Profiles of Non-medical Drivers and Health Burden Associated With Care Seeking Among Rural Women

Abstract

Background:

Non-medical drivers of health, including social, demographic, and geographic conditions, play an important role in shaping women’s healthcare use in rural settings. How these intersecting factors form patterns that influence care seeking remains underexamined.

Methods:

A cross-sectional survey (n = 159) was conducted through public health service sites across 29 rural counties in Texas, United States. K-means clustering identified groups of women with shared patterns of health burden, health-related social needs, and geographic proximity to care. Chi-square tests and one-way ANOVA assessed demographic differences, and a Generalized Linear Model estimated the proportion of times participants reported receiving needed care based on profile membership and demographics.

Results:

Seven clusters, interpreted as profiles, reflected variation in combinations of health burden, health-related social needs, and geographic proximity. Cluster membership and difficulty affording healthcare were significantly associated with higher odds of receiving needed care. Women in profiles characterized by multiple co-occurring constraints, as well as one profile marked by elevated health-related social needs, had significantly lower odds of receiving care compared to the reference profile with no constraints.

Conclusions:

Findings suggest that profile-based, analytic approaches may help clarify differences in unmet care needs and inform more targeted efforts to address delayed or forgone care in rural settings.

Keywords

access to care cluster analysis community health rural health social determinants of health Texas women’s health underserved communities

Introduction

Persistent differences in health outcomes are closely tied to the social and structural conditions in which people live.¹ These contextual factors include interconnected demographic, socioeconomic, geographic, and place-based conditions that can increase exposure to adverse life circumstances, shape the navigation capacity of institutional systems, and create substantial barriers to accessing the healthcare, social services, and resources essential for building and maintaining overall health and well-being.² Often referred to as non-medical drivers, such factors tend to operate in combination rather than in isolation, influencing both patterns of engagement with healthcare systems, and, ultimately, health outcomes.³ In rural areas, influential intersecting qualities of place include less availability of primary and specialty care providers, fewer public social support services, long travel distances to care, and constrained employment opportunities that reduce insurance coverage and income, limiting the ability to absorb costs and overcome such barriers to seeking and receiving care when needed.^4,5

With approximately 16% of its 30.5 million residents living in rural areas,⁶ Texas has the largest rural population in the United States and offers a relevant case study for examining differences in rural care-seeking patterns common at the state level.⁷ These differences are driven by a combination of institutional (eg, absence of Medicaid expansion), geographic (eg, expansive rural land areas), and sociodemographic (eg, unemployment, uninsurance rates, food and housing insecurity) conditions.⁷ Recent evidence suggests that these intersecting conditions shape care-seeking behavior and accessibility in ways not fully captured by single-factor or demographic approaches, particularly in under-resourced regions.⁸ General care availability also remains a concern, with rural Texas counties being more likely than urban counties to be classified as primary care shortage areas.⁹

Women living in rural areas across the United States experience these conditions acutely, contributing to differences in health behavior and outcomes when compared to those in urban areas.^10,11 Rural women in Texas are no exception and face disproportionate risks across several indicators, especially as multiple such social and geographic determinants (such as physical location, community-level vulnerability, and insurance access) frequently cooccur (Supplemental Figure A.1). These co-occurring conditions can influence whether and when women seek care, contributing to worse health outcomes for rural women compared to men and their urban counterparts, including higher rates of chronic disease, obesity, and hospitalization, as well as lower rates of preventive screening and later-stage cancer diagnoses with poorer survival rates.^12-14 They also contribute to reduced engagement in maternal care.¹⁵ In 2020, only 68% of Texas mothers initiated prenatal care during the first trimester (the lowest rate of any U.S. state), with even lower rates observed among Hispanic and non-Hispanic Black women.¹⁵ Delayed or forgone care represents not only a marker of unmet needs, but also a potential mechanism through which non-medical drivers translate into long-term health differences among groups.¹⁶

This study examines an application of understanding care seeking using survey data from rural women to identify shared contextual profiles defined by indicators of health burden, health-related social needs, and geographic care availability. By assessing how these profiles relate to getting care when needed, the findings demonstrate an analytic approach that can offer insight into localized drivers of delayed or forgone care and may be useful for service planning in comparable rural settings.

Methods

Study Area

This study focused on rural counties in Texas’ Public Health Region 4/5, serviced by jurisdiction 4/5N. Rurality was determined using the U.S. Department of Agriculture’s (USDA) 2023 Rural-Urban Continuum Codes (RUCC); categories 4 to 9 were considered rural.¹⁷ Of 38 counties in the region, 29 met the inclusion criteria (Supplemental Figure A.2).

