Caregivers’ Health Literacy and Gaps in Children’s Medicaid Enrollment: Findings from the Carolina Oral Health Literacy Study

Background and Objectives Recent evidence supports a link between caregivers’ health literacy and their children’s health and use of health services. Disruptions in children’s health insurance coverage have been linked to poor health care and outcomes. We examined young children’s Medicaid enrollment patterns in a well-characterized cohort of child/caregivers dyads and investigated the association of caregivers’ low health literacy with the incidence of enrollment gaps. Methods We relied upon Medicaid enrollment data for 1208 children (mean age = 19 months) enrolled in the Carolina Oral Health Literacy project during 2008–09. The median follow-up was 25 months. Health literacy was measured using the Newest Vital Sign (NVS). Analyses relied on descriptive, bivariate, and multivariate methods based on Poisson modeling. Findings One-third of children experienced one or more enrollment gaps; most were short in duration (median = 5 months). The risk of gaps was inversely associated with caregivers’ age, with a 2% relative risk decrease for each added year. Low health literacy was associated with a modestly elevated risk increase [Incidence Rate Ratio (IRR) = 1.17 (95% confidence interval (CI) 0.88–1.57)] for enrollment disruptions; however, this estimate was substantially elevated among caregivers with less than a high school education [IRR = 1.52 (95% CI 0.99–2.35); homogeneity p<0.2]. Conclusions Our findings provide initial support for a possible role of caregivers’ health literacy as a determinant of children’s Medicaid enrollment gaps. Although the association between health literacy and enrollment gaps was not confirmed statistically, we found that it was markedly stronger among caregivers with low educational attainment. This population, as well as young caregivers, may be the most vulnerable to the negative effects of low health literacy.


Introduction
Health Insurance Coverage Affects Appropriate Health Care during the First Years of Life Children's health insurance coverage is a major determinant of preventive and continuous health care [1][2][3]. Low-income children enrolled in Medicaid are more likely to receive timely preventive care, including well-child visits, compared with their counterparts without insurance coverage [4][5][6][7][8]. Nationally representative US data confirm that being uninsured is associated with low levels of adherence to professional recommendations for wellchild visits, with the lowest rates found among children eligible but not enrolled in public insurance [9]. This is an important observation because well-child visits and preventive care during the first years of life have demonstrable benefits [10][11][12]. Similarly, continuity of care has been linked to favorable health outcomes, including lower emergency department utilization [13]. These observations are aligned with the concept of a 'medical home' which, according to the American Academy of Pediatrics, encompasses the characteristics of pediatric care considered essential for all children'' and is defined as a ''model of primary care that is accessible, continuous, comprehensive, familycentered, coordinated, compassionate, and culturally effective'' [14][15].
Current legislation supports the expansion in Medicaid coverage in 2014 [16] and has the potential to benefit a large and vulnerable component of the US population. Insurance coverage is an important enabling factor for health care use overall and its absence leads to under-utilization and unmet medical needs [8]; however, its availability is not sufficient for enrollment or the establishment of a medical home [10]. To affect health outcomes, the expansion of coverage must be met with adequate enrollment.
To What Extent are Disruptions in Health Insurance Detrimental for Children? Current evidence demonstrates that disruptions (''gaps'') in coverage are likely to decrease both access to a medical home and the likelihood of seeking routine care for young children [17][18][19]. A 2006 report [20] found that between 18-36% of children in four states had insurance gaps over a period of up to 2 years. Other recent studies confirmed that insurance gaps may be frequent, with 25% of children experiencing a gap in Medicaid enrollment over a 12-month period in Oregon [17], and 14-61% of children in five other states experiencing similar gaps over a 4-year period [21]. Because of its frequency and negative effects, unstable health insurance coverage is a phenomenon that needs further investigation as a risk factor for children's health [2,[22][23].
What is known about the Correlates of Health Insurance Enrollment Gaps?
State Medicaid programs have a compelling interest in enrolling and maintaining coverage for eligible children but there is a paucity of evidence to inform policy makers on relevant strategies and best practices to accomplish these goals [24]. Although state programs strive to keep enrollment and renewal procedures simple, most reports on enrollment gaps implicate caregiverrelated factors. For example, studies by DeVoe et al. [17] and Satchell and Pati [22] relied upon nationally representative data and found that poverty and maternal education were the strongest correlates of enrollment gaps. Gaps may be attributed to changes in family circumstances that render parents and children ineligible; nevertheless, caregivers' difficulty in navigating the complex administrative processes can also explain why those eligible for public or private insurance are challenged to enroll and maintain enrollment. The challenges of the required forms and paperwork are well-documented [20,[25][26]. As one example, recent findings indicate that Medicaid enrollment forms may be excessively demanding to read and comprehend [27]; in fact, most states fail to meet their self-established readability criteria [28].

