Objective
The objective of the study was to examine mental health referrals outcomes among obstetric patients at risk for depression.
Study Design
Fifty-one perinatal women who were offered mental health referrals were queried about their behaviors at 4 steps in the treatment engagement process and factors facilitating or impeding each step.
Results
Although 59% of at-risk women accepted mental health referrals, only 27% ultimately engaged in treatment. Women who proactively sought help via a hotline were more likely to accept referrals ( P < .001), contact a referred provider ( P < .001), and engage in treatment ( P < .05) than those who received unsolicited referrals after screening at-risk for depression. Barriers to successful treatment linkage were identified at the patient, provider, and system levels.
Conclusion
Only a minority of women who are at risk for perinatal depression and receive mental health referrals ultimately engage in treatment. Successful linkage may be enhanced via interventions targeting identified barriers; such interventions require prospective evaluation.
Depression is among the most common of perinatal complications, occurring in up to 19% of pregnant or postpartum women. Antepartum depression can compromise pregnancy outcome, and postpartum depression can disrupt the parent-infant relationship and have a negative impact on childhood development. Perinatal depression screening is recognized as an effective means of identifying women at risk for depression during and after pregnancy. The American College of Obstetricians and Gynecologists recommends screening pregnant and postpartum women for depression, and at least 3 state governments have enacted legislation to encourage screening.
Despite the collective momentum favoring perinatal depression screening, its effectiveness can be compromised if appropriate evaluation and treatment of at-risk patients does not follow. Obstetricians often rely on referral to mental health providers to evaluate and treat their patients who screen positive (at-risk) for perinatal depression. Unfortunately, research suggests that primary care-based mental health referral results in a low rate of successful linkage to treatment for depressed patients.
Based on a review of multiple controlled trials of care management and on data from their own practice, Solberg et al identified limited uptake of referral by patients and low follow-through from referring physicians. Mojtabi found that more than 30% of survey respondents in a nationally representative adult sample reported an unmet need for treatment of depressive symptoms, even if subjects actively sought treatment. Utilizing data from a telephone survey of primary care physicians, Trude and Stoddard characterized the failure to connect patients to mental health services as a supply side flaw of the health care system wherein inadequate referral systems between medical and mental health services hamper access.
To date, no studies have tracked depression treatment uptake among perinatal women and gathered qualitative data on the help-seeking experience from the patient’s perspective, and only 1 study tracked a cohort of perinatal women to observe their mental health treatment outcomes after receiving referrals.
To gauge the impact of clinician behavior on treatment follow-through, Flynn et al conducted a longitudinal study tracking depression screening scores, diagnosis for major depressive disorder, impact of inclusion of high Edinburgh Postnatal Depression Scale (EPDS) score in chart on physician discussion of depression treatment, and depression treatment usage from 3 months prior to first prenatal appointment through 6 weeks postpartum. They concluded that physician discussion of depression treatment doubled treatment uptake in the short term for patients with scores 10 or greater on the EPDS. In contrast to the research presented here, Flynn et al did not seek qualitative information from study subjects explaining behaviors or barriers at each step of the referral uptake process.
In the current study, we conducted extended interviews in a group of women at risk for depression as identified via positive screening or after calling a perinatal mental health hotline. In both instances, standardized telephone assessment and referrals were made by trained mental health staff. The subsequent survey and interviews of consenting women were designed to discern barriers (and facilitators) of referral acceptance and treatment engagement for these at-risk women.
Materials and Methods
To understand patient behaviors and barriers at each step of the referral process, this study examined outcomes among a sample of 51 perinatal women who were offered mental health referrals during the course of their obstetric care and agreed to participate in a program evaluation study. Eligible subjects were derived from a preexisting department-based universal depression screening program; the program also sponsors a perinatal depression hotline that is available throughout the state of Illinois via a partnership with the Department of Public Health. Both the screening program and the hotline service were components of CHAPERONE (Creating Healthy Antepartum and Postpartum Expectations, Realities and Outcomes for New and Expectant mothers), a department-based program that utilizes centralized processing of depression screening coupled to referral of at-risk women to an established network of community-based mental health providers. The details of this program have been described elsewhere.