Participants

The study presents cross-sectional data collected from adult (over 18 years old) individuals identifying as female (henceforth, women) receiving health services within the study area. Data were collected online through the survey platform Qualtrics. Participants were primarily recruited in person at Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) service centers, particularly those serving clients from rural counties. While at the centers for appointments, which included WIC voucher pick-up as well as other general health and education services, participants were recruited both actively by staff and passively via advertisements. A small portion of participants were recruited at community-health events near WIC centers. Participants received a small bag of personal hygiene items (worth less than $15) if they completed the survey on-site; otherwise, they received no incentive.

Survey Construction and Measures

The survey included up to 50 items assessing demographics, health status, geographic access, socioeconomic factors, health information preferences, and care-seeking behaviors. After screening for completeness and consistency, 159 responses were retained. Variables were then operationalized and prioritized for analysis due to sample size.¹⁸

Demographic characteristics were obtained from self-report. Items prioritized were age, pregnancy status within the past year, race/ethnicity, insurance status, employment status, and difficulty affording healthcare. Age was continuous and included to account for differing health needs of younger and older women. Employment, insurance coverage, and the ability to afford healthcare were included as they are often associated with greater access to and use of care.⁵ Insurance and pregnancy statuses were assessed with yes-no questions. Employment statuses included employed full-time, employed part-time, seeking work, retired, and not currently working. Race/ethnicity categories included American Indian/Alaska Native, Asian, Black/African American, Hispanic/Latino, Middle Eastern/North African, Native Hawaiian/Pacific Islander, White, and other. Employment and race/ethnicity categories were dichotomized due to low sample sizes. Difficulty affording all healthcare costs was assessed using a 4-point Likert-type scale and dichotomized based on the 2 highest points of the scale.

Health burden was defined by having a chronic condition, as well as ratings of physical and mental health and health-related quality of life, widely used and reliable survey measures that have been associated with healthcare utilization.¹⁹ Chronic condition status was assessed with a yes-no question and included to reflect increased care needs or exacerbated health issues.²⁰ Health quality was self-assessed using a 5-point Likert-type scale and considered poor if the worst 2 scale points were rated for current mental and physical health and if the worst 3 scale points were rated for health-related quality of life.

Health-related social needs were assessed with ratings of access to basic resources needed for living (transportation, food, housing, and income).² Access to each of these resources was rated using a 3-point Likert-type scale. Geographic availability of care was represented by using counts of up to 5 healthcare facility types within 30 min, a threshold at which travel time begins to significantly deter care-seeking behavior.²¹

The primary outcome was getting different types of care when needed. Delays or failures in receiving needed care are linked to poorer outcomes, including reduced long-term care utilization, worsened chronic disease, and higher mortality.²² Care access frequency was calculated as a proportion, summing the number of (up to 13) care types received when needed and dividing by the total number of care types received and not received.

Analytical Approach

Cluster analysis was used to identify subgroups of rural women with shared patterns of health burden, health-related social needs, and geographic availability of care and examine their collective impact on receiving care when needed. Latent class analysis was not used due to the small sample and numerous categorical variables, which risked overfitting and unstable results.²³ Data were first standardized and explored with K-means clustering using the scikit-learn library in Python, assessing the contribution of various individual measures to defining contextual profiles. To improve performance and interpretability, 3 binary composite variables were created: health burden, health-related social needs, and geographic availability (Supplemental Table 1). Simplifying variables reduced sparse feature spaces and “cube corner” effects that complicate cluster geometry.²⁴ The cutoffs selected reflected indicators of contextual constraints and included high health-related social needs (≥2 indicators), low geographic availability of care (≤2 facilities within 30 min), and high health burden (having a chronic condition or ≥2 poor health indicators). While some specificity was lost, dimensionality reduction improved stability and preserved the ability to examine the overlapping and mutually reinforcing contextual factors shaping care access.²

The consolidated variables were explored again with K-means clustering to identify natural groupings (contextual profiles) within the data. K-means was selected due to its computational efficiency, widespread use in applied research, and ease of interpretation.²⁴ The final seven-cluster solution demonstrated strong performance across validation metrics, indicating reliable, distinct, and well-defined clusters (silhouette score, 0.93; Davies-Bouldin Index, 0.403; Calinski-Harabasz Index, 513.7).

Once clusters were established, one-way ANOVA and Chi-square tests of independence with post-hoc adjusted standardized residual analysis examined if clusters differed by demographic characteristics. A Generalized Linear Model (GLM) with binomial distribution and a log link function was used to model the percentage of times successfully getting care when needed based on cluster membership while accounting for age; employment, insurance, and pregnancy status; identifying as a person of color (POC); and difficulty affording healthcare costs. Statistical tests were performed using SPSS, version 28.0 (Chicago, IL, USA). P values greater than 0.05 were considered significant.

