Maternal Risk Factors Associated with Poor Postpartum Healthcare Utilization

Objectives: Perceptions regarding the benets of postpartum care among mothers and clinicians often differ. Clinicians generally perceive postpartum care as preventative, whereas pregnant and postpartum women often lack knowledge about its preventative benets. As a result many women choose not to return for scheduled postpartum care visits. Methods: To examine if clinically relevant demographic and birth related factors are informative predictors for postpartum healthcare follow-up care, we conducted a population based cohort study of all women who delivered a child in 2012 – 2015 at the New York Mount Sinai Hospital Obstetrics and Gynecology Ambulatory Practice. Data was ascertained from electronic health records. Results: Of the 4,240 unique women who delivered between 2012-2015 at the Mount Sinai Hospital OB/GYN Ambulatory Practice, 1,685 (39.7%) did not return for their postpartum care follow-up appointment. The number of prenatal visits, maternal age, and parity were signicantly associated with postpartum care follow-up. Conclusion for Practice: The purpose of this study was to determine identiable factors associated with reduced postpartum healthcare follow-up utilization. Several clinically relevant variables were associated with the reduced likelihood for attending postpartum care visits. Because pregnant women represent a medically captured population, the results of this study point to the need to increase postpartum healthcare literacy during perinatal appointments especially among younger mothers, women who have had previous deliveries, and those with fewer prenatal visits.


Introduction
The most vulnerable time for both the mother and her newborn occurs not during pregnancy, but rather after delivery, in the postpartum period (Cheng et al. 2006;WHO, 2014). Psychiatric complications, abnormalities in maternal-infant attachment as well as life threatening physical conditions such as hemorrhage, infection, and anemia are readily treatable when identi ed. Unfortunately, the importance of postpartum healthcare remains poorly understood and follow-up visits irregularly attended (Chu et al. 2007).
Towards the effort of better understanding opportunities for intervention, a number of studies have recently begun exploring the various factors that may in uence a woman's decision to utilize available postpartum services. Importantly, women living in low SES, whom reportedly comprise the group at greatest risk for complications after childbirth, are also the group least likely to take advantage of available postpartum services (Department of Health and Human Services, 1998). Even when nancial concerns are eliminated through government subsidy, barriers to postpartum care utilization remain. the primary aim of this study was to utilize electronic health records (EHR) from the OBGYN-AP as a means to determine if clinically relevant maternal and obstetric risk factors are associated with poor postpartum care follow-up. Understanding the risk factors associated with decreased utilization will not only address a current gap in the literature, but may also assist in developing interventions to improve postpartum healthcare attendance and postpartum mental health outcomes among those at greatest risk for postpartum depression (Chu et al. 2007; Morgan et al. 2018).

Study population
The study cohort consisted of the entire population of women seen in the Mount Sinai Hospital OB/GYN Ambulatory Practice (OBGYN-AP) between January 1, 2012 and December 31, 2015. The Mount Sinai Hospital OBGYN-AP serves a population of primarily Hispanic and African-American women who reside in the inner city. Of the women who receive care in the Mount Sinai Hospital OBGYN-AP, 88% are enrolled in a United States government funded healthcare plan. The study was performed in compliance with the Mount Sinai School of Medicine Program for the Protection of Human Subjects and in accordance with the Health Insurance Portability and Accountability Act (HIPAA) security rule guidelines enacted in 2003.

Outcomes
Postpartum care follow-up is de ned as a medical evaluation visit within 6 weeks after childbirth.

Exposures
A requirement for an exposure variable was two-fold. First, the variable must be available as an electronic health record (EHR) at the Mount Sinai Hospital OBGYN-AP and must represent evidence of a veri able postpartum visit. Second, because it is possible that a woman appeared at the clinic, checked in, but subsequently declined clinical evaluation or was not seen by a practitioner, the variable recorded must also be considered "clinically relevant" to assure practitioner involvement during the visit.
Since 2010, the Mount Sinai Hospital OBGYN-AP has been prospectively capturing all healthcare utilization using EHR. Diagnostic information is based on the International Classi cation of Diseases (ICD) codes. Dates of inclusion were selected to conform to the rst and last complete years of available data at the time of analysis.

