Racial Disparities in Diagnosis of Postpartum Mood and Anxiety Disorders Among Symptomatic Medicaid Enrollees, 2012–2015
Abstract
Objective:
This study quantified the prevalence of postpartum mood and anxiety disorder (PMAD) diagnoses among symptomatic Michigan Medicaid enrollees and explored factors associated with receiving a diagnosis.
Methods:
Data sources comprised Michigan Medicaid administrative claims and Phase 7 Michigan Pregnancy Risk Assessment Monitoring System (MI-PRAMS) survey responses, linked at the individual level. Participants were continuously enrolled in Michigan Medicaid, delivered a live birth (2012–2015), responded to the survey, and screened positive for PMAD symptoms on the adapted two-item Patient Health Questionnaire. Unadjusted and adjusted weighted logistic regression analyses were used to predict the likelihood of having a PMAD diagnosis (for the overall sample and stratified by race).
Results:
The weighted analytic cohort represented 24,353 deliveries across the 4-year study. Only 19.8% of respondents with symptoms of PMAD had a PMAD diagnosis between delivery and 3 months afterward. Black respondents were less likely to have PMAD diagnoses (adjusted odds ratio [AOR]=0.23, 95% CI=0.11–0.49) compared with White respondents. Among White respondents, no covariates were significantly associated with having a diagnosis. However, among Black respondents, more comorbid conditions and more life stressors were statistically significantly associated with having a diagnosis (AOR=3.18, 95% CI=1.27–7.96 and AOR=3.12, 95% CI=1.10–8.88, respectively).
Conclusions:
Rate of PMAD diagnosis receipt differed by race and was low overall. Black respondents were less likely than White respondents to receive a diagnosis. Patient characteristics influencing diagnosis receipt also differed by race, indicating that strategies to improve detection of these disorders require a tailored approach.
HIGHLIGHTS
•
Only about 20% of patients with symptoms of postpartum mood and anxiety disorders (PMADs) had a contemporaneous PMAD diagnosis.
•
The probability of having a PMAD diagnosis was three times higher for White patients than for Black patients with PMAD symptoms, even after the analyses were controlled for other factors.
•
Racial disparities in PMAD diagnosis receipt and documentation require tailored interventions.
Maternal mental disorders are the most common pregnancy-related complication, affecting roughly one in five perinatal patients (1, 2). Postpartum mood and anxiety disorders (PMADs) are the most common maternal mental disorders. They are associated with missed routine health services, increased emergency health service use, and maternal morbidity and mortality (3–5).
Despite their high prevalence and disease burden, PMADs often remain underdiagnosed. Low diagnosis rates contribute to even lower treatment rates and exacerbate poor outcomes. Untreated PMADs are associated with worse maternal and infant health outcomes (6, 7), and the societal cost of untreated PMADs is estimated at $14 billion annually in the United States alone (8).
The association between race and PMAD diagnosis receipt is complex and may be confounded by other socioeconomic factors and measurement biases. Low socioeconomic status, indicated by factors such as low levels of education and income, has been associated with increased risk for PMADs (9, 10). Although some studies (9, 11–13) have found in unadjusted analyses that people of color are at an increased risk for PMADs, these associations often became nonsignificant when the analyses were controlled for socioeconomic factors. In addition, at least one study (10) found that Black and Hispanic women had lower rates of postpartum depression, even after other factors were considered. Differences in PMAD diagnosis rate may be further confounded by racial differences in illness severity (13).
People of color face additional barriers to accessing health care and experience lower treatment initiation than White patients (14) but increased use of hospital-based care (e.g., inpatient hospitalizations) for postpartum depression (15). These treatment disparities may be partly caused by racial differences in access to care, clinical experiences, and care preferences (16); however, the potential influence of these factors on PMAD diagnosis rates is unknown.
The Michigan Pregnancy Risk Assessment Monitoring System (MI-PRAMS) is an ongoing, representative survey of childbearing people, conducted at 2–6 months after childbirth by the Michigan Department of Health and Human Services (MDHHS) in conjunction with the Centers for Disease Control and Prevention (17, 18). The survey asks respondents about factors and experiences that may affect mothers and infants, including the presence of PMAD symptoms in the early postpartum period.
Michigan Medicaid insures >40% of births in the state of Michigan (19, 20). Medicaid recipients are more likely to have lower socioeconomic status, to experience preexisting health conditions, and to have worse birth outcomes than are privately insured individuals (21–24). MDHHS manages Michigan Medicaid’s administrative claims data, which document patients’ clinical diagnoses, including PMAD.
This study created a unique linkage between the MI-PRAMS survey and Michigan Medicaid claims data to assess the relationship between self-reported PMAD symptoms and clinical PMAD diagnoses among patients insured by Michigan Medicaid and who gave birth between 2012 and 2015. This study quantified the PMAD diagnosis rate, identified characteristics associated with receiving a diagnosis, and explored whether these characteristics differed by race. We hypothesized that Black Medicaid enrollees with PMAD symptoms would have a lower PMAD diagnosis rate than would White Medicaid enrollees with PMAD symptoms because of systemic differences in access to and quality of care.
Methods
Design
This retrospective, cross-sectional, observational study characterized the relationship between self-reported PMAD symptoms and a clinical PMAD diagnosis. We estimated the PMAD diagnosis rate among Michigan Medicaid enrollees experiencing childbirth (2012–2015) and self-reported PMAD symptoms. We also identified patient characteristics associated with receiving a PMAD diagnosis (overall and stratified by race).