Ethics Approval

The study was performed in compliance with relevant laws and institutional guidelines. All participants’ privacy rights were observed, and participants provided informed consent via the survey interface prior to participation.

Results

Participant Characteristics

Table 1 summarizes demographic characteristics of participants. Slightly over half of participants identified as white (54.1%). The mean participant age was 32.7 years (SD = 11.7). Most were employed either full- or part-time (57.2%). Approximately 23.2% of respondents reported having chronic health conditions. The majority (82.4%, 131) indicated they had children; the average number of children was 2.1 (SD = 1.4). A smaller share reported being born outside the United States (10.0%) or speaking a primary language other than English (8.0%).

Table 1.

Demographic Characteristics of Participants.

Characteristic	Number	%
Race
White	86	54.1
Person of color	73	45.9
Employment status
Full/part-time	91	57.2
Other (retired, caregiver, etc.)	68	42.8
Coverage status
Has insurance	112	70.4
Has difficulty affording care	72	45.3
Health burden
Has a chronic condition	37	23.2
Poor physical health	21	13.2
Poor mental health	34	21.4
Poor health-related quality of life	62	39.0
Pregnant within the last year	59	37.1
Health-related social needs and geographic availability
Transportation insecurity	25	15.7
Food insecurity	65	40.9
Housing insecurity	39	24.5
Income insecurity	83	52.2
Has geographic access to care	110	69.2
	Range	Mean (SD)
Age	18-78	32.7 (11.7)

Contextual Profiles from the Clustering Approach

The clustering approach identified 7 distinct clusters, interpreted as profiles, in the sample based on indicators of contextual constraints including high health burden, high health-related social needs, and low geographic availability of care. Table 2 summarizes these dimensional characteristics of women in each cluster. Figure 1 shows the scatter plots of the characteristics from 2 angles to visualize the position and separation of each cluster, with lines from the origin marking the central point (centroid).

Table 2.

Profiles of Contextual Constraints Identified Through K-means Clustering.

Cluster label	n (%)	High health burden (%)	High health-related social needs (%)	Low geographic access (%)	Success rate getting care
Cluster 0 (reference/no constraints)	38 (23.9)	0	0	0	91.6% ± 18.6
Cluster 1 (high constraints)	22 (13.8)	100	59	100	73.5% ± 32.5
Cluster 2 (low constraints)	27 (17.0)	0	0	100	91.6% ± 17.5
Cluster 3 (moderate constraints)	20 (12.6)	0	100	100	71.0% ± 27.1
Cluster 4 (low constraints)	21 (13.2)	0	100	0	87.9% ± 19.2
Cluster 5 (low constraints)	18 (11.3)	100	0	0	93.7% ± 10.3
Cluster 6 (moderate constraints)	13 (8.2)	100	100	0	74.1% ± 32.1

Indicators of contextual constraints are shown for women in each cluster. Cluster 0 is the reference cluster with no indicators. Low constraint is defined by the presence of 1 indicator; moderate, by the presence of 2 indicators; and high, by the presence of all indicators.

Figure 1.

Health burden, health-related social needs, and geographic availability of care visualized for women across clusters.

Cluster 0, with 23.9% of women, represented the reference (control) cluster (with no indicators of constraint). Low constraint profiles (with 1 indicator) were represented by Cluster 2 (17%), Cluster 4 (13.2%), and Cluster 5 (11.3%). Moderate constraint profiles (with 2 indicators) were Clusters 3 (12.6%) and 6 (8.2%). The high constraint profile (with all indicators) was represented by Cluster 1 (13.8%). Overall, 41.5% of women fell into low profiles, 20.8% into moderate profiles, and 13.8% into the high profile.

Table 3 presents demographic characteristics across clusters. The highest rate of insurance coverage was in Cluster 5, a low constraint profile, while Cluster 6, a moderate profile, had the lowest coverage. Employment rates were also highest in Cluster 5 (72.2%) and lowest in Cluster 6 (46.2%). Women of color were most represented in Cluster 4 (61.9%), a low constraint profile, and least represented in Cluster 5 (33.3%). Difficulty affording healthcare was most frequently reported in Cluster 1 (63.6%), the high constraint profile. Cluster 4 had the most women who had been pregnant within the past year (57.1%), while Cluster 5 had the least (22.2%).

Table 3.

Distribution of Demographic Characteristics Across Clusters.