The Role of Health Literacy
A recent report [29] identified low health literacy as a risk factor for lack of adult health insurance and outlined the direct implications for the anticipated Medicaid expansion and implementation under the Affordable Care Act (ACA). Several studies have documented an association between caregiver health literacy and child health and services' use [30][31][32][33]. Health literacy's relationship to Medicaid enrollment gaps may offer one potential explanation of the relationship between parental health literacy and child health outcomes. To the best of our knowledge, no data exist to date regarding the possible association between caregivers' health literacy and gaps in their children's public health insurance enrollment. Our aims were to 1) describe and characterize gaps in Medicaid enrollment of young, low-income children of English-speaking families in North Carolina, and 2) identify caregiver factors associated with this phenomenon, including health literacy.

Ethics statement
Written consent and HIPPA approval were obtained by all participants and the study was approved by the Biomedical Institutional Review Board at the University of North Carolina-Chapel Hill (approval #07-0837).

Parent study sampling frame
This investigation utilizes data obtained in the context of the Carolina Oral Health Literacy (COHL) Study, which initially enrolled 1,405 clients of the Special Supplemental Nutrition Program for Women, Infants and Children (WIC) in seven counties in NC: Brunswick, Buncombe, Burke, New Hanover, Orange, Robeson, and Wake [34]. The study's aim was to investigate the impact of caregivers' health literacy on their children's oral health and use of services. The study population consisted of low-income predominantly female caregivers who were 18 years of age or older, English-speaking, and the primary caregiver of a Medicaid-eligible healthy child. Purposeful quota sampling [35] was used to ensure adequate representation of African American (AA; 40%) and American Indian (AI; 20%) participants. Detailed descriptions of the study procedures, interview instruments and baseline outcomes have been reported previously [36][37][38]. In brief, non-random WIC sites were selected based on geographic region, rural or urban makeup, population demographics, existence of very active WIC clinics, and established working relationship with the investigators [34]. Consecutive WIC-attending clients were approached by study personnel while in the waiting areas and asked if they would to 8 questions included in the study's eligibility screener. Caregivers who were determined to be eligible and agreed to participate were accompanied to a private area to complete the in-person study interview. Of note, to qualify for WIC, caregivers must have an income ,185 percent of the federal poverty level.

Data Sources
Data were collected in 30-minute in-person interviews with one of the two trained study interviewers between July 2008-July 2009 in the domains of socio-demography, caregivers' self-reported personal and children's general and dental health status, dental knowledge and behaviors, quality of life, dental neglect, general self-efficacy, and health literacy. Study personnel screened eligible participants to ensure that unique children were enrolled in the cohort. In June 2011, we obtained Medicaid enrollment and claims files for children from the NC Division of Medical Assistance, ranging from January 2004 to December 2010, a period that covered the lifetime Medicaid enrollment history for all children up to the end of CY-2010. We merged these files with interview data that were obtained at baseline using the children's unique Medicaid identification number that was provided by the caregivers at study enrollment. All COHL children were Medicaid-eligible or enrolled at baseline, as per our inclusion requirements. Of the 1405 caregivers recruited, 1242 had children with a Medicaid identification number at the baseline interview. Children's unique Medicaid identification number was used to link interview and Medicaid claims data. Subsequently, we excluded dyads wherein children had no Medicaid data (n = 31; 2% of total), caregivers under 18 years old (n = 2; 0.2%), and one caregiver (0.1%) who had missing information on health literacy. Thus, the final analytical sample included 1208 child/caregiver dyads.