Data Sources
This study used secondary data from the MI-PRAMS survey and from Michigan Medicaid claims. MDHHS routinely collects and manages the MI-PRAMS survey and Michigan Medicaid claims as part of its routine operations. Institutional review boards at the University of Michigan Medical School (HUM00148854) and at MDHHS (201811-10-EA) approved this research.
Sample
The study sample was drawn from MI-PRAMS survey respondents who gave birth during 2012–2015 and were enrolled in Michigan Medicaid at the time of delivery. To ensure all perinatal health care interactions were documented in Medicaid claims, we included only respondents with continuous Medicaid enrollment from 9 months before delivery through 3 months after childbirth (N=2,781). In addition, we included only enrollees who self-reported PMAD symptoms on the two-item Patient Health Questionnaire (PHQ-2) embedded within the MI-PRAMS survey, for a total unweighted sample of 542. When weighted, this sample represents 24,353 enrollees.
Measurements
We defined the presence of self-reported PMAD symptoms as a positive result on the PHQ-2, a mental health screening tool used in clinical and public health settings (25–27). The MI-PRAMS survey uses adapted PHQ-2 response options of “always,” “often,” “sometimes,” “rarely,” and “never.” We defined a positive screening result as a response of “often” or “always” to either PHQ-2 item, consistent with prior use of this survey item. A positive screening result indicates the presence of PMAD symptoms but does not necessarily reflect a clinical diagnosis (28–30). A systematic review (31) of postpartum depression screening tools found that the PHQ-2 has an average sensitivity of 79% (95% CI=62%–100%) and an average specificity of 83% (95% CI=79%–91%).
Our independent variable was receipt of a PMAD diagnosis, which we studied overall and stratified by race. We defined PMAD diagnosis as the presence of at least one inpatient or two outpatient Medicaid claims that included ICD-9 or ICD-10 codes for anxiety or depression between the delivery and 3 months after childbirth. Medicaid claims document clinicians’ diagnoses but do not contain information on how clinicians diagnose conditions. Diagnostic protocols may vary by clinician. We curated anxiety and depression ICD-9 and ICD-10 codes from the Healthcare Cost and Utilization Project (32). This list of PMAD ICD-9 and ICD-10 codes is available in the online supplement to this article.
Covariates included maternal age, race (White, Black, or other/unknown), Obstetric Comorbidity Index (OBCMI) (33) score (0 or 1 vs. ≥2, where possible scores range from 0 to 45 and higher scores indicate more comorbid conditions), premature birth (defined as <37 weeks of gestation), cesarean birth, satisfaction with prenatal care (satisfied vs. not satisfied), satisfaction with delivery (satisfied vs. not satisfied), receipt of postpartum checkup (yes vs. no), and number of major life stressors during the year before delivery (0–3 vs. ≥4).
The OBCMI is a validated composite measure of antepartum health factors, such as diabetes or hypertension, that is used to identify childbearing patients at high risk for severe morbidity (33). We measured life stressors by using responses to a series of MI-PRAMS survey questions asking patients to affirm whether they had experienced any of 14 life stressors during the year before delivery. Potential stressors included illness of a family member, divorce or separation, moving, homelessness, job loss, reduced work hours, work-related separation from partner, arguing with partner, partner not wanting the pregnancy, difficulty paying bills, partner or self in jail, substance use by someone close, or death of someone close. We dichotomized the number of stressors at the median (0–3 vs. ≥4 stressors).
Analysis
We weighted the analytic sample according to the MI-PRAMS weighting strategy (34). The unit of analysis was an individual delivery.
Univariate analyses included measures of central tendency, with weighted means for continuous variables and weighted frequencies for categorical variables. We examined differences between Black and White patients by using Student t tests for continuous variables and Pearson chi-square tests for categorical covariates.
We calculated unadjusted and adjusted weighted logistic regression models to predict receipt of a PMAD diagnosis. We performed these models with the analyses controlled for race and stratified by race. Stratified subgroup analyses included only White and Black subgroups, because other racial-ethnic categories had too few individuals for meaningful interpretation. We performed data management and logistic regression in R, version 4.0.3, by using the “svyglm” function (binomial family with link=“logit”) from the “survey” package, and we calculated predicted probabilities in Stata, version 17, by using the “margins” command.
Results
In total, 7,257 respondents completed MI-PRAMS surveys from 2012 through 2015, representing annual, unweighted response rates ranging from 49.7% to 52.1% (35–38). Michigan Medicaid insured 3,909 (53.9%) of these respondents at the point of delivery, and 2,781 (71.1%) of those respondents were continuously enrolled in Medicaid from 9 months before delivery through 3 months after childbirth. Of these respondents, 542 (19.5%) had positive results on PMAD symptom screening. After the analyses were weighted according to MI-PRAMS weighting methodology, these 542 respondents represented 24,353 deliveries across the 4-year study. The study cohort diagram is available in the online supplement .