During the study period (June 2006 to November 2007), 958 patients met eligibility criteria; 379 of these (40%) provided consent to be contacted for research, and 51 subjects were recruited to participate in the study (screen positive, n = 28; hotline call, n = 23). This sample size was targeted based on recommended sample sizes for qualitative research. With approval from the institutional review board, women were contacted an average of 9 weeks (range, 2–24 weeks) after being offered mental health referrals and invited to participate in this study and provide informed consent.
Study invitation was intentionally delayed to afford women adequate time to follow through on any mental health referrals they had received unencumbered by the study inquiry; women who had declined mental health referrals were able to be interviewed, on average, earlier than those who indicated their acceptance of referrals during phone evaluation and triage. Among the screened group who were approached for participation, 59% consented to the study as compared with 26% of eligible women who were invited from the hotline group.
Participants completed a mixed-methods telephone interview at a time of their convenience to examine their behaviors at several key steps in the referral uptake process and to determine what factors in their experience facilitated or impeded each step. Our assessment of the treatment engagement process focused on 4 distinct ordinal steps that were selected to best explore patient, provider, and systems barriers: (1) verbally accepting a mental health referral during the phone evaluation; (2) contacting a mental health provider after receiving the referral information; (3) seeing a mental health provider for an initial session; and (4) returning for additional mental health treatment thereafter.
All interviews were conducted by mental health professionals with expertise in perinatal care. The interview consisted of fixed-response survey data regarding specific behaviors (eg, did you contact the counselor to whom you were referred?) and semistructured qualitative follow-up (eg, why did you not contact the counselor?). The interviews ranged from 15 to 45 minutes in length.
Interview data were audio recorded, transcribed verbatim (with identifiers removed), and imported into NVivo version 7 data management software (QSR International, Doncaster, Victoria, Australia). Qualitative data were independently coded by 3 trained raters using a comparative case study design to identify barriers and facilitators to mental health treatment. Coding was reconciled across raters to ensure accuracy and consistency. NVivo software was used to facilitate qualitative data analysis and retrieval. Demographic and fixed-response survey data were assigned numeric values, coded as participant attributes, and analyzed via 2-tailed Fisher’s exact test using GraphPad QuickCalcs software (GraphPad Inc, San Diego, CA).
Results
The demographic and obstetric characteristics of the study sample are reported in Table 1 . As compared with the overall group of eligible participants, study subjects were more likely to be privately insured (86.4% of participants vs 59.4% of overall eligible sample; P = .009). There was no difference between the participant group and the overall eligible group in marital status (partnered vs unpartnered) or gestational time (antepartum vs postpartum).
| Characteristic | % |
|---|---|
| Privately insured | 88 |
| Publicly insured or uninsured | 12 |
| Partnered | 88 |
| Unpartnered | 13 |
| Antepartum | 47 |
| Postpartum | 53 |
| High-risk pregnancy a | 12 |
| Multiple gestation | 4 |
| History of mental illness | 43 |
| On psychotropic medication | 10 |
a Defined as receiving obstetric care from a maternal-fetal medicine specialist.
Mental health treatment data were initially partitioned between screen-positive subjects vs hotline callers and then subsequently compared. Table 2 illustrates the stepwise treatment engagement outcomes for each of these groups. The hotline group was more likely than the screened group to accept a referral (87% vs 36%; P < .001), contact a provider (70% vs 18%; P < .001), see a provider for an initial session (52% vs 14%; P < .01), and continue treatment beyond the initial session (43% vs 14%; P < .05). However, similar proportions (50% of hotline callers and 60% of screened women) who initially accepted referral in both groups did not follow through with at least 2 mental health visits.
