Preferences for Video Consultations in Family Health Stations: A Discrete Choice Experiment Among Ultra-Orthodox Jewish Women

Abstract

Background

Video consultation in family health stations offers potential for improving healthcare access for cultural and religious minority populations who face unique challenges in healthcare utilization. Ultra-Orthodox Jewish women provide a case study for understanding preferences in hard-to-reach communities.

Objective

We aimed to elicit preferences for video consultations versus in-clinic consultations at family health stations among Ultra-Orthodox women, and to demonstrate the application of discrete choice experiment methodology in religious and culturally sensitive research.

Methods

A face-to-face, paper-based, discrete choice experiment was conducted. Following culturally adapted qualitative interviews and focus groups, four attributes were examined: consultation timing, waiting time, child presence requirement, and nurse familiarity. Participants evaluated 12 choice scenarios comparing video consultations, using rabbinically approved devices, with traditional in-clinic consultations. A mixed logit model analyzed preferences and heterogeneity across demographic subgroups.

Results

A total of 174 women participated, with a mean age of 30.0 ± 6.1 years and a mean of 3.45 ± 2.1 children. Waiting time was the strongest predictor of consultation preference (β = −1.50, p < 0.001), followed by consultation hours (β = 1.10, p < 0.001), provider familiarity (β = 0.55, p < 0.001), and child presence requirement (β = −0.51, p < 0.001). Consultation type (video vs in person) was not independently significant. Subgroup analyses showed that older women had weaker preferences for video consultation. Policy simulation estimated that optimized video consultation configurations could capture 58–78% of current in-person consultations.

Conclusions

This study demonstrates how discrete choice experiments can successfully elicit preferences in hard-to-reach cultural and religious communities. Service design must address specific population needs including scheduling flexibility, continuity of care, and family logistics. These findings inform culturally sensitive telemedicine implementation for diverse minority populations.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40271-026-00814-5.

Key Points for Decision Makers

Video consultation implementation in family health stations serving cultural and religious minority populations must prioritize reduced waiting times, evening appointment availability, continuity of care with familiar nurses, and consultations that do not require child presence to achieve optimal uptake rates.

Successful telemedicine adoption in closed communities requires culturally adapted technology solutions, including rabbinically-approved devices with restricted internet access. Policymakers must collaborate with religious leadership to ensure technology aligns with community values while maintaining healthcare quality and accessibility.

Implementing discrete choice experiments with hard-to-reach populations requires culturally sensitive methodologies including community-matched researchers, religious leadership approval, mixed-methods approaches, and adapted data collection tools. This framework can guide preference elicitation for healthcare innovations in diverse minority communities.

Introduction

Video consultation (VC) is a telemedicine technology that enables medical interactions between patients and healthcare providers through digital interaction using both image and voice, via either asynchronous (store-and-forward) communication, whereby feedback is delivered without real-time face-to-face contact, or synchronous (online) communication [1, 2]. As healthcare systems worldwide face increasing demands and resource constraints, VCs offer potential benefits in improving accessibility, also for underserved populations with unique cultural and religious needs [3, 4]. The adoption of digital health technologies among culturally and religiously distinct populations presents both opportunities and challenges that extend beyond technological infrastructure to encompass deeply held beliefs, community norms, and traditional practices that shape healthcare-seeking behaviors [5]. Understanding how different cultural and religious groups navigate the intersection of traditional values and modern healthcare technology has become increasingly important for health systems seeking to reduce disparities and improve access to care [6].