Table 1 contains cohort characteristics overall and stratified by race. Among all respondents with PMAD symptoms, only 19.8% of delivery records included a PMAD diagnosis code between delivery and 3 months after childbirth. A greater percentage of White respondents had a PMAD diagnosis compared with Black respondents (23.8% vs. 8.9%, p<0.001), and a smaller percentage of White respondents had a premature delivery compared with Black respondents (11.1% vs. 21.2%, p=0.015). Other covariates, including age, dichotomized OBCMI score, cesarean birth, satisfaction with prenatal care, satisfaction with delivery, receipt of postpartum checkup, and dichotomized number of life stressors, did not vary by race.
| Characteristic | All (N=24,353 deliveries) | White (N=13,166 respondents) | Black (N=8,069 respondents) | | |||
|---|---|---|---|---|---|---|---|
| N | % | N | % | N | % | p | |
| Diagnosis receipt | | | | | | | <.001 |
| Yes | 4,814 | 19.8 | 3,135 | 23.8 | 718 | 8.9 | |
| No | 19,539 | 80.2 | 10,031 | 76.2 | 7,351 | 91.1 | |
| Age (mean±SD years) | 25.3±5.2 | | 25.0±4.5 | | 25.4±5.8 | | .452 |
| Race | | | | | | | <.001 |
| White | 13,166 | 54.1 | 13,166 | 100.0 | 0 | — | |
| Black | 8,069 | 33.1 | 0 | — | 8,069 | 100.0 | |
| Other/unknown | 3,117 | 12.8 | 0 | — | 0 | — | |
| Obstetric Comorbidity Index score | | | | | | | .567 |
| 0 or 1 | 16,360 | 67.2 | 8,964 | 68.1 | 5,224 | 64.7 | |
| ≥2 | 7,993 | 32.8 | 4,202 | 31.9 | 2,845 | 35.3 | |
| Gestational age | | | | | | | .015 |
| Full term | 21,118 | 86.7 | 11,702 | 88.9 | 6,360 | 78.8 | |
| Premature | 3,236 | 13.3 | 1,465 | 11.1 | 1,709 | 21.2 | |
| Delivery | | | | | | | .467 |
| Vaginal | 16,221 | 66.6 | 8,486 | 64.4 | 5,539 | 68.6 | |
| Cesarean | 8,132 | 33.4 | 4,681 | 35.6 | 2,530 | 31.4 | |
| Satisfaction with prenatal care | | | | | | | .118 |
| Satisfied | 19,411 | 80.0 | 9,878 | 75.0 | 6,608 | 82.8 | |
| Not satisfied | 4,850 | 20.0 | 3,288 | 25.0 | 1,370 | 17.2 | |
| Satisfaction with delivery | | | | | | | .075 |
| Satisfied | 17,259 | 71.3 | 8,538 | 65.0 | 5,987 | 75.1 | |
| Not satisfied | 6,963 | 28.7 | 4,595 | 35.0 | 1,984 | 24.9 | |
| Postpartum checkup | | | | | | | .103 |
| Received | 19,045 | 78.5 | 10,574 | 80.3 | 5,660 | 71.1 | |
| Not received | 5,202 | 21.5 | 2,593 | 19.7 | 2,304 | 28.9 | |
| N of stressors | | | | | | | .305 |
| 0–3 | 12,308 | 51.8 | 6,257 | 48.1 | 4,186 | 54.5 | |
| ≥4 | 11,441 | 48.2 | 6,760 | 51.9 | 3,492 | 45.5 | |
a
Unweighted total sample N=542. All Ns are weighted and were calculated by using the Michigan Pregnancy Risk Assessment Monitoring System weighting strategy (34). Weighted Ns include fractions of people and have been rounded to the nearest whole person; rounding may cause columns to exceed the total N. Not every subcategory sums to the overall total due to missing responses; percentages were calculated on the basis of subcategory totals.
Table 2 contains odds ratios for the unadjusted and fully adjusted models for the overall weighted cohort. In unadjusted logistic regression analyses among the overall analytic sample, Black respondents were less likely to have a PMAD diagnosis than were White respondents (OR=0.31, 95% CI=0.16–0.60, p<0.001). In adjusted logistic regression analyses, Black respondents were still less likely to have a PMAD diagnosis than White respondents (adjusted OR=0.23, 95% CI=0.11–0.49, p<0.001). No other covariates were associated with receipt of a PMAD diagnosis in unadjusted or adjusted analyses in the overall analytic sample.
| Variable | Unadjusted | Adjusted | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | p | Adjusted OR | 95% CI | p | |
| Age | 1.03 | .97–1.10 | .301 | 1.02 | .94–1.09 | .659 |
| Black race (reference: White) | .31 | .16–.60 | <.001 | .23 | .11–.49 | <.001 |
| Other/unknown race (reference: White) | 1.42 | .48–4.21 | .521 | 1.68 | .55–5.17 | .362 |
| Obstetric Comorbidity Index score ≥2 (reference: 0 or 1) | 1.77 | .83–3.75 | .136 | 1.63 | .69–3.81 | .262 |
| Premature birth (reference: full-term birth) | 1.26 | .51–3.07 | .617 | 1.79 | .66–4.84 | .252 |
| Cesarean birth (reference: vaginal birth) | 1.35 | .63–2.89 | .438 | 1.23 | .55–2.77 | .616 |
| Not satisfied with prenatal care (reference: satisfied) | 2.01 | .85–4.74 | .112 | 2.38 | .88–6.41 | .086 |
| Not satisfied with delivery (reference: satisfied) | .68 | .28–1.64 | .388 | .44 | .18–1.10 | .081 |
| No postpartum checkup (reference: checkup) | 1.30 | .54–3.13 | .558 | 1.53 | .59–3.99 | .385 |
| ≥4 stressors (reference: 0–3) | 1.13 | .53–2.40 | .755 | .99 | .44–2.25 | .980 |
Table 3 shows odds ratios for the unadjusted and adjusted models, stratified by race (weighted cohort). No variables were significantly associated with receipt of a PMAD diagnosis among White respondents in unadjusted or adjusted models. Among Black respondents, older age (OR=1.08, 95% CI=1.00–1.15, p=0.036), higher OBCMI score (OR=4.55, 95% CI=1.94–10.68, p=0.001), and greater number of life stressors (OR=3.20, 95% CI=1.34–7.66, p=0.009) were statistically significantly associated with receipt of a PMAD diagnosis in unadjusted models. Higher OBCMI score (adjusted OR=3.18, 95% CI=1.27–7.96, p=0.013) and greater number of life stressors (adjusted OR=3.12, 95% CI=1.10–8.88, p=0.033) remained statistically significantly associated with receipt of a PMAD diagnosis after the analyses were adjusted for all covariates. Figure 1 displays weighted adjusted odds ratios and 95% confidence intervals for characteristics associated with receipt of a PMAD diagnosis for the cohort overall, for Black patients only, and for White patients only.
