Association of Health Literacy Among Nulliparous Individuals and Maternal and Neonatal Outcomes

Key Points Question Is maternal health literacy associated with maternal and neonatal health outcomes? Findings In this cohort study of 10 038 nulliparous individuals, those with inadequate health literacy had greater odds of cesarean delivery, major perineal lacerations, and neonates with small for gestational age status, low birth weight, and low 5-minute Apgar score, although absolute differences were generally small. Meaning These findings suggest that inadequate health literacy is common among pregnant individuals in the US and is associated with a variety of adverse maternal and neonatal outcomes.


Introduction
Health literacy is the degree to which individuals can obtain, process, and understand the basic health information and services they need to make appropriate health decisions. 1,2 Health literacy is considered an important social determinant of health, but it is not the same as education or language. [1][2][3] It has been well documented that even individuals with strong educational attainment and linguistic skills may have trouble obtaining, understanding, and using health information, given the complexity of the health care system and of health information. Nearly one-half of individuals in the US have low or marginal health literacy, commonly referred to as inadequate health literacy. 1,2,4 Health literacy has been previously associated with health outcomes. There are 3 primary mechanisms through which health literacy is thought to affect health: health care access and use, patient-practitioner interactions, and health behavior, each of which may be influenced by patient and health system factors. 5,6 These theoretical causal pathways have been demonstrated in many contexts, with inadequate health literacy linked to a variety of nonobstetric adverse health care processes, behaviors, and outcomes. 6,7 Such adverse effects include lower use of preventive care, less adherence to treatment, more hospital stays, and higher mortality rates. 7 Because of the deep and broad impact of health literacy on health, addressing health literacy is a national priority; objectives of the Healthy People 2020 initiative include improving health literacy as one domain of social determinants of health interventions. 8 Despite ample data on health literacy in nonobstetric medicine, it has rarely been studied in obstetrics, and when it has been studied, it has been primarily to assess knowledge. 9 For example, inadequate health literacy has been associated with poorer understanding of reproductive health topics, such as contraception, prenatal genetic screening, diabetes management during pregnancy, and preeclampsia. [9][10][11][12][13][14][15] It has also been associated with disease-specific health behaviors, such as not seeking preconception care among individuals with diabetes. 9,16,17 On the basis of the abundance of evidence on health literacy as a social determinant of health in other areas of medicine, it has been hypothesized that inadequate health literacy may deepen existing inequities in maternal and neonatal outcomes. 2,3 However, to our knowledge, the association of health literacy with outcomes has not been studied in a general obstetric population. Thus, our objective was to evaluate the association between health literacy and maternal and neonatal outcomes.

Study Overview
In this study, more than 10 000 nulliparous individuals with singleton gestations were recruited in the first trimester to undergo longitudinal assessments. Recruitment was conducted at geographically diverse locations and was designed to sample a population reflective of the general US population. Individuals were eligible for enrollment if they had a live singleton gestation, had no previous pregnancy beyond 20 weeks' gestation, and were between 6 weeks 0 days and 13 weeks 6 days of gestation at recruitment. Exclusion criteria included maternal age younger than 13 years, history of 3 or more spontaneous abortions, current pregnancy complicated by suspected fatal fetal malformation or known aneuploidy, assisted reproduction with a donor oocyte, multifetal reduction, or plan to terminate the pregnancy.

Data Collection
Data were collected via multiple sources, including in-person interviews, surveys, and medical record review. Participants completed 3 study visits with trained research personnel. At least 30 days after delivery, trained and certified record abstractors reviewed the medical records of all participants and their neonates and recorded final birth outcomes. 18 Full details of the study protocol previously have been published. 18 This analysis addresses maternal health literacy as the exposure of interest. To measure health literacy, all participants completed the Rapid Estimate of Adult Literacy in Medicine-Short Form (REALM-SF) at visit 2, which was between 16 and 21 weeks' gestation. 18,19 The REALM-SF is a 7-item word recognition test-the words are behavior, exercise, menopause, rectal, antibiotics, anemia, and jaundice-commonly used for quick assessment of health literacy. It was derived from the 66-item version of the REALM, both of which have been validated in diverse settings outside of pregnancy. 19 Participants are shown a written list of words and asked to say each word aloud. If pronunciation is incorrect or not attempted, a point is not awarded for that word. The number of correctly pronounced words is tallied to generate the total score. Scores are categorized according to grade level using standardized scoring criteria. 19 In nuMoM2B, the REALM-SF was performed only among participants who chose to complete interviews in English, including individuals who may have spoken English as a second language. Individuals who spoke English as a second language were included in the primary analysis because they chose to complete all aspects of the study in English, which suggests they are likely to attempt health care communication in English. This analysis included all participants with available REALM-SF data.