Characteristic	Cluster 0 (n = 38) (%)	Cluster 1 (n = 22) (%)	Cluster 2 (n = 27) (%)	Cluster 3 (n = 20) (%)	Cluster 4 (n = 21) (%)	Cluster 5 (n = 18) (%)	Cluster 6 (n = 13) (%)	Test statistic	P-value	Effect size
Age (mean)	31.8	35.7	32.7	30.1	29.6	35.6	35.8	F = 1.03	.41	η² = .04
Person of color (%)	16 (42.1)	13 (59.1)	11 (40.7)	8 (40)	13 (61.9)	6 (33.3)	6 (46.2)	χ² = 5.64	.46	Cramer’s V = .19
Employed (%)	22 (57.9)	12 (54.5)	17 (63.0)	10 (50.0)	11 (52.4)	13 (72.2)	6 (46.2)	χ² = 3.37	.76	Cramer’s V = .15
Has insurance (%)	30 (78.9)	14 (63.6)	19 (70.4)	11 (55.0)	13 (61.9)	18 (100)	7 (53.8)	χ² = 14.11	.03*	Cramer’s V = .30
Difficulty affording healthcare (%)	15 (39.5)	14 (63.6)	11 (40.7)	6 (30)	9 (42.9)	10 (55.6)	7 (53.8)	χ² = 6.82	.34	Cramer’s V = .21
Pregnant within last year	13 (34.2)	5 (22.7)	11 (40.7)	9 (45.0)	12 (57.1)	4 (22.2)	5 (38.5)	χ² = 3.37	.76	Cramer’s V = .15

P < .05.

A one-way ANOVA and chi-square tests of independence assessed statistical differences in demographic characteristics among clusters. Post hoc analyses with a Bonferroni correction (α_adj = .004) were conducted using adjusted standardized residuals to identify cells contributing to overall significance. Only having insurance demonstrated a significant association with cluster membership, χ²(6, N = 159) = 14.11, P < .05, with women in Cluster 5 being significantly more likely to have insurance (adjusted residual = 2.92, P < .001).

Impact of Cluster-Profile on Receiving Care

A GLM with binomial distribution and a log link function modeled the percentage of times getting care when needed based on cluster membership, accounting for age; employment, insurance, and pregnancy status; identifying as a POC; and difficulty affording healthcare. An Omnibus test was significant, χ²(6, n = 151) = 98.80, P < .001, indicating a better fit than the intercept-only model. When compared to a reduced model with only demographic predictors, a likelihood ratio test indicated the full model with cluster membership provided a significantly better fit to the data, χ²(6) = 70.32, P < .001. Despite some overdispersion in both models, goodness of fit statistics also supported this conclusion, with the full model demonstrating lower deviance (deviance/df = 3.09) compared to the reduced model (deviance/df = 3.46), as did Akaike’s and Bayesian information criteria (AIC = 584.11 vs. 643.43, BIC = 650.10 vs. 678.96).

Results (Table 4) indicated that both cluster membership and difficulty affording health care were significant predictors of receiving care when needed. Compared to women in Cluster 0, those in Cluster 1 (OR = 0.28, P < .001), Cluster 3 (OR = 0.15, P < .001), Cluster 4 (OR = 0.47, P = .020), and Cluster 6 (OR = 0.26, P < .001) had significantly lower odds of receiving care when needed. Difficulty affording health care was the only demographic characteristic independently associated with reduced odds of receiving care when needed (OR = 0.42, P < .001); no others were significant (all P > .05).

Table 4.

GLM Results Examining Receiving Needed Care Based on Cluster Membership and Demographic Characteristics.

	B	Std. error	95% Wald confidence interval		Hypothesis test			Odds radio (expB)
Variable	B	Std. error	Lower	Upper	Wald Chi-Square	df	Sig.	Odds radio (expB)
(Intercept)	3.08	0.42	2.25	3.91	52.76	1	0.00	21.70
Cluster 6 (moderate constraints)	−1.33	0.32	−1.97	−0.69	16.77	1	0.00*	0.26
Cluster 5 (low constraints)	0.43	0.40	−0.35	1.21	1.16	1	0.28	1.54
Cluster 4 (low constraints)	−0.76	0.33	−1.39	−0.12	5.43	1	0.02*	0.47
Cluster 3 (moderate constraints)	−1.88	0.30	−2.47	−1.30	39.82	1	0.00*	0.15
Cluster 2 (low constraints)	−0.62	0.33	−1.26	0.02	3.60	1	0.06	0.54
Cluster 1 (high constraints)	−1.28	0.28	−1.84	−0.72	20.24	1	0.00*	0.28
Cluster 0 (reference/no constraints)	0^a
POC (yes)	0.05	0.17	−0.28	0.39	0.10	1	0.75	1.06
POC (no)	0^a
Employed (yes)	−0.02	0.18	−0.38	0.34	0.02	1	0.90	0.98
Employed (no)	0^a
Has insurance (yes)	−0.31	0.20	−0.71	0.08	2.40	1	0.12	0.73
Has insurance (no)	0^a
Difficulty affording healthcare (yes)	−0.88	0.18	−1.24	−0.52	23.20	1	0.00*	0.42
Difficulty affording healthcare (no)	0^a
Pregnant within the last year (yes)	0.10	0.20	−0.29	0.49	0.26	1	0.61	1.11
Pregnant within the last year (no)	0^a
Age	0.00	0.01	−0.02	0.02	0.00	1	0.99	1.00

Events: times_got_care.