The integration of telemedicine services into primary care settings as family health stations presents a promising opportunity to enhance healthcare delivery to cultural and religious minority groups, such as Ultra-Orthodox (UO) Jewish women [7, 8]. While the UO Jewish community provides a unique case study, similar challenges in balancing religious observance with healthcare technology adoption have been documented across various faith-based communities worldwide, including conservative Christian, Muslim, and other religious groups where traditional values significantly influence health-related decision making [9, 10]. These communities often share common characteristics such as an emphasis on modesty, gender separation, community-based decision making, and concerns about digital exposure, making insights from one context potentially relevant for understanding technology adoption patterns in other religiously observant populations. The UO population often faces barriers to healthcare access, family size, transportation limitations, and specific religious practices that may complicate attendance at traditional in-clinic consultations (ICCs) [11, 12]. This study focuses specifically on women because they typically serve as primary health managers for their families; our previous research found that 85% of UO women reported being responsible for managing their children’s and family’s health matters [13]. Despite potential benefits, the implementation of VC technology in culturally sensitive healthcare settings remains complex and requires careful consideration of user preferences and needs [14].

Previous research has demonstrated that the successful adoption of telemedicine depends significantly on understanding key stakeholders’ preferences [15, 16]. Our earlier study examining VC preferences in Israel’s general population revealed important insights regarding factors influencing the choice between VCs and ICCs in primary care settings among key stakeholders: patients, primary care physicians, and policy makers, showing the importance of the time to the next available consultation, waiting times, familiarity with the physician, and the quality of the consultation itself [13, 17, 18]. However, these general population findings may not fully apply to cultural and religious minority groups whose healthcare needs and preferences may differ substantially from the general population [19, 20]. Research from diverse geographic and cultural contexts has consistently shown that preferences for healthcare delivery modalities can vary significantly based on the cultural background, religious observance level, and community integration, highlighting the necessity of population-specific preference elicitation rather than extrapolation from majority population studies [21, 22].

The current study focuses specifically on UO women’s preferences regarding VCs with nurses at family health stations as compared to traditional in-clinic visits. Family health stations (known as “Tipat Chalav”) provide preventive care, developmental assessments, vaccinations, and parental guidance for children aged 0–5 years, serving as the primary point of contact for maternal and child health in Israel. Israel operates a universal healthcare system under the National Health Insurance Law (1994), through which all residents are entitled to a defined basket of health services delivered by four competing non-profit Health Maintenance Organizations. Family health station services are funded by the Ministry of Health and provided free of charge to all residents. Telemedicine services, including VCs, have been integrated into all four Health Maintenance Organizations for over 7 years at no additional cost to patients, ensuring equitable access regardless of socioeconomic status [23]. For UO women, who often have larger families and face unique modesty and cultural considerations, the option of VCs could potentially address several barriers to care while respecting cultural sensitivities [23–25].

However, technology accessibility presents a significant consideration in this population. Many UO women do not own smartphones or have restricted access to general Internet services because of religious guidelines [26]. Instead, they typically access the Internet at work or home primarily for health-related purposes. As revealed in our previous research, UO women expressed willingness to participate in VCs only through rabbinically approved “kosher” devices—specialized communication devices that are certified by rabbinic authorities specifically for video calls and communication with healthcare providers, while blocking access to the general Internet [13, 27]. This phenomenon of religiously mediated technology adoption where religious authorities provide frameworks for acceptable use of modern technologies is not unique to the UO Jewish community and has been observed in other religiously conservative groups seeking to benefit from technological advances while maintaining religious standards [28].

Understanding the preferences of this population is crucial for designing telemedicine services that are both effective and culturally appropriate. To quantify these preferences, we employed a discrete choice experiment (DCE) methodology, which examines individual choice preferences among several options. This method has been widely used in health economics and policy-making research to provide a better understanding of people’s preferences regarding healthcare use [29, 30].

This study aims to identify the key factors that influence UO women’s decisions to choose VCs over in-clinic visits for family health services, and to quantify the relative importance of these factors. The findings will contribute to developing culturally sensitive telemedicine implementation strategies that can improve healthcare accessibility for this unique population while respecting their religious and cultural values.

Methods

A DCE is a methodology that examines individual choice preferences among several options. The present study framework followed recommended steps, including selecting key attributes and assigning levels to the attributes; developing an experimental design, scenarios, and surveys; administering surveys to elicit preferences; and analyzing data [27–30], and is reported in accordance with the DIRECT (Discrete Choice Experiments Reporting) checklist [31].