Outcomes
On the basis of the established causal pathway models for health literacy, 5,6 we hypothesized that health literacy would be associated with outcomes via their impact on pregnancy-related knowledge, behaviors, communication, and care access, all of which could be affected by patient and health system factors. Using this pathway, we selected maternal and neonatal outcomes that could plausibly be affected by factors such as patient-practitioner interactions, health behaviors, or health Gynecologists. 20 Perineal laceration was evaluated because of its potential association with patientpractitioner interactions. Neonatal outcomes included small for gestational age (SGA) status neonate (<10th percentile by Alexander criteria 21 ), low birth weight (<2500 g), very low birth weight (<1500 g), macrosomia (>4500 g), preterm birth (<37 weeks' gestation), admission to the neonatal intensive care unit, 5-minute Apgar score less than 4, and receipt of neonatal respiratory support. Maternal demographic and clinical characteristics measured via assessments at visit 1 (between 6 and 14 weeks) were assessed as potentially confounding factors. Self-reported race and ethnicity were assessed because they were considered important social determinants of health.

Statistical Analysis
For this analysis, consistent with definitions and terminology of prior literature, REALM-SF scores were analyzed by dichotomizing them as inadequate vs adequate, in which inadequate scores were everything less than a high school level (ie, score <7). 19,22 Such scores were labeled inadequate because scores do not indicate an absolute characteristic about an individual, but rather indicate a relational difference regarding the ability to have the best success in interfacing with the health care system. We examined differences between baseline demographic and clinical characteristics by health literacy status using χ 2 and t tests. We assessed differences between maternal and neonatal outcomes by health literacy using χ 2 tests. Bivariable and multivariable Poisson regression models were constructed to estimate relative risks of each outcome, with adequate health literacy as the referent. Multivariable models accounted for potential confounders determined on the basis of a priori hypotheses using a directed acyclic graph (eFigure in the Supplement). Two sensitivity analyses were performed. In the first, we excluded education from the multivariable models, given the possible collinearity of health literacy with educational attainment; all other variables from the primary model were retained in these models. Second, we excluded participants who spoke English as a second language; these models included all variables in the primary model. Self-reported race and ethnicity were not examined as effect modifiers because there is no reason the effect of health literacy should differ by racial or ethnic identity.
Analyses were performed using StataMP statistical software version 15.0 (StataCorp). All statistical tests were 2-tailed and considered significant at the P < .05 level. This analysis was completed from July to December 2019.

Results
Of 10 038 participants in nuMoM2B, 9341 had REALM-SF data available for analysis and were eligible for inclusion; the mean (SD) age was 27.0 (5. 6  Very low birth weight, macrosomia, and preterm birth were more frequent among participants with inadequate health literacy but did not differ after adjusting for potential confounders. However, participants with inadequate health literacy had neonates with greater relative risk of 5-minute Apgar score less than 4 (aRR, 2.78; 95% CI, 1.16-6.65).
Both sensitivity analyses supported the findings of the primary analysis ( Table 3). In both sensitivity analyses, the findings did not change, with all RRs remaining within 10% of the primary analysis.

Discussion
In this study, nearly 1 in 5 pregnant individuals had inadequate health literacy. Inadequate health literacy was associated with differences in maternal and neonatal outcomes, including greater risk of cesarean delivery, major perineal laceration, SGA status, low birth weight, and low 5-minute Apgar score. Although the absolute differences were small, these differences may have a meaningful public health impact on a population level. In addition, as has been demonstrated repeatedly in other contexts, health literacy, education, and language are discrete constructs. Health literacy is not simply a reflection of educational attainment or primary language, but rather is a distinct social determinant of perinatal health.