Selected Exposure variable
The Edinburgh Postnatal Depression Scale (EPDS; Cox et al. 1987) is universally provided to all women at the OBGYN-AP who return for their postpartum visit as an initial step towards assessing mood change in the postpartum period. Each woman is offered a validated version of the EPDS in her own language in the examination room prior to meeting with the medical provider. The EPDS is a ten-item self-report instrument designed to assess symptoms associated with depression using a scale of 0-30. While it cannot be used to provide a diagnosis of depression, it is an effective 1 st stage screening tool for measuring the postpartum mood changes associated with postpartum depression (Cox et al. 1987). Because a "hard stop" was programmed into the postpartum medical record, clinicians must enter the EPDS mood assessment score into the chart before it can be closed and the patient can be released.

Statistical Analysis
The aim of the analysis was two-fold. First, we aimed to determine the rate of woman who delivered a baby at the Mount Sinai Hospital and did not return for postpartum care at the OBGYN-AP. Second, we aimed to explore whether there was any association between postpartum care utilization at the OBGYN-AP and demographics, maternal behaviors and obstetric outcomes. Covariates were selected based on previous research into postpartum healthcare compliance as well as postpartum mood change (Gaynes et al. 2005; Silverman et al., 2017) and can be found in Table 1.
We estimated the relative risk (RR) of postpartum follow-up care rate ratios from Poisson regression models tted to the data. RRs were calculated for each of the covariates in the model. The associated two-sided 95% Wald-type con dence intervals (CI) were calculated, corresponding to a statistical test on the two-sided 5% level of signi cance. Data management and all statistical models were conducted using SAS software, version 9.4.

Demographic and Delivery Characteristics of the Study Cohort
Between 2012 and 2015, there were 4,240 singleton deliveries from unique women. The mean age of the mothers was 27 years (Range 14-51; SD=6.34). 2,206 women (52.0%) self-identi ed as Hispanic/Latino; 1,304 women (30.8%) identi ed as African American; 196 women (4.7%) identi ed as White/Caucasian and; 108 women (2.5%) identi ed as Asian, the remaining 426 women (10.0%) identi ed as either Native American, Paci c Islander or "Other." Vaginal deliveries accounted for 2755 deliveries (65%) whereas 1485 women (35%) delivered via cesarean section. The average gestational age of delivery was 38.4 weeks (range 16-46; SD=2.5); 479 (11%) were born preterm (before 37 weeks gestation) and 97 (2%) were born very preterm (before 32 weeks gestation). 1604 (38%) of the deliveries were identi ed as "high risk", threatening the health of the mother and/or the fetus. 795 (19%) of the delivered infants were initially sent to the neonatal intensive care unit (NICU). Additional characteristics of the cohort are provided in Table 1.

Postpartum care follow-up and EPDS Screening
Of the 4240 unique women who delivered between 2012 and 2015, 2555 (60.3%) returned for their postpartum visit with 2532 (99.1%) of these women having an EMR record of being screened for postpartum mood change in the form of a clinician recorded EPDS score. Of the 23 women who did not have a recorded EPDS score in their medical record, nine included a clinician note indicating that the patient was in fact screened, two had a note indicating that the patient "declined" screening, four EPDS assessments were noted as incomplete secondary to a language barrier and one was noted incomplete due to "illiteracy." Factors associated with poor postpartum care utilization Poor care utilization was signi cantly associated with younger maternal age, RR = 1. 13