Outcome Definition and Covariates
The study's primary analytical endpoint was the incidence of Medicaid enrollment disruptions or ''gaps''. During the study period (2004-2010) Medicaid eligibility criteria in NC, as well as procedures of enrollment and re-enrollment varied for different categories of child/caregiver dyads; however, these criteria included, among others, income, resources, residence, and proof of citizenship. Re-enrollment processes included periodic verifications of eligibility criteria at intervals ranging between 3-12 months, unless changes were self-reported by the beneficiaries. The current list of eligibility criteria and required documentation can be found at: http://www.ncdhhs.gov/dma/forms/famchld. pdf. For the purposes of this study we defined gap as a period during which a child was Medicaid-eligible but not enrolled, which was preceded and followed by periods of enrollment, as described by Copper et al. [39]. This definition of enrollment disruption was conservative, in a manner that prevented children with a permanent change in Medicaid-eligibility from being classified as having a gap. In such instances, there would be no subsequent reenrollment in the NC files and the child would be labeled ''lost to follow-up'' rather than having an enrollment gap.
Apart from caregivers' difficulty with necessary forms and processes and administrative reasons, there are several other reasons why enrollment interruptions occur for children covered by Medicaid. Among others, these include: a child/family moving out of state, a child's acquisition of comprehensive private health insurance, a caregiver's request for termination of coverage, and family income changes. Any of these result in permanent eligibility changes without subsequent re-enrollment. We cannot ascertain with certainty the reason for permanent eligibility changes and some may be due to difficulty in processes/procedures; however, our goal for using a conservative definition of gaps was to ensure we were capturing administrative-related lapses due to parental actions that resulted in an interruption in enrollment.
Enumeration of enrolled/non-enrolled months began at the month of the baseline COHL interview and ended in December 2010. For each child, we identified and counted enrolled/nonenrolled periods and calculated their duration in months. We calculated 'enrollment proportion' as the ratio of months enrolled divided by months eligible with a range of 0-100. History of Medicaid enrollment was available from January 2004-December 2010, a period that included all available Medicaid data up to 2010 for all children. To further characterize patterns of Medicaid enrollment, we calculated periods of non-enrollment, enrollment gaps, and percent time enrolled both during the study duration and the entire period available.
We measured health literacy using a comprehension and numeracy-based instrument, the Newest Vital Sign (NVS). The NVS is a nutritional label accompanied by six questions, requiring approximately three minutes for completion. The NVS is a validated test with good psychometric properties and it has been used extensively since its introduction [40][41]. According to Weiss et al. [40] who developed and tested the NVS, the instrument is reliable (Cronbach alpha .0.76 in English), correlates with the test of functional health literacy in adults (TOFHLA), requires 3 minutes for administration and, thus, is suitable for use as a quick screening test for limited health literacy. According to Osborn et al. [42], the NVS is able to reliably identify individuals with limited literacy skills, which was the group of interest in this study (vs. the upper end of the literacy spectrum). In the context of our study, the NVS's Cronbach's alpha was 0.75. The NVS score ranges from 0-6, wherein values of 0-1 are indicative of ''low'', 2-3 ''moderate'' and 4-6 ''higher health literacy'' [42]. We used this 3-level categorical classification as well as a dichotomous one where NVS score 0-1: ''low health literacy'' and 2-6: ''higher health literacy''. The examination of the latter dichotomy was driven by our working hypothesis which considered low caregiver health literacy as a predictor of children's Medicaid enrollment gap; in this scenario, a contrast between the lowest category of the NVS index (containing individuals most likely to face difficulties with forms and administrative processes) and higher ones was the most meaningful one.
Socio-demographic covariates included caregivers' race, sex, age, education, marital status, number of children, and child's age. Race was self-reported and categorized as white, African American (AA), and American Indian (AI). Caregivers' age was measured in years and classified as a tertile-categorical variable for descriptive purposes and a continuous variable for analytical purposes. Education was grouped in four categories: less than high school (HS) education, HS/General Education Diploma (GED), some technical or college training, and college or higher education. Marital status was classified as single, married, and divorced/separated/other. Children's age was measured in months and was categorized in five groups: 0-11, 12-23, 24-25, 36-47, and 48-59 months.
As a measure of caregivers' self-perceived health status, we used the National Health and Nutrition Examination Study (NHANES) self-reported item containing five possible answers: excellent, very good, good, fair, and poor. For analytical purposes, the responses were converted to a dichotomous variable, excellent/very good/ good versus fair/poor.

Data availability
We have created and made available a de-identified dataset to accompany this manuscript, as supplemental material available online. This tab-delimited dataset (Dataset S1) contains all original and derived variables described and used for this study's analyses, including a data dictionary as a separate spreadsheet.