Trials: number_care_needed.

Model: (Intercept), Cluster, POC, employed_binary, self_insured_binary, difficult_afford_health, pregnant_binary, Age.

P < .05.

Set to zero because this parameter is redundant.

Discussion

Seven clusters, interpreted as profiles, were identified based on contextual constraint indicators of health burden, health-related social needs, and geographic availability of care. Cluster membership significantly predicted getting care when needed, more so than individual demographic factors. Apart from one profile (4), those with fewer indicators of constraint (0, 2, 5) showed no impact on receiving care when needed, whereas profiles with more indicators (1, 3, 6) were associated with worse odds. Demographic variables were not significant, except for difficulty affording care. Limited sample size likely contributed to the lack of demographic differences, a finding deviating from broader literature.²⁵

Notably, women in some low-moderate constraint profiles (3 and 4) with low health burden reported worse care access than those in higher constraint profiles, suggesting compounding effects of social and geographic barriers even without health issues, as both showed high health-related social needs. Although those groups indicated fewer indicators of ongoing health issues warranting care, their lower reported access could indicate possible difficulties obtaining acute care for emergent issues, an important concern given that limited emergency care availability in rural areas is linked to worse health outcomes, including higher mortality.²⁶ The relatively high proportion of pregnant women in Cluster 4, while not statistically significant, may have influenced this pattern and could emerge more clearly with a larger sample size. This is consistent with prior research showing that pregnant women experiencing multiple health-related social needs or socioeconomic difficulties often receive fewer prenatal services and procedures, even when they have Medicaid.²⁷ Nonetheless, the findings also indicated that coverage-related factors still mattered within the context of constraints: insurance status emerged as the only statistically significant demographic variable differing among clusters, and difficulty affording care was the only significant demographic predictor of reduced odds of receiving care when needed. This aligns with evidence emphasizing the enabling role of insurance in encouraging the use of health services in rural and under-resourced contexts.^8,28-30

More generally, the findings add to a growing body of work demonstrating that care use is rarely constrained by a single determinant, but instead reflects the interaction of multiple co-occurring needs.^3,31 Consistent with prior studies showing that multidimensional needs profiles can better explain variation in care use than isolated indicators,^32-34 especially when insurance coverage and financial resources intersect with geographic barriers,³² the findings emphasize that layered constraints common in rural areas may compound unmet need even among individuals without chronic conditions or self-reported poor health.³⁵ That characteristics such as employment, insurance, or race alone were not independently predictive further underscores the inadequacy of unidimensional approaches and the need for more holistic assessment frameworks that move beyond basic demographic targeting to optimize services and expand access points.

Although broader reforms addressing fundamental issues with healthcare costs, insurance coverage, and structural drivers of health-related social needs are often concentrated at institutional or policy levels, such frameworks can combine individual-level health-related social needs screening with area-level measures to improve targeting of outreach, navigation, and wraparound services at more local levels.^34,36 This type of practice-, community-, and/or locality-oriented analysis may be particularly useful for identifying types of individuals at risk for delayed or avoided care, whose personal needs may differ from the average socioeconomic conditions of typical residents^31,33 and, if persistently unaddressed, may increase the likelihood of escalating needs that are more intensive to meet, especially in already under resourced rural communities. It also supports a more complete understanding of constraints facing local community or clinic populations, specifically, which constraints are most prevalent, which cooccur to form locality-based patterns, and what proportions of individuals experience low, moderate, or high cumulative access or care-use burden.

Such insights can help both community health systems and local service providers identify the extent of overlapping constraints, as well as the specific combinations potentially contributing to missed appointments or delayed care, and thereby prioritize more relevant, multi-domain solutions or support for these groups. At the community level, care models might be adapted to better leverage local infrastructure (eg, co-located health and social services, mobile clinics, community care teams, expanded community-systems partnerships) and encourage more precise matching of patients to bundled supports rather than generic referrals.^33,34,36 In this context, simultaneous efforts to strengthen local affordability supports, streamline care navigation, and better coordinate service and follow-up referrals may further reduce cumulative barriers,³⁴ especially in rural communities with provider or service shortages^37-39 where improved coordination may enhance the efficiency and responsiveness of existing infrastructure.