Results in Context
Health literacy is associated with health learning, which has documented downstream effects on health knowledge, behaviors, and outcomes. 6 For these reasons, health literacy has been associated with many adverse health behaviors and outcomes, including poorer use of preventive health care, suboptimal management of chronic diseases, poorer self-reported health status, increased frequency of hospitalizations, greater likelihood of medication errors, excess adult mortality, and higher health care costs. 1,7,8,23,24 The outcomes of our study suggest that corollaries to these adverse health outcomes exist in the obstetric context. Prior limited work in obstetrics suggests that inadequate health literacy is associated with poorer pregnancy-specific knowledge and some health behaviors. 9 Our study extends this work by demonstrating that in a population of pregnant individuals enrolled in a longitudinal study, these associations between inadequate health literacy and adverse outcomes exist. The association of health literacy with health behaviors and outcomes may hold even greater import in pregnancy than in nonobstetric settings because pregnancy is a time of rapid knowledge acquisition, increased demand for health behavior changes, and intensive need for physician counseling. Failure to achieve optimal health in pregnancy may alter the risk profile for the individual and family across the life cycle.

Research Implications
Future investigation must explore the mechanisms between health literacy and obstetric outcomes.
Proposed mechanisms are diverse and may include differential adherence to prenatal care or health behavior guidelines (eg, nutrition and weight gain), differential understanding of counseling, alterations of the effects of psychosocial stressors, or differential engagement or self-advocacy. 9,34 In addition, inadequate health literacy may compound practitioner biases or exacerbate poor patient-practitioner communication. Better understanding of pregnancy-specific mechanisms may as well as greater study on the unique learning and motivation features of pregnancy. 36 The role of health literacy in postpartum morbidity, including perinatal depression, also warrants investigation. 9,15 In addition, health literacy is complex and multifaceted. 28 Future investigations should investigate other aspects of health literacy, such as print literacy, oral literacy, health numeracy, and comprehension, in obstetrics. 23 Studying the full meaning of health literacy also includes understanding how different types of health literacy may be more relevant for certain populations.

JAMA Network Open | Obstetrics and Gynecology
For example, health numeracy skills 37,38 may be particularly important for individuals with gestational diabetes, who must manipulate numbers to self-manage nutrient intake and medication administration. 12

Strengths and Limitations
A major strength of this study is the use of a large sample of pregnant individuals that is representative of the US population. The cohort is extraordinarily well characterized and includes detailed prospective assessments of health literacy prior to the occurrence of outcomes. 18 The assessment of health literacy via the REALM-SF substantially enhances the fidelity of literacy evaluation, in contrast to investigations that use education as a proxy for literacy.
However, there are limitations to consider. As with all studies of health literacy, this study is observational and subject to unmeasured confounding. Second, health literacy is a skill that may evolve, and this study captured health literacy at a single time point in midpregnancy; however, it is unlikely that health literacy levels would be expected to change substantively during pregnancy.
Third, as noted already, the REALM-SF has limitations as a measure of health literacy. For example, it is primarily an assessment of reading and pronunciation, rather than comprehension, and, in contrast to newer measures, it does not comprehensively assess all types of health literacy, such as health numeracy. Fourth, although this is one of the largest studies of health literacy in pregnancy, the study may be underpowered to detect differences in uncommon events, such as postpartum readmission.
Fifth, participants were enrolled in a longitudinal investigation that began in early pregnancy and were recruited from large medical centers; thus, the findings may not be fully generalizable. The young age of the population and their enrollment in this study may be underlying reasons for the lower prevalence of inadequate health literacy than that of general population.

Conclusions
In summary, this cohort study found that inadequate health literacy was common among pregnant individuals in the US. Health literacy is an important, yet often underrecognized, social determinant of health, and may be associated with health disparities. Our work suggests that health literacy may be independently associated with the risk of adverse perinatal outcomes, separate from educational attainment and language. These findings represent a call to action for the development of evidence-based interventions to optimize communication and enhance health education.