| Variable | White (N=13,166 respondents) | Black (N=8,069 respondents) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unadjusted | Adjusted | Unadjusted | Adjusted | |||||||||
| OR | 95% CI | p | Adjusted OR | 95% CI | p | OR | 95% CI | p | Adjusted OR | 95% CI | p | |
| Age | 1.03 | .92–1.15 | .638 | 1.00 | .89–1.13 | .983 | 1.08 | 1.00–1.15 | .036 | 1.05 | .98–1.12 | .179 |
| Obstetric Comorbidity Index score ≥2 (reference: 0 or 1) | 1.33 | .46–3.84 | .599 | 1.41 | .43–4.59 | .563 | 4.55 | 1.94–10.68 | .001 | 3.18 | 1.27–7.96 | .013 |
| Premature birth (reference: full-term birth) | 1.91 | .52–7.09 | .328 | 2.08 | .47–9.21 | .332 | 1.81 | .78–4.21 | .166 | 1.82 | .74–4.48 | .192 |
| Cesarean birth (reference: vaginal birth) | .95 | .32–2.81 | .929 | .80 | .29–2.23 | .672 | 1.93 | .86–4.30 | .109 | 1.44 | .61–3.41 | .402 |
| Not satisfied with prenatal care (reference: satisfied) | 2.08 | .66–6.53 | .207 | 2.29 | .69–7.63 | .175 | 1.56 | .60–4.02 | .361 | 1.23 | .34–4.39 | .748 |
| Not satisfied with delivery (reference: satisfied) | .62 | .19–2.01 | .427 | .38 | .12–1.24 | .107 | 1.61 | .68–3.80 | .280 | 1.53 | .41–5.75 | .529 |
| No postpartum checkup (reference: checkup) | 2.17 | .64–7.33 | .208 | 2.07 | .60–7.19 | .249 | .53 | .20–1.43 | .210 | .52 | .16–1.67 | .271 |
| ≥4 stressors (reference: 0–3) | .63 | .23–1.75 | .372 | .49 | .16–1.48 | .202 | 3.20 | 1.34–7.66 | .009 | 3.12 | 1.10–8.88 | .033 |
a
There are no estimates for race in stratified models because race was the stratifying variable.
FIGURE 1. Adjusted odds of having a postpartum mood and anxiety disorder among symptomatic Medicaid-insured postpartum patients, for the overall weighted cohort and stratified by racea
aAdjusted for age (continuous), Obstetric Comorbidity Index (OBCMI) score (0 or 1 vs. ≥2), premature birth (vs. full term), cesarean birth (vs. vaginal), satisfaction with prenatal care (satisfied vs. not satisfied), satisfaction with delivery (satisfied vs. not satisfied), receipt of postpartum checkup (yes vs. no), number of life stressors in the year before delivery (0–3 vs. ≥4), and race (overall cohort only). Overall weighted cohort, N=24,353 deliveries; White respondents, N=13,166; Black respondents, N=8,069. AOR, adjusted odds ratio.
Figure 2 displays weighted adjusted predicted probabilities of having a PMAD diagnosis for the overall cohort, for Black patients only, and for White patients only. After the models were adjusted for all variables, the predicted probability of having a PMAD diagnosis was 23.8% for White respondents and 7.2% for Black respondents. Sensitivity analyses adjusted for additional measures collected during the MI-PRAMS survey. The additional measures we explored, consisting of enrollees’ smoking status, education, marital status, and experiences with racism, produced similar results.
FIGURE 2. Adjusted predicted probability of having a postpartum mood and anxiety disorder among symptomatic Medicaid-insured postpartum patients, for the overall weighted cohort and stratified by racea
aAdjusted for age (continuous), Obstetric Comorbidity Index (OBCMI) score (0 or 1 vs. ≥2), premature birth (vs. full term), cesarean birth (vs. vaginal), satisfaction with prenatal care (satisfied vs. not satisfied), satisfaction with delivery (satisfied vs. not satisfied), receipt of postpartum checkup (yes vs. no), number of life stressors in the year before delivery (0–3 vs. ≥4), and race (overall cohort only). Overall weighted cohort, N=24,353 deliveries; White respondents, N=13,166; Black respondents, N=8,069. Error bars represent the 95% confidence intervals for each predicted probability.
Discussion
In this study representing 4 years of deliveries covered by Michigan Medicaid, most enrollees experiencing PMAD symptoms did not receive a PMAD diagnosis. Diagnosis rates were especially low among Black patients experiencing PMAD symptoms.