At the practice level, these insights could translate into concrete strategies to expand access to patient groups experiencing high cumulative burden. Depending on profiles of needs, strategies might include expanded clinic hours, flexible or same-day scheduling options, team-based staffing models centralizing advanced practice providers and care coordination teams, local transportation partnerships, and/or service outreach through mobile units to community public services sites (eg, WIC clinics, schools, libraries).⁴⁰ These types of expanded services or availability options might help mitigate non-insurance-related cost burdens, such as lost wages or transportation expenses associated with attending appointments, that compound access challenges in rural settings.^5,40 Additionally, as provider communication and trust-building interventions have been shown to improve adherence to recommended care,⁴¹ including in the context of cost-related nonadherence,⁴² incorporating a profile-informed understanding of patients’ access constraints could enhance these strategies by helping to tailor communication and recommendations to their specific circumstances.

Ultimately, the findings suggest that integrating individual-level needs mapping into rural health planning may help clarify intersecting factors shaping differences in care seeking³¹ and support more targeted local service delivery at different entry points, as reflected in successful multilevel, socioecological programs currently implemented in many rural and underserved communities.^43-45

Strengths and Limitations

This study offers insights drawn from a localized context not easily captured in larger population-level or aggregated datasets.⁴⁶ Nonetheless, several limitations must be considered. The cross-sectional design and non-random sampling limit causal inference. The modest sample size also constrained analytic power and limited variable inclusion. The use of composite variables may have reduced insight into within-group variability, although this is a known trade-off in clustering approaches.⁴⁷ Future work could address these limitations, for example, by redeploying the survey to increase sample size, using qualitative methods to examine the lived experiences of women from each profile type to add further contextual richness to understanding their care access, or applying longitudinal designs to assess whether profiles analyzed at a population-level might predict behavioral outcomes and health trajectories over time.

Conclusion

This study identified 7 distinct profiles among rural Texas women, reflecting variation in health burden, health-related social needs, and geographic care availability. Profiles with multiple co-occurring constraints had significantly lower odds of receiving care when needed, underscoring that combinations of non-medical social and geographic factors shape care seeking beyond demographics or health status alone. In contrast, a profile with high health burden but fewer non-medical factors demonstrated relatively good access, highlighting the protective roles of geographic proximity and reduced resource strain. The profile-based analytic approach illustrated here may be useful to apply in other settings to better characterize co-occurring needs of local patient populations, generate hypotheses about the implications of delayed or forgone care in specific groups, evaluate the impact of intervention programs on care seeking that address multiple non-medical drivers of health, and inform future service planning at practice and community levels to support earlier engagement with care in rural health systems.

Supplemental Material

sj-docx-1-jpc-10.1177_21501319261432462 – Supplemental material for Profiles of Non-medical Drivers and Health Burden Associated With Care Seeking Among Rural Women

Supplemental material, sj-docx-1-jpc-10.1177_21501319261432462 for Profiles of Non-medical Drivers and Health Burden Associated With Care Seeking Among Rural Women by Elizabeth Ekren, Saksham Adhikari, Shadi Maleki, Jessica K. Sexton, Christina Aubert, Maria Tomasso, Russell Hopkins and Melinda M. Villagran in Journal of Primary Care & Community Health

Footnotes

Acknowledgements

The authors acknowledge the support of Drs. Michelle Crum, Catherine Goetz, Ilia Pashintsev, and Samiksha Srivastava at the University of Texas at Tyler Health Science Center.

ORCID iDs

Elizabeth Ekren

Jessica K. Sexton

Ethical Considerations

This study was conducted under the ethical standards of the Texas State University Institutional Review Board (#9510, June 18, 2024).

Consent to Participate

All participants provided informed electronic consent through a secure survey platform prior to participation.

Consent for Publication

Not applicable.

Author Contributions

E.E. led project conceptualization, survey design and administration, statistical analysis and oversight, and manuscript drafting and supervision. S.A. conducted data cleaning and management, statistical analyses (including clustering code development), figure generation, and contributed to the methods section. S.M. contributed to survey design and administration, map production, and manuscript writing and editing. J.K.S. led development of the discussion section, provided subject-matter expertise, and contributed to manuscript editing. C.A. oversaw recruitment and data collection, supported survey administration and data management, and contributed to manuscript editing. M.T. reviewed the statistical analyses, refined analytic code, and contributed to manuscript editing. R.H. and M.M.V. contributed to project conceptualization, supervision, resource administration, and manuscript editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this research was provided by the Community Health and Economic Resilience Research Center of Excellence at Texas State University.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Participant data are not available due to participant privacy considerations. Python code is available from the authors upon reasonable request.

Supplemental Material

Supplemental material for this article is available online.

References

Al Rifai

Jain

Khan

, et al. State-level social vulnerability index and healthcare access: the behavioral risk factor surveillance system survey. Am J Prev Med. 2022;63(3):403-409.

Ashokan

Rajagopal

Krishnaswamy

Sheela

Marimuthu

Pothumani

DS.

