. Vesga-López O, Blanco C, Keyes K, et al. Psychiatric disorders in pregnant and postpartum women in the United States. Arch Gen Psychiatry. 2008;65:805–815.
11. Johnson JO, Downs B. Maternity leave and employment patterns: 1961–2000. Current Population Report. Washington, DC: US Census Bureau; 2003:70–103.
12. Greenberg P, Kessler R, Birnbaum H, et al. The economic burden of depression in the United States: how did it change between 1990 and 2000? J Clin Psychiatry. 2003;64:1465–1475.
13. Pignone MP, Gaynes BN, Rushton JL, et al. Screening for depression in adults: a summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 2002;136:765–776.
14. Stewart WF, Ricci JA, Chee E, et al. Cost of lost productive work time among US workers with depression. JAMA. 2003;289:3135–3144.
15. Rosenbaum S. Negotiating the new health system: purchasing publicly accountable managed care. Am J Prev Med. 1998;14:67–71.
16. Baicker K, Chandra A. The labor market effects of rising health insurance premiums. J Labor Econ. 2006;24:609–634.
17. Webster J, Pritchard M, Linnane J, et al. Postnatal depression: use of health services and satisfaction with health-care providers. J Qual Clin Pract. 2001;21:144–148.
18. Dennis C-L. Influence of depressive symptomatology on maternal health services utilization and general health. Arch Womens Ment Health. 2004;7:183–191.
19. Brody D, Thompson T, Larson D, et al. Recognizing and managing depression in primary care. Gen Hosp Psychiatry. 1995;17:93–107.
20. Katon W, Berg A, Robins A, et al. Depression—medical utilization and somatization. West J Med. 1986;144:564–568.
21. Simon G, Vonkorff M, Barlow W. Health care costs of primary care patients with recognized depression. Arch Gen Psychiatry. 1995;52:850–856.
22. Callahan C, Hui S, Nienaber N, et al. Longitudinal study of depression and health services use among elderly primary care patients. J Am Geriatr Soc. 1994;42:833–838.
23. Johnson J, Weissman M, Klerman G. Service utilization and social morbidity associated with depressive symptoms in the community. JAMA. 1992;267:1478–1483.
24. Manning W Jr, Wells K. The effects of psychological distress and psychological well-being on use of medical services. Med Care. 1992;30:541–553.
25. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36:1–10.
26. McGovern P, Dowd B, Gjerdingen D, et al. Postpartum health of employed mothers 5 weeks after childbirth. Ann Fam Med. 2006;4:159–167.
27. McGovern P, Dowd B, Gjerdingen D, et al. The determinants of time off work after childbirth. J Health Polit Policy Law. 2000;25:527–564.
28. Bhandari A, Wagner T. Self-reported utilization of health care services: improving measurement accuracy. Med Care Res Rev. 2006;63:217–235.
29. 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.
30. Ware J, Kosinski M, Turner-Bowker D, Gandek B. Version 2 of the SF-12 Health Survey. Lincoln, RI: QualityMetric Inc; 2002.
31. Gjerdingen DK, Froberg DG, Chaloner KM, et al. Changes in women's physical health during the first postpartum year. Arch Fam Med. 1993;2:277–283.
32. Sherbourne CD, Stewart AL. The MOS Social Support Survey. Soc Sci Med. 1991;32:705–714.
34. McGovern P, Dowd B, Gjerdingen D, et al. Time off work and the postpartum health of employed women. Med Care. 1997;35:507–521.
35. Wang PS, Patrick A, Avorn J, et al. The costs and benefits of enhanced depression care to employers. Arch Gen Psychiatry. 2006;63:1345–1353.
36. Institute for Women's Policy Research. Women's Economic Status in the States: Wide Disparities by Race, Ethnicity and Region. Washington, DC: Institute for Women's Policy Research; 2004.
©2012The American College of Occupational and Environmental Medicine