Evaluation Plan for Postpartum Depression Screening Initiative
Evaluation Plan for Postpartum Depression Screening Initiative
Although a number of screening and treatment programs for postpartum depression have been implemented, many of these programs have not been studied to determine efficacy (reviewed by Yawn et al., 2012b). This lack of evidence has prevented a number of agencies and organizations from issuing recommendations, including the American College of Obstetrics and Gynecology and the U.S. Preventive Services Task Force.
The Institute of Medicine's (2001) report, titled "Crossing the Quality Chasm: A New Health System for the 21st Century," proposed six aims to improve health care in America. These aims were providing safe, effective, patient-centered, timely, efficient, and equitable care. One of the rules outlined to help achieve these aims was to ensure that patients received care based on the best scientific evidence available. In keeping with this goal of providing evidence-based care, an evaluation plan is outlined below for a Postpartum Depression Screening Initiative (PDSI) implemented at a local health care organization.
Several studies have attempted to determine if the screening for, and treatment of, postpartum depression improves care outcomes. Pignone and colleagues (2002) performed a meta-analysis of six adult depression screening efficacy studies, published between 1994 and 2001, and found a two to three-fold increase in the diagnosis of depression and a 7% decline in symptoms six months post screening (Pignone et al., 2002). Yonkers and colleagues (2009) assessed the efficacy of the federal Healthy Start depression initiative in the state of New Jersey and found no advantage to screening for pregnant and postpartum women seeking care in publicly-funded facilities. A more recent meta-analysis of the research literature found screening for adult depression in primary care settings provided no benefit (Gilbody, Sheldon, and House, 2008). In addition, an evaluation of a state program in New Jersey, which screened women for postpartum depression, found no increase in treatment-seeking behavior among low-income women on Medicaid (Kozhimannil, Adams, Soumerai, and Huskamp, 2011). All of these studies, including those encompassed by the meta-analyses, had methodological problems, including non-randomized samples and insufficient statistical power.
A recent randomized, controlled study used the Edinburgh Postnatal Depression Scale (EPDS) to evaluate a postpartum depression screening and treatment initiative for a Hong Kong sample (Leung et al., 2011). Follow-up evaluations at 6 and 18 months revealed the initiative resulted in a lower risk of depression (Risk Ratio = 0.59, 95% CI 0.39-0.89) after 6 months only. The evaluation also revealed a significant reduction in the number of visits to the doctor's office for the infant during the first six months postpartum. Of the battery of instruments used to assess depression at 6 and 18 months, the EPDS instrument provided the greatest discrimination (p = 0.001) between the intervention and control groups. Leung and colleagues (2011) also noted that the EPDS was the only instrument used in the study that had been previously validated, with sensitivity, specificity, and positive predictive values of 82%, 86%, and 44%, respectively.
Another recent randomized, controlled study examined the efficacy of primary care screening and treatment for postpartum depression in the U.S. population (Yawn et al., 2012a). The intervention involved clinician training in screening, diagnosis, follow-up, and management of postpartum depression. The screening involved two survey instruments: the EPDS and the 9-item Patient Health Questionnaire (PHQ-9). Since intervention involved clinician training, clinics, rather than patients, were randomized across 21 different states. Analysis of patient demographic information revealed that women seeking care at clinics within the intervention group tended to be less educated, economically disadvantaged, and unmarried. Since all three of these variables have been associated with poorer depression treatment outcomes, intervention group patients would tend to skew the results of the study in favor of no effect. The results revealed that PHQ-9 scores were not significantly different between the two groups; however, intervention group patients were significantly more likely to receive a diagnosis of depression (66% vs. 41%, p = 0.0001), medication (56% vs. 35%, p < 0.0001), and counseling (20% vs. 11%, p = 0.02). In addition, intervention group patients were more likely to have improved depression symptoms 12 months after screening (adjusted odds ratio = 1.74, 95% CI 1.05-2.86). A meta-analysis of studies that used the PHQ-9 found a pooled sensitivity and specificity of 80% and 92%, respectively (Gilbody, Richards, Brealey, and Hewitt, 2007).
The current Postpartum Depression Screening Initiative (PDSI) was implemented without consideration for the need to randomize patients or clinicians. A control population has not been defined either, but should a sister public health department be willing, their patients could function as a control population if a PDSI has not been implemented. This sister organization would have to be willing to enroll patents in the study, and in the absence of outside funding, be willing to cover any costs incurred by the study. The choice of a sister organization should depend on the similarity of the patient socioeconomic demographics and a willingness to become a stakeholder in improving the quality of care for women suffering from postpartum depression.
In the absence of a control population, a retrospective study could be done to assess the mental health of patients who were seen at the public health department prior to implementation of the PDSI. The EPDS and PHQ-9 instruments could be used to jog the memory of former patients concerning their mental health. The medical records of these patients could also be reviewed for evidence of mental health problems. Leung and colleagues (2011) discovered the infants of depressed mothers tended to visit the pediatrician significantly more often, which suggests this type of independent measure could also be used.
The possible stakeholders include obstetrician/gynecologists (Ob/Gyn), pediatricians, nurses, and mental health workers (Price, Corder-Mabe, and Austin, 2012, p. 448). Phase two of the PDSI involved educating Ob/Gyns and their nursing staff concerning the value of, and methods for, postpartum screening. Interviews of these providers would likely generate valuable insights into whether the intervention changed their practice methods. These interviews should be done at the end of Phase three.
The weekly data log sheets are filled out by the Ob/Gyn nurses during, or shortly after, the first offer to screen for depression during a patient visit. These data logs are accessible to the evaluation team; therefore the need to contact Ob/Gyns and their nursing staff would be limited to situations where the data logs are unclear. A review of the data logs should be done during and at the end of Phase three, to minimize the loss of information due to poor memory.
The remaining stakeholders, pediatricians and mental health providers, would be relevant to the evaluation because they could provide information concerning PDSI outcomes. Pediatricians may become aware of parental mental health issues affecting their ability to provide infant care, thus interviewing these care providers may provide valuable insight into PDSI efficacy (Price, Corder-Mabe, and Austin, 2012, p. 448). Interviews of mental health providers will provide data for the number of referrals patients take advantage of, if treatment was offered after further evaluation, whether treatment was undertaken by the patient, treatment adherence, treatment efficacy, and treatment complications. The time frame for interviewing pediatricians and mental health providers should be between 1 and 2 years after birth of the child.
An important independent control for the interview data and the data logs are the medical records for mother and child. Information concerning follow-ups, referrals, treatment choice, treatment adherence, and outcomes should be contained in these records and thus offer an opportunity to validate the information contained in the data logs and gleamed from interviews. Gaining access to these records will require provider and organizational consent. Review of these records should be done at least one year after the birth of the child.