Extent of Underdiagnosis
These results showed that only one in five respondents with PMAD symptoms received a PMAD diagnosis. Furthermore, the discrepancy between the diagnosis rate and PHQ-2 screening results (39–41) indicated that many true PMAD cases were not diagnosed within the study period. Our estimated diagnosis rate was slightly lower than that indicated in previous literature. One meta-analysis (42), which was based primarily on medical records review literature, found that less than one-third of postpartum depression cases are diagnosed. The difference between our observed diagnosis rate and that of previous literature may indicate that screening tools are less accurate than medical records review and may point to additional underdiagnosis of postpartum anxiety (43). Furthermore, the relatively short span of our study (delivery through 3 months after childbirth) may have produced an underestimate of diagnoses.
Lack of familial support, child care, and health knowledge are barriers to receiving maternal mental health care (44). Of note, our results showed that receipt of the routine 6-week postpartum checkup was not associated with having a PMAD diagnosis. This finding may indicate that barriers to diagnosis occur within clinical care and during the diagnostic process itself. Some patients may feel uncomfortable disclosing symptoms, clinicians may inconsistently interpret PMAD symptoms across patient populations, or competing clinical needs may reduce time for discussion of mental health symptoms.
Our study focused on PMAD diagnosis in the early postpartum period, because clinical care typically diminishes in the late postpartum period. There are no additional obstetric checkups in the late postpartum period (45), and Medicaid coverage ends for enrollees without other qualifying criteria 2 months after childbirth (46, 47), underscoring the importance of timely detection and of creating multiple points of care after delivery.
Racial Disparities in Diagnosis Receipt
Race was the most pervasive and prominent factor associated with receipt of a PMAD diagnosis. Although Black and White respondents reported symptoms at similar rates, and our cohort was narrowed to include only respondents with PMAD symptoms, the rate of PMAD diagnosis was lower for Black respondents than for White respondents.
Our findings were consistent with other research showing that Black postpartum patients face additional individual, structural, and historical barriers to care compared with White patients (16). Historical medical discrimination and exploitation have contributed to stigma regarding mental health care in Black communities (48, 49). Black perinatal patients are less likely to receive adequate care and may face racism and provider bias in clinical settings (16, 50–52). As a result, Black postpartum patients may rely less on clinical mental health support and more on spiritual counseling (53). A dearth of culturally inclusive care options may prevent Black postpartum patients from seeking clinical care for mental health concerns.
The diagnostic process itself may prevent Black postpartum patients from receiving timely and accurate care. The Structured Clinical Interview for DSM-IV is a semistructured guide commonly used to confirm psychiatric diagnoses; however, the interview incurs a high level of participant burden, and noncompletion of interviews is more common among Black patients than among White patients (54). Moreover, although Black patients may prefer nonclinical mental health support, research comparing patient preferences with type of treatment received has demonstrated little association between patient preference and treatment modality (53, 55). Thus, patient affinity for or aversion to mental health treatment does not fully explain these disparities.
The factors associated with receipt of a diagnosis differed by race. These differences may provide additional clues to the underlying reasons for higher rates of underdiagnosis among Black respondents. Among Black respondents with symptoms, those with more comorbid conditions and those with more stressors were likelier to receive a psychiatric diagnosis. Patients with more comorbid conditions may interact with the health care system more frequently, incurring more opportunities for screening and diagnosis. A high number of life stressors, including events such as homelessness and incarceration, may reflect greater severity of illness or may indicate access to services, such as social work, that may increase the likelihood of receiving a psychiatric diagnosis.
Effects of Inaccurate Clinical Documentation on Health Equity
The differential rates of PMAD diagnosis for Black and White respondents indicated that PMAD diagnoses were not equally represented in administrative health claims. Without accurate and unbiased assessment of the prevalence of PMAD, such claims data cannot be leveraged to equitably identify or manage maternal mental health conditions.
Underrepresentation of Black patients with PMAD symptoms may create biased data, leading to biased results erroneously indicating that this population is at lower risk for illness. This underrepresentation may perpetuate underdiagnosis in subsequent patients, because clinical support technology, such as machine learning (56), is increasingly used with existing records to help clinicians predict and identify PMADs. Clinical support technology that is trained on biased data will produce biased predictions that could further exacerbate racial disparity in underdiagnosis. Thus, accurate, timely diagnosis of PMADs is not only critical for health equity but also for equitable health informatics.
Understanding and Improving Underdiagnosis
Measuring the factors associated with underdiagnosis is a first step in improving PMAD detection and treatment. Interviews with patients and providers as well as clinical observation of the shared decision making process that leads to clinical diagnosis of maternal mental disorders may elucidate why some patients receive diagnoses and others do not. Because factors associated with screening varied between White and Black patients, our work indicates the importance of drawing on experiences from an array of patient populations.
Structural and cultural components of the health care system likely shape the individual patient-provider interactions that lead to underdiagnosis. The literature (53, 57, 58) indicates that fear of stigma and lack of trust remain barriers to care-seeking behavior, particularly for Black patients. Implementing culturally sensitive care, including strategies such as using nonstigmatizing language to describe mental health, ensuring that providers reflect patients’ identities, and fostering patient-provider relationships (59, 60), may mitigate underdiagnosis. These strategies not only promote care-seeking behavior but may also combat provider bias.
Strengths
Strengths of this study include its novel use of a combination of MI-PRAMS survey data and Michigan Medicaid claims data. This linkage provided insight into the interplay between nonclinical information (e.g., frequency of symptoms) and clinical information (e.g., psychiatric diagnosis) at a larger scale than previously possible and enabled us to control the analyses for nonclinical factors (e.g., stress). This linkage leveraged the strengths of Medicaid claims, which are comprehensive and clinically documented, and those of MI-PRAMS data, which represent patients’ perspectives and provide statistical representation of the population studied. Furthermore, the MI-PRAMS survey uses a strong and established survey methodology. The MI-PRAMS weighting scheme allowed us to estimate study results at a larger scale than would be possible with survey responses alone.