Statistical Analyses
We used descriptive statistics for our initial data presentation and x 2 tests to compare the distribution of low/adequate health literacy and Medicaid enrollment gaps across strata of covariates. Because the occurrence of enrollment gaps is a time-dependent event, ignoring the time at-risk could introduce bias of substantial magnitude and unknown direction. For this reason, we used multivariate Poisson regression models that accounted for time at risk using months enrolled as the exposure and a log-link function. Although our modeling approach did not consider the length of each gap, we support that it is superior to the enumeration of months not enrolled due to reasons outlined above. To identify a final model, we employed a backward hierarchical variable selection strategy [43], starting with a full model that included all covariates and a p,0.2 criterion. We included caregivers' race, age, and general health status a priori in all models and obtained exponentiated model coefficients corresponding to Incidence Rate Ratios (IRR) and 95% confidence intervals (CI). For the interpretation of results, we followed an effect-estimation (estimation of the magnitude and precision of the association) versus hypothesis-testing (determination of the presence of an association or not) approach [44].
For sensitivity analysis, we developed a second identical model considering all Medicaid enrollment history available, prior to and after enrollment in the COHL study. All Medicaid enrollment history covered the lifetime period for all enrolled children, up to December 2010. In addition, to explore for potential vulnerable sub-groups, we conducted a post hoc homogeneity analysis to determine whether educational attainment and race modified the association between caregivers' health literacy and insurance gaps. To do so, we compared education level and race-specific estimates against the main effect estimate using a homogeneity x 2 test. Because these post hoc stratified analyses are de facto underpowered, we used an a priori p value criterion of ,0.2 for the detection of heterogeneity [45]. We used Stata version 12.1 (StataCorp LP, College Station, TX) and SAS version 9.3 (SAS Institute, Cary, NC) software for data analyses.

Results
The primary caregivers' socio-demographic characteristics and responses to the self-reported health status items, both overall and stratified by health literacy group, are presented in Table 1. There was a 2:2:1 distribution of white, AA, and AI participants with the vast majority of all participants being female. Caregivers' and children's mean age at the baseline interview was 27 years and 17 months, respectively. A small proportion (11%) of caregivers reported their general health as fair or poor. Fifteen percent of participants were classified as low health literacy based on their NVS score. This proportion was higher among AA and AI participants and among those who were younger, had lower educational attainment, and worse self-reported health status.
The median follow-up time was 25 months (mean = 24; range = 18-29), with 11% of children experiencing an interruption of Medicaid enrollment with no subsequent re-enrollment. This latter phenomenon was almost twice as high among children of caregivers with some technical/college or higher education versus those with less than high school education (15% versus 8%). Thirty percent of children experienced at least one enrollment gap ( Table 2). Relative to gaps, the majority of the children (90%) experienced only one, 9% had two, and 1% had three. Twenty percent of these interruptions had duration of one month, 50% four months or less, while 21% were 12 months or longer. As anticipated, children with no gaps were enrolled longer, on average for six more months compared to those with gaps. Relative to bivariate associations, continuous enrollment was more frequent among children of older caregivers, older children, and those with more siblings (Table 2). Continuous enrollment was more frequent among children whose caregivers reported having better health but the difference was small (3%).
We present the final Poisson regression model (A) for enrollment gap incidence in Table 3. After adjusting for race, education, gender, and all other covariates, inclusion in the low health literacy group was associated with a modestly elevated risk for an enrollment gap (adjusted IRR = 1.17; 95% CI 0.89-1.54). The risk of gaps was inversely associated with caregivers' age, with a 2% relative risk decrease for each added year. An association between female gender and decreased risk of enrollment gaps was also noted; however, this finding was based on an analysis including only 46 male caregivers and thus should be viewed with caution. Stratified analyses (Table 4) revealed that educational attainment significantly modified the association between literacy and gaps (homogeneity p,0.2), with the highest risk estimates found among caregivers with less than high school education (IRR = 1.52; 95% CI 0.99-2.35). Race-stratified estimates did not depart substantially from homogeneity. Inspection of the exploratory models (B) that considered all available Medicaid enrollment history did not reveal any important differences compared to our main analyses.