Social vulnerability assessment in the health and disease context: review. Discov Soc Sci Health. 2024;4(1):55.

Butler

Petterson

Phillips

Bazemore

AW.

Measures of social deprivation that predict health care access and need within a rational area of primary care service delivery. Health Serv Res. 2013;48(2 Pt 1):539-559.

Anderson

Saman

Lipsky

Lutfiyya

MN.

A cross-sectional study on health differences between rural and non-rural U.S. counties using the County Health Rankings. BMC Health Serv Res. 2015;15:441.

Gong

Phillips

Hudson

Curti

Philips

BU.

Higher U.S. rural mortality rates linked to socioeconomic status, physician shortages, and lack of health insurance. Health Aff (Millwood). 2019;38(12):2003-2010.

U.S. Census Bureau. Nation’s urban and rural populations shift following 2020 Census. 2022. Accessed August 5, 2025. https://www.census.gov/newsroom/press-releases/2022/urban-rural-populations.html. Accessed August 15, 2024.

Ekren

Maleki

Curran

Watkins

Villagran

MM.

Health differences between rural and non-rural Texas counties based on 2023 County Health Rankings. BMC Health Serv Res. 2025;25(2):2.

MacDougall

Mork

Hanson

Smith

CH.

Rural-urban differences in health care unaffordability. J Rural Health. 2024;40(2):376-385.

U.T. Tyler Hibbs Institute. Hibbs Institute health index: Examining health conditions in the region (an update). 2023. Accessed August 5, 2025. https://www.uttyler.edu/academics/colleges-schools/business/centers/hibbs-institute/files/reports/2023/hibbs-institute-health-index-updated-2022-2023.pdf.

10.

Ziller

Lenardson

Paluso

Janis

Preventive Health Service Use Among Rural Women. PB-73: University of Southern Maine, Muskie School of Public Service, Maine Rural Health Research Center; 2019.

11.

Larson

Correa-de-Araujo

Preventive health examinations: a comparison along the rural-urban continuum. Women’s Health Issues. 2006;16(2):80-88.

12.

Zhan

Lin

Racial/Ethnic, socioeconomic, and geographic disparities of cervical cancer advanced-stage diagnosis in Texas. Women’s Health Issues. 2014;24(5):519-527.

13.

Cubbin

La Frinere-Sandoval

QNNB

Widen

EM.

Social inequities in cardiovascular disease risk factors at multiple levels persist among mothers in Texas. Int J Environ Res Public Health. 2025;22(3):404.

14.

Borders

Rohrer

Cardarelli

KM.

Gender-specific disparities in obesity. J Community Health. 2006;31(1):57-68.

15.

Texas Health and Human Services. Maternal and child health epidemiology: 2022/2023 healthy Texas mothers and babies data book. 2023. Accessed June 10, 2025. https://www.dshs.texas.gov/sites/default/files/mch/pdf/HTMB_Data_Book_2022-2023_revFeb2025.pdf.

16.

Krukowski

Jacobson

John

, et al. Correlates of early prenatal care access among U.S. women: Data from the Pregnancy Risk Assessment Monitoring System (PRAMS). Matern Child Health J. 2022;26(2):328-341.

17.

U.S. Department of Agriculture (USDA). USDA ERS - Documentation. 2024. Accessed August 5, 2024. https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/#:~:text=For%20Rural%2DUrban%20Continuum%20Codes,below%20250%2C000%20(code%203).

18.

Steyerberg

Eijkemans

Harrell

Habbema

JD.

Prognostic modeling with logistic regression analysis: in search of a sensible strategy in small data sets. Med Decis Making. 2001;21(1):45-56.

19.

Short

Goetzel

Pei

, et al. How accurate are self-reports? Analysis of self-reported health care utilization and absence when compared with administrative data. J Occup Environ Med. 2009;51(7):786-796.

20.

Vogeli

Shields

Lee

, et al. Multiple chronic conditions: prevalence, health consequences, and implications for quality, care management, and costs. J Gen Intern Med. 2007;22 Suppl 3:391-395.

21.

Mseke

Jessup

Barnett

Impact of distance and/or travel time on healthcare service access in rural and remote areas: a scoping review. J Transp Health. 2024;37:101819.

22.

Rose

Bonnell

Clifton

Natkin

Hitt

O’Rourke-Lavoie

Outcomes of delay of care after the onset of COVID-19 for patients managing multiple chronic conditions. JABFM. 2022;35(6):1081-1091.

23.

Sinha

Calfee

Delucchi

KL.

Practitioner’s guide to latent class analysis: methodological considerations and common pitfalls. Crit Care Med. 2021;49(1):e63-e79.

24.

Hennig

Meila

Murtagh

Rocci

Handbook of Cluster Analysis: Chapman and Hall/CRC; 2015.

25.