Limitations
This study had some limitations. Generalizability was limited by the exclusion of patients without continuous Medicaid enrollment. Because Medicaid claims are observational data, reflecting real-world conditions in which clinicians use various diagnostic approaches, we could not control the analyses to account for the methods clinicians used to make their diagnoses.
Surveys may introduce systematic errors, such as recall, self-report, or response bias. Furthermore, some survey measures, including the adapted PHQ-2 and number of major life stressors, lacked validation. The MI-PRAMS survey measures symptoms via a modified version of the PHQ-2 that uses an unvalidated answer bank (always, often, sometimes, rarely, never) instead of the validated answer bank (not at all, several days, more than half of days, nearly every day).
In addition, our data were relatively old and may not represent current trends. Professional organizations now recommend universal PMAD screening (61), and the COVID-19 pandemic has negatively affected maternal mental health (62). Our data, however, may serve as a baseline for future evaluation of racial disparities in maternal mental health. Although trends may have changed, systemic racism is unlikely to have abated since data collection.
Finally, this study focused on the relatively narrow time frame between delivery and 3 months after childbirth, but PMADs frequently occur during pregnancy and up to 1 year after delivery. Thus, the findings from the short observational period in our study likely represent underestimates of the number of patients who ultimately receive a PMAD diagnosis within the full postpartum period of 1 year after delivery.
Conclusions
PMAD diagnosis is a critical step in the pathway to treatment and remission. Unfortunately, many PMAD cases are not diagnosed. This study found a paltry 19.8% overall diagnosis rate among symptomatic Medicaid enrollees. Overall diagnosis rate and patient characteristics associated with receiving a diagnosis also varied by race. Future research should further explore underlying mechanisms for these varying diagnosis rates. Improving clinical detection of PMADs will increase access to treatment and improve health outcomes for postpartum patients.
Acknowledgments
The authors thank the Michigan Department of Health and Human Services as well as the Susan B. Meister Child Health and Evaluation Research Center, University of Michigan, for their support.
Supplementary Material
File (appi.ps.20230094.ds001.docx)
- Download
- 185.71 KB
References
1.
Gavin NI, Gaynes BN, Lohr KN, et al: Perinatal depression: a systematic review of prevalence and incidence. Obstet Gynecol 2005; 106:1071–1083
2.
Gaynes BN, Gavin N, Meltzer-Brody S, et al: Perinatal depression: prevalence, screening accuracy, and screening outcomes: summary. Evid Rep Technol Assess (Summ) 2005; 119:1–8
3.
Ding XX, Wu YL, Xu SJ, et al: Maternal anxiety during pregnancy and adverse birth outcomes: a systematic review and meta-analysis of prospective cohort studies. J Affect Disord 2014; 159:103–110
4.
Trost SL, Beauregard JL, Smoots AN, et al: Preventing pregnancy-related mental health deaths: insights from 14 US maternal mortality review committees, 2008–17. Health Aff 2021; 40:1551–1559
5.
Slomian J, Honvo G, Emonts P, et al: Consequences of maternal postpartum depression: a systematic review of maternal and infant outcomes. Womens Health 2019; 15:1745506519844044
6.
Davalos DB, Yadon CA, Tregellas HC: Untreated prenatal maternal depression and the potential risks to offspring: a review. Arch Womens Ment Health 2012; 15:1–14
7.
Misri S, Kendrick K: Treatment of perinatal mood and anxiety disorders: a review. Can J Psychiatry 2007; 52:489–498
8.
Luca DL, Margiotta C, Staatz C, et al: Financial toll of untreated perinatal mood and anxiety disorders among 2017 births in the United States. Am J Public Health 2020; 110:888–896
9.
Guintivano J, Manuck T, Meltzer-Brody S: Predictors of postpartum depression: a comprehensive review of the last decade of evidence. Clin Obstet Gynecol 2018; 61:591–603
10.
Ertel KA, Rich-Edwards JW, Koenen KC: Maternal depression in the United States: nationally representative rates and risks. J Womens Health 2011; 20:1609–1617
11.
Dagher RK, Pérez-Stable EJ, James RS: Socioeconomic and racial/ethnic disparities in postpartum consultation for mental health concerns among US mothers. Arch Womens Ment Health 2021; 24:781–791
12.
Liu CH, Tronick E: Rates and predictors of postpartum depression by race and ethnicity: results from the 2004 to 2007 New York City PRAMS survey (Pregnancy Risk Assessment Monitoring System). Matern Child Health J 2013; 17:1599–1610
13.
Wei G, Greaver LB, Marson SM, et al: Postpartum depression: racial differences and ethnic disparities in a tri-racial and bi-ethnic population. Matern Child Health J 2008; 12:699–707
14.
Kozhimannil KB, Trinacty CM, Busch AB, et al: Racial and ethnic disparities in postpartum depression care among low-income women. Psychiatr Serv 2011; 62:619–625
15.
Chan AL, Guo N, Popat R, et al: Racial and ethnic disparities in hospital-based care associated with postpartum depression. J Racial Ethn Health Disparities 2021; 8:220–229
16.
Mensah M, Ogbu-Nwobodo L, Shim RS: Racism and mental health equity: history repeating itself. Psychiatr Serv 2021; 72:1091–1094
17.