Discussion
In this study of approximately 1200 young children of lowincome predominantly female caregivers enrolled in WIC, 30% had a gap in Medicaid enrollment over 25 months. Most experienced only one gap of short duration. Our findings revealed a weak association between caregivers' health literacy and their children's Medicaid enrollment gaps; in the entire cohort, independent of race, education, and age, low health literacy was associated with a non-significant gap risk increase of about 20%. Nevertheless, this effect was substantially elevated among caregivers with less than high school education compared to those with high school or higher education; caregivers with low education may be the most vulnerable to the effects of low health literacy. Moreover, above and beyond other caregiver factors, caregivers' age was inversely associated with the risk of gaps.
The concept of Medicaid's serving as a revolving-door for some children has been recognized and more steps must be taken by state Medicaid programs to alleviate this problem. While we cannot confirm it using our data, this phenomenon may be more closely related to administrative issues (i.e. difficulty with the enrollment and maintenance process) rather than eligibility disruptions. States are required to record and report on children's coverage stability but this has generally not been undertaken [46]. It is not uncommon for children who are uninsured for less than a full year to be reported as insured and this cohort of children has received less attention in public health research and policy discussions [23]. Olson [2] found that children insured for part of the previous year had higher proportions of delayed care and unmet medical care versus those uninsured for the full year. A similar finding was reported by DeVoe and colleagues [17], who compared children with gaps longer than six months versus those never insured.
Children from poor and near-poor families have been reported to be 4-5 times more likely to have lapsed coverage than those from high-income families [22]. At the same time, decreased preventive care utilization among children who had Medicaid insurance gaps was irrespective of their insurance status after discontinued enrollment [47]. Taken together, these findings underscore the dramatic effects that interruptions may have for children, their families, and the health care system. The association between Medicaid interruptions and subsequent children's health outcomes and related costs is fertile ground for future research.
As more health literacy findings are reported, the multi-level effects of health literacy are becoming increasingly evident. Health literacy is recognized already as having a pivotal role in numerous health outcomes [48][49], insurance status [29], and health disparities [50][51], and recently was termed the ''sixth vital sign'' [52]. Strategies that attempt to circumvent or mitigate the effects of low health literacy, such as the Federal Plain Language Act and simplification of enrollment procedures are promising [53], but more steps towards implementation are necessary. Although it is impossible to quantify the precise contributors to the relationship between health literacy and Medicaid enrollment gaps from our data, it appears that the readability of Medicaid forms [27] and user-friendliness of the enrollment renewal process [26,[54][55] are potential targets for addressing literacy-sensitive concerns. This notion was recently mirrored by findings reported by Pati et al. [56] who found that mothers with inadequate health literacy were less likely to receive child care subsidy than those with adequate health literacy, implicating the application processes as health literacy-sensitive step.
Our findings on the incidence of enrollment gaps are consistent with previous reports of Medicaid insurance gaps among young children. Previously published estimates [20,46,57] of any gap of enrollment were 25% over a 12-month period in Oregon, 26-35% over 18 months in Ohio, 18% over a 2-year period in Louisiana, 25% over 12 months in Rhode Island, and 33% over an 18-month period in in Virginia. State, community, and local populationspecific parameters may account for these relatively small variations. Our finding of increased risk for children's enrollment gaps among younger caregivers is a likely reflection of additional stressors and strains experienced by young parents/caregivers compared to older ones that could include education, employment, transportation, or others.
The finding of a weak association between enrollment gaps and literacy in the full cohort analysis requires further investigation. First, it must be acknowledged that in absence of statistical confirmation, an association may in fact be non-existent. Further, it is possible that the negative effects of low health literacy are manifested only among vulnerable subsets of the population, including those with low educational attainment, a scenario supported by our data. It is also possible that truly a weak association exists; this result might be a reflection of the NC Medicaid Program's success in addressing issues with the enrollment and maintenance processes and caregiver community support (i.e. an effect of WIC and practice health coordinators). In fact, while NC had the highest readability level goal (6-8 grade) for enrollment forms among all states with such guidelines, it was 40 th in terms of actual readability level [28]. This parallels our finding of a 52% risk increase for enrollment gaps among caregivers in the lowest educational attainment group. Nevertheless, other differences in caregivers' age, family conditions, or children's health care needs may influence the likelihood to enroll in Medicaid and remain continuously enrolled. As an example, Cooper et al. [39] found that among children with asthma, those without insurance gaps had more hospitalizations compared with those with insurance gaps. This finding was a possible consequence of health    problem-initiated and enrollment pattern driven care-seeking behavior.
Evidence supporting an association between parental health literacy and child health insurance status has been reported by large, rigorous studies, including those by Pati et al. [56] and Yin et al. [58]. Although these studies did not investigate enrollment gaps, they add to the growing evidence base linking health literacy with children's insurance status. The findings of the DeVoe et al. [17] and Satchell and Pati [22] studies underscored the importance of socio-economic stressors, including poverty and maternal education as strong predictors of enrollment gaps. This effect was reflected in our study's stratified analyses, wherein the effect of low health literacy varied significantly according to education level, with the largest estimates found among the lowesteducated caregivers.
Our findings have limitations. This WIC study cohort was not a probability sample, an important factor because WIC enrollment alone has been shown to confer benefits for child health behaviors and outcomes [59][60]. The WIC program is the largest public health nutrition program in the US and, as such, serves a large percentage of low-income children and their mothers. To put this in perspective, in fiscal year 2013, more than 2 million women and infants nationwide participated in WIC, whereas there were 264,000 WIC participants in North Carolina. Our sample size was modestly sized, including approximately 1,200 dyads and 199 caregivers with low health literacy; however, it must be noted to the best of our knowledge no other cohorts to-date exist with person-level data on parents' health literacy and children's Medicaid enrollment information. For many practical reasons, our population was limited to English-speaking caregivers over the age of 18, limiting the generalizability of our findings beyond that demographic group. Although there is a Spanish version of the NVS tool, most of our survey instruments at the time of the study had not been developed for or validated in Spanish. Because Hispanic children are often the most likely to have insurance gaps [22], the absence of this demographic group is an obvious study limitation. Future studies among these high-risk, vulnerable groups are warranted. Another limitation that must be pointed out is that NVS is only one of many available instruments available to assess health literacy; as a comprehension and numeracy-based test, it is not a comprehensive health literacy assessment.
Understanding Medicaid enrollment gaps is a complex issue. State differences in the enrollment and re-enrollment processes warrant the investigation of the association between health literacy and enrollment gaps in diverse population-based samples in different states. It is possible that enrollment and re-enrollment in Medicaid require cognitive and/or executive functioning skills not well-captured by the instrument used in this study. Prolonged interruptions could be due to changes in eligibility, but gaps can be due to administrative issues on both the program and the client's sides. Our data included rich information from the perspective of the client, but lacked information from the perspective of the program.