Singh

Daus

Allender

, et al. Social determinants of health in the United States: addressing major health inequality trends for the nation, 1935-2016. Int J MCH AIDS. 2017;6(2):139-164.

26.

Miller

KEM

James

Holmes

van Houtven

. The effect of rural hospital closures on emergency medical service response and transport times. Health Serv Res. 2020;55(2): 288-300.

27.

Taylor

Liu

T-L

Howell

EA.

Insurance differences in preventive care use and adverse birth outcomes among pregnant women in a Medicaid nonexpansion state: a retrospective cohort study. J Womens Health. 2020;29(1):29-37.

28.

Allen

Wright

Broffman

The impacts of Medicaid expansion on rural low-income adults: lessons from the Oregon health insurance experiment. Med Care Res Rev. 2018;75(3):354-383.

29.

Finkelstein

Taubman

Wright

, et al. The Oregon health insurance experiment: evidence from the first year. Q J Econ. 2012;127(3):1057-1106.

30.

Jenkins

Pho

Nance

, et al. Influence of insurance type on healthcare utilization among rural people who use drugs. Drug Alcohol Depend. 2025;269:112542.

31.

Brignone

LeJeune

Mihalko

Shannon

Sinoway

LI.

Self-reported social determinants of health and area-level social vulnerability. JAMA Netw Open. 2024;7(5):e2412109.

32.

Kullgren

McLaughlin

CG.

Beyond affordability: the impact of nonfinancial barriers on access for uninsured adults in three diverse communities. J Community Health. 2010; 35(3):240-248.

33.

Dev

Allen

Cass

Canavan

Velmineti

Goodman

DJ.

Characterizing intersecting social determinants of health during pregnancy: a descriptive cross-sectional analysis from a northern New England health system. Front Med. 2025;12:1658735.

34.

Eke

Aneesa

Hogans-Mathews

Speights

JB.

Community perceptions of the health equity navigators program: a cross-sectional study in rural and medically underserved areas of Georgia, USA. BMC Public Health. 2025;25(1):4212.

35.

Stringhini

Carmeli

Jokela

, et al. Socioeconomic status and the 25 × 25 risk factors as determinants of premature mortality: a multicohort study and meta-analysis of 1·7 million men and women. Lancet. 2017;389(10075):1229-1237.

36.

Gizaw

Astale

Kassie

GM.

What improves access to primary healthcare services in rural communities? A systematic review. BMC Prim Care. 2022;23(1):313.

37.

Goldfield

Fuller

RL.

Access to affordable, high-quality health insurance for rural residents and its impact on their health and on rural hospitals. J Ambul Care Manage. 2019; 42(2):78-85.

38.

Torres

Richman

Schreier

Vohra

Verbanac

An evaluation of a rural community-based breast education and navigation program: highlights and lessons learned. J Cancer Educ. 2019;34(2):277-284.

39.

Thompson

Rivelli

Schneider

, et al. Adaptations to a patient navigation program for follow-up colonoscopy in rural primary care practices. J Eval Clin Pract. 2024;30(7):1457-1466.

40.

Chen

, Access improvement in healthcare: a 12-step framework for operational practice. Front Health Serv. 2025;4:1487914.

41.

Zolnierek

Dimatteo

MR.

Physician communication and patient adherence to treatment: a meta-analysis. Med Care. 2009;47(8):826-834.

42.

Piette

Heisler

Krein

Kerr

EA.

The role of patient-physician trust in moderating medication nonadherence due to cost pressures. Arch Intern Med. 2005;165(15):1749-1755.

43.

Seguin-Fowler

Eldridge

Rethorst

, et al. Improvements and maintenance of clinical and functional measures among rural women: strong Hearts, Healthy Communities-2.0 cluster randomized trial. Circ Cardiovasc Qual Outcomes. 2022;15(11):e009333.

44.

Salihu

Wilson

King

Marty

Whiteman

VE.

Socio-ecological model as a framework for overcoming barriers and challenges in randomized control trials in minority and underserved communities. Int J MCH AIDS. 2015;3(1):85-95.

45.

Whitman

Lew N

Chappel

Aysola

Zuckerman

Sommers

. Addressing social determinants of health: Examples of successful evidence-based strategies and current federal efforts. 2022. Accessed July 1, 2025. https://aspe.hhs.gov/sites/default/files/documents/6ba4bbb2e9c9551355a6926f023f1585/SDOH-Evidence-Review.pdf

46.

Ziller

Milkowski

A century later: rural public health’s enduring challenges and opportunities. Am J Public Health. 2020;110(11):1678-1686.

47.

Wani

AA.

Comprehensive analysis of clustering algorithms: exploring limitations and innovative solutions. PeerJ Comput Sci. 2024;10:e2286.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.15 MB