PRAMS Methodology. Atlanta, Centers for Disease Control and Prevention, 2021. https://www.cdc.gov/prams/methodology.htm. Accessed Sept 4, 2021
18.
Pregnancy Risk Assessment Monitoring System (PRAMS): Phase 7 Core Questionnaire—Final. Atlanta, Centers for Disease Control and Prevention, 2012. https://www.cdc.gov/prams/pdf/questionnaire/Phase-7-Core-Questions-508.pdf
19.
Births Financed by Medicaid. San Francisco, KFF, 2021. https://www.kff.org/medicaid/state-indicator/births-financed-by-medicaid/?currentTimeframe=0&selectedRows=%7B%22states%22:%7B%22michigan%22:%7B%7D%7D%7D&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D. Accessed Feb 3, 2022
20.
Live Births to Mothers Who Had Prenatal Care in the First Trimester. Lansing, Michigan Department of Health and Human Services, Division for Vital Records and Health Statistics, 2021. https://www.mdch.state.mi.us/osr/natality/PNCObject.asp?Char=244&Row=STANDARD&Col=RACE&Stat=NxP&Average=A&TableType=G&RowGroup=AGE&COLGROUP=RACE_FIXED. Accessed Feb 2, 2022
21.
Wang E, Glazer KB, Howell EA, et al: Social determinants of pregnancy-related mortality and morbidity in the United States: a systematic review. Obstet Gynecol 2020; 135:896–915
22.
Markus AR, Krohe S, Garro N, et al: Examining the association between Medicaid coverage and preterm births using 2010–2013 National Vital Statistics birth data. J Child Poverty 2017; 23:79–94
23.
Creanga AA, Bateman BT, Kuklina EV, et al: Racial and ethnic disparities in severe maternal morbidity: a multistate analysis, 2008–2010. Am J Obstet Gynecol 2014; 210:435.e1–435.e8
24.
Admon L, Winkelman T, Moniz M, et al: Chronic disease prevalence among mothers delivering in the United States, 2004–2013 [abstract]. Obstet Gynecol 2017; 129:S152–S153
25.
Kurtz S, Levine J, Safyer M: Ask the question: screening for postpartum mood and anxiety disorders in pediatric primary care. Curr Probl Pediatr Adolesc Health Care 2017; 47:241–253
26.
Kroenke K, Spitzer RL, Williams JBW: The Patient Health Questionnaire–2: validity of a two-item depression screener. Med Care 2003; 41:1284–1292
27.
Ko JY, Rockhill KM, Tong VT, et al: Trends in postpartum depressive symptoms—27 states, 2004, 2008, and 2012. MMWR Morb Mortal Wkly Rep 2017; 66:153–158
28.
Pooler J, Perry DF, Ghandour RM: Prevalence and risk factors for postpartum depressive symptoms among women enrolled in WIC. Matern Child Health J 2013; 17:1969–1980
29.
Stone SL, Diop H, Declercq E, et al: Stressful events during pregnancy and postpartum depressive symptoms. J Womens Health 2015; 24:384–393
30.
Shin D, Lee KJ, Adeluwa T, et al: Machine learning–based predictive modeling of postpartum depression. J Clin Med 2020; 9:2899
31.
Ukatu N, Clare CA, Brulja M: Postpartum depression screening tools: a review. Psychosomatics 2018; 59:211–219
34.
Clinical Classifications Software (CCS) for ICD-9-CM. Rockville, MD, Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project, 2017. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. Accessed Dec 9, 2021
33.
Bateman BT, Mhyre JM, Hernandez-Diaz S, et al: Development of a comorbidity index for use in obstetric patients. Obstet Gynecol 2013; 122:957–965
34.
Describing the PRAMS Sample Design for SUDAAN, SAS Complex Survey, SPSS Complex Samples Modules, STATA, and R. Atlanta, Centers for Disease Control and Prevention, 2021. https://dph.georgia.gov/document/document/analysis-tips/download. Accessed Sept 4, 2021
35.
Haak P, Sauter H: Michigan PRAMS Indicator Tables: 2012 Birth Year. Lansing, Michigan Department of Health and Human Services, 2017. https://www.michigan.gov/mdhhs/-/media/Project/Websites/mdhhs/Folder3/Folder21/Folder221/MI_PRAMS_2012_Tables_final_-_secured.pdf?rev=0aa6f89e307b411e8299711b53de8ada&hash=F43463829F1BA562BF90426011F0F459
36.
Haak P, Paciorek J, Sauter H: Michigan PRAMS Indicator Tables: 2013 Birth Year. Lansing, Michigan Department of Health and Human Services, 2017. https://www.michigan.gov/mdhhs/-/media/Project/Websites/mdhhs/Folder3/Folder31/Folder2/Folder131/Folder1/Folder231/2017-08-21_MI_PRAMS_2013_Tables_final.pdf?rev=c1837673d49b427f8f67c4a9f8517e5b&hash=8AD3724FF4B24BAAB71429CAF013B33E
37.
Haak P, Paciorek J, Sauter H: Michigan PRAMS Indicator Tables: 2014 Birth Year. Lansing, Michigan Department of Health and Human Services, 2017. https://www.michigan.gov/mdhhs/-/media/Project/Websites/mdhhs/Folder3/Folder31/Folder2/Folder131/Folder1/Folder231/2017-08-21_MI_PRAMS_2014_Tables_final.pdf?rev=2e59926b078344fda95f41c5e434c59b&hash=8249487E9D0CF4D0F872FC35BAFB2517
38.