Summary and implications
Health insurance gaps have been linked to increased health disparities [61], leading to a poorer quality of health care [62]. Accordingly, correcting problems with children's health insurance enrollment and coverage should be a priority for stakeholders. We found that 30% of young children experienced one or more enrollment gaps over a 2-year period. Our data did not reveal any important association between caregivers' health literacy and the incidence of these gaps; however, they indicated that low health literacy may be a risk factor for enrollment gaps among the lowesteducated caregivers, who may be the most vulnerable to its detrimental effects. Moreover, caregivers' young age showed an association with increased risk for enrollment gaps. Considering the economy of scale of the anticipated health insurance coverage expansion in 2014, insurance coverage and continuity issues are certain to become ever more important challenges. Because health literacy issues are, unlike many other social determinants, amenable to interventions, it is critical that they are kept on the agendas of policy makers. Although age is a non-intervenable factor, it may be pertinent to consider caregivers' (young) age as a possible additional risk factor for their children's public insurance enrollment disruptions.
The Congressional Budget Office projects that the currently 34 million people that are covered by Medicaid will increase to 47 million in 2014 with the implementation of the ACA [63]. Considering this huge expansion, health insurance coverage should be examined from a social and behavioral perspective to inform and guide states' actions in the implementation of the Medicaid coverage expansion [16,[64][65][66]. Gaps in insurance coverage should be considered with a rigor similar to nonenrollment. More research is needed to comprehensively characterize health system and individual-level risk factors for public health insurance gaps; however, it must be acknowledged that population-wide, policy solutions such as continuous eligibility and automatic re-enrollment in public insurance may be more effective approaches compared to ''individual risk factor'' strategies. Beyond the obvious utility of identifying susceptible groups or individuals (i.e., young caregivers of children), it is critical that planned interventions are evidence-based with attention to the critical issues of health literacy and cultural appropriateness.

Supporting Information
Dataset S1 Tab-delimited dataset containing all original and derived variables described and used for this study's analyses, including a data dictionary as a separate spreadsheet. (ZIP)