Discussion
In a large urban community population-based study we explored postpartum medical care utilization and whether clinically relevant maternal and obstetric risk factors were associated with postpartum follow-up among women living in low-SES whom had recently given birth. Our results con rm that even though postpartum care utilization is well understood by clinicians to be an important preventative care component of maternal health, many women nevertheless fail to utilize these services, even when they are available. More importantly, women living in low-SES, who are also among the group reportedly at Awareness of health literacy disparities among low-SES and minority populations has been steadily increasing (Nesbitt and Palomarez, 2016). However, the relationship between low-SES and health service utilization is complex with factors that remain mostly undetermined or poorly understood (Muennig et al. 2005). What is known however is that the relative increase in morbidity with lower income accounts for a greater loss of health than any risk factor other than normal human aging (Cooper et al. 2012). The widely held hypothesis that poor health literacy across virtually all areas of medical care is a major determinant in the observed disparities among the lower-SES, minority populations and those with lower academic achievement (Mantwill et al. 2015). To this point, the American Congress of Obstetricians and Gynecologists (ACOG) released a Committee Opinion noting the need for increased attention to patients health literacy needs in the effort to promote better health outcomes as they relate to reproductive health (ACOG Committee Opinion No. 676, 2016). Supporting this recommendation, a recent meta-analysis exploring literacy and health disparity con rmed that health literacy is directly related to health utilization as it relates to reproductive health (Mantwill et al. 2015). Unfortunately, the majority of studies surveyed only focused on the relationship between health literacy and prenatal care, as opposed to postnatal care. The few studies surveyed that were focused on postnatal care were generally limited to birth outcomes, glucose follow-up or breast feeding behaviors (Lu and Prentice, 2002). More so, these studies generally sampled those who returned for postpartum care thereby assessing those who were either already literate about the bene ts of postpartum care or simply compliant patients.
Given the current attention postpartum depression has been receiving, it is important to note that lower health literacy has also been shown to be associated with increased postpartum depression symptomatology (Weiss et al. 2009). Similarly, other studies have shown that improving health literacy around postpartum mood improves postpartum depressive symptomatology in those women who choose to return for screening (Howell et al. 2012). While this association seems to be promising, the association may in part represent a causality dilemma. That is, those women experiencing the more severe depressive symptomatology likely do not possess the necessary resources or do not perceive the need (Nadeem et al. 2009) to return for their follow-up care appointment. This is something that should be explored further. Nevertheless, our ndings further underscore that given how pregnant and peripartum women represent a medically captured population, additional efforts towards increasing health literacy during these appointments may encourage more equitable care, further promote healthy reproductive behaviors as well as assist in increasing postpartum service utilization among those at greatest risk.
Although these ndings are of potential importance, we do recognize some possible study limitations. First, the population explored is a clinical sample from a care-based clinic. As a convenience sample, it may not represent the entire population of women. Second, we only assessed each woman once during the sampling years. While it is probable that some woman gave birth more than once during the 2012-2015 time period assessed, their attendance or absence, whether repeated or not was not included. This methodology was chosen to avoid the problem of correlated observations between successive births and seemed to be the clearest approach. Third, the data we utilized was from electronic medical records and as such available data is subject to clinician detection, diagnostic certainty and recording accuracy. This is a well-understood limitation of medical registries and registry-based resolution. Finally, by design, the study's outcome variable only represented treatment-seeking behavior. While, we did not assess the reasons why women elected to return, it is possible that some women returned without choice or without an understanding of choice. As such, the actual incidence of postpartum follow-up, as a function of voluntary behavior may have been overestimated.
Finally, in the effort to further reduce maternal morbidity and mortality, recent recommendations from the American College of Obstetricians and Gynecologists (ACOG) have noted the importance of a comprehensive 'fourth trimester' of care, transitioning from the traditionally recommended single postpartum visit within the rst 6 weeks to multiple visits over the three months following child delivery. As noted in the ACOG Committee Report, "All women should ideally have contact with a maternal care provider within the rst 3 weeks postpartum" (ACOG Committee Opinion No. 736, 2018). Similarly, in recent years there has been both lay and legislative initiatives to provide and in some cases mandate universal depression screening in the postpartum period. However, these efforts will remain largely ineffective as long as a large percentage of high-risk women continue to choose not to return for postpartum healthcare services. Given that pregnant women represent a population that is receiving regular medical care, the results of this study point to the need and opportunity to address the importance of postpartum care follow-up during perinatal appointments and in the immediate window following delivery. Indeed, novel approaches towards enhancing this understanding and/or conveying the importance of postpartum healthcare to those least likely to make use of these services are needed. Further research should therefore focus on interventions to improve postpartum health literacy for those at greatest risk for poor postpartum healthcare utilization.

Conclusion
We set out to examine if any clinically relevant demographic and birth related factors were informative regarding reduced postpartum care follow-up utilization. While the importance of postpartum healthcare utilization has gained considerable attention, the lack of postpartum care follow-up remains a signi cant problem. Several clinically relevant variables were associated with reduced postpartum care follow-up.
Because pregnant women represent a medically captured population, the results of this study point to the need to increase postpartum care literacy during perinatal appointments. Funding: None to report Author Contributions: Dr. Silverman and Dr. Loudon had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis; Study concept and design: All authors; Acquisition, analysis, or interpretation of data: All authors; Drafting of the manuscript: Silverman, Loudon; Critical revision of the manuscript for important intellectual content: All authors; Statistical analysis: Silverman; Administrative, technical, or material support: All authors.

Acknowledgements: None
Availability of Data and Materials: All data presented in this manuscript are stored by the research group of the authors on secure servers at the Icahn School of Medicine at Mount Sinai and handled con dentially. Access to the data is only available to Drs. Silverman and Loudon and cannot be shared secondary to institutional policy.