Haak P, Sauter H, Paciorek J: Michigan PRAMS Indicator Tables: 2015 Birth Year. Lansing, Michigan Department of Health and Human Services, 2018. https://www.michigan.gov/mdhhs/-/media/Project/Websites/mdhhs/Folder3/Folder74/Folder2/Folder174/Folder1/Folder274/2018-05-01_MI_PRAMS_2015_Tables_final.pdf?rev=352c638b89ca4e8da220d12af916c0ef&hash=3E0F8B7722D6D9CC55A17ABC9F102B09
39.
Gjerdingen D, Crow S, McGovern P, et al: Postpartum depression screening at well-child visits: validity of a 2-question screen and the PHQ-9. Ann Fam Med 2009; 7:63–70
40.
Slavin V, Creedy DK, Gamble J: Comparison of screening accuracy of the Patient Health Questionnaire–2 using two case-identification methods during pregnancy and postpartum. BMC Pregnancy Childbirth 2020; 20:211
41.
Bennett IM, Coco A, Coyne JC, et al: Efficiency of a two-item pre-screen to reduce the burden of depression screening in pregnancy and postpartum: an IMPLICIT network study. J Am Board Fam Med 2008; 21:317–325
42.
Cox EQ, Sowa NA, Meltzer-Brody SE, et al: The perinatal depression treatment cascade: baby steps toward improving outcomes. J Clin Psychiatry 2016; 77:1189–1200
43.
Fairbrother N, Corbyn B, Thordarson DS, et al: Screening for perinatal anxiety disorders: room to grow. J Affect Disord 2019; 250:363–370
44.
Grissette BG, Spratling R, Aycock DM: Barriers to help-seeking behavior among women with postpartum depression. J Obstet Gynecol Neonatal Nurs 2018; 47:812–819
45.
ACOG Committee opinion no 736: optimizing postpartum care. Obstet Gynecol 2018; 131:e140–e150
46.
Status of State Medicaid Expansion Decisions: Interactive Map. San Francisco, KFF, n.d. https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map. Accessed June 29, 2022
47.
Medicaid Expansion in Michigan. San Francisco, KFF, The Kaiser Commission on Medicaid and the Uninsured, 2014. https://files.kff.org/attachment/medicaid-expansion-in-michigan-2-fact-sheet
48.
Gaston GB, Earl TR, Nisanci A, et al: Perception of mental health services among Black Americans. Soc Work Ment Health 2016; 14:676–695
49.
Amankwaa LC. Postpartum depression, culture and African-American women. J Cult Divers 2003; 10:23–29
50.
Nelson T, Brown MJ, Garcia-Rodriguez I, et al: Gendered racism, anxiety, and depression: the mediating roles of gendered racialized stress and social isolation. Ethn Health 2023; 28:12–28
51.
Adebayo CT, Parcell ES, Mkandawire-Valhmu L, et al: African American women’s maternal healthcare experiences: a critical race theory perspective. Health Commun 2022; 37:1135–1146
52.
Mehra R, Boyd LM, Magriples U, et al: Black pregnant women “get the most judgment”: a qualitative study of the experiences of Black women at the intersection of race, gender, and pregnancy. Womens Health Issues 2020; 30:484–492
53.
Bodnar-Deren S, Benn EKT, Balbierz A, et al: Stigma and postpartum depression treatment acceptability among Black and White women in the first six-months postpartum. Matern Child Health J 2017; 21:1457–1468
54.
Gjerdingen D, McGovern P, Center B: Problems with a diagnostic depression interview in a postpartum depression trial. J Am Board Fam Med 2011; 24:187–193
55.
Zittel-Palamara K, Rockmaker JR, Schwabel KM, et al: Desired assistance versus care received for postpartum depression: access to care differences by race. Arch Womens Ment Health 2008; 11:81–92
56.
Cellini P, Pigoni A, Delvecchio G, et al: Machine learning in the prediction of postpartum depression: a review. J Affect Disord 2022; 309:350–357
57.
Lara-Cinisomo S, Clark CT, Wood J: Increasing diagnosis and treatment of perinatal depression in Latinas and African American women: addressing stigma is not enough. Womens Health Issues 2018; 28:201–204
58.
Recto P, Champion JD: “We don’t want to be judged”: perceptions about professional help and attitudes towards help-seeking among pregnant and postpartum Mexican-American adolescents. J Pediatr Nurs 2018; 42:111–117
59.
Di Florio A, Putnam K, Altemus M, et al: The impact of education, country, race and ethnicity on the self-report of postpartum depression using the Edinburgh Postnatal Depression Scale. Psychol Med 2017; 47:787–799
60.
Khan S, Lovell K, Lunat F, et al: Culturally-adapted cognitive behavioural therapy based intervention for maternal depression: a mixed-methods feasibility study. BMC Womens Health 2019; 19:21
61.
ACOG Committee opinion no 757 summary: screening for perinatal depression. Obstet Gynecol 2018; 132:1314–1316
62.
Hessami K, Romanelli C, Chiurazzi M, et al: COVID-19 pandemic and maternal mental health: a systematic review and meta-analysis. J Matern Fetal Neonatal Med 2022; 35:4014–4021
Information & Authors
Information
Published In
History
Received: 22 February 2023
Revision received: 30 May 2023
Accepted: 6 July 2023
Published online: 27 September 2023
Published in print: February 01, 2024
Keywords
Authors
Competing Interests
The authors report no financial relationships with commercial interests.
Funding Information
This research was supported by the Blue Cross Blue Shield of Michigan Foundation (002618.II) and by the Eisenberg Michigan Mental Health Integration Partnership Scholars Award (F052384).
Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).