Identification of Attributes and Levels

Identifying key attributes and assigning levels to the attributes were done using qualitative research methods according to established guidelines [32–34], reported according to COREQ (Consolidated Criteria for Reporting Qualitative Research) [35] and published [13]. A three-stage mixed-methods approach was employed to ensure culturally sensitive and methodologically rigorous attribute identification. Stage 1 involved 33 semi-structured interviews with key stakeholders, including 22 UO women who had recently utilized family health station services, along with men, rabbis, and healthcare providers, to identify initial attributes and levels through a qualitative analysis. Stage 2 employed an attribute-ranking exercise with 88 UO women to refine and prioritize the identified attributes based on their importance to the target population. Stage 3 utilized cognitive interviews with 15 UO women to validate the DCE questionnaire, ensuring its cultural appropriateness, comprehensibility, and feasibility within this closed community. Throughout all stages, extensive cultural adaptations were implemented, including the use of female UO interviewers, religiously appropriate language, community leader endorsement, and careful attention to modesty requirements, demonstrating the cultural sensitivity essential for meaningful engagement with this population. Data analysis was performed inductively, using a thematic content analysis method, followed by a ranking exercise based on the results gleaned from the interviews and group discussions.

The analysis revealed four key attributes that influenced UO women’s preferences regarding VCs versus ICCs at family health stations: (1) consultation timing (regular hours/evening hours); (2) waiting time; (3) child presence requirement (consultation requires the child’s presence/no child presence required); and (4) familiarity with the nurse. Regarding waiting time levels, these were selected based on triangulation of women’s self-reported waiting times during preliminary interviews, which ranged from near-immediate entry to approximately 1 hour, and Israeli Ministry of Health published data on Health Maintenance Organization waiting times. The 5-minute level represents the minimum realistic wait, while 40 minutes captures the upper range of typical waiting times reported by women attending family health stations.

Regarding child presence, while certain consultations such as vaccinations and initial wellness visits require in-person attendance, VCs with the child present can be appropriate for specific scenarios, such as developmental assessment follow-ups or consultations where observation of the child in their natural home environment may facilitate cooperation and provide clinically valuable information. The DCE scenarios focused exclusively on consultation types amenable to video delivery, including postpartum follow-up consultations, developmental assessment completions, counseling sessions on age-appropriate topics (e.g., weaning, nutrition transitions), and postpartum depression monitoring. Consultations requiring physical procedures, such as vaccinations and initial physical examinations, were not included in the choice scenarios.

Importantly, a fundamental prerequisite for examining preferences was that the device used for VCs must be rabbinically approved and “kosher,” exclusively designated for healthcare provider communication while closed to open Internet networks. This condition was explicitly stated in the DCE questionnaire to ensure that all preferences elicited were based on this culturally acceptable technological framework. Table 1 presents the attributes, descriptions, levels, hypotheses, and labels for the regression.

Table 1.

Attributes and levels

Attribute	Description	Level	Hypothesisa
Consultation timing (HOUR)	Hours when family health station services are available	Regular hours (0), evening hours (1)	+
Queuing time before consultation (WAIT)	Typical waiting time before the consultation	5 minutes (0), 40 minutes (1)	−
Child’s presence at visit (CHILD)	Whether the child’s presence is required for the visit (e.g., development assessment) or not required (e.g., guidance on nutrition)	Not required (0), Required (1)	?
Familiarity with the nurse (KNOW)	Whether the consultation is with a nurse the parent knows or an unfamiliar nurse	Unfamiliar nurse (0), Familiar nurse (1)	+

^aThe Hypothesis column indicates the expected direction of preference: (+) higher attribute level expected to increase preference for video consultation, (−) higher attribute level expected to decrease preference for video consultation, (?) direction uncertain a priori

Experimental Design

A labeled-choice experiment was employed to compare preferences between VC and ICC services. This approach was chosen as it tends to provide more realistic and feasible scenarios, thereby enhancing the validity of findings [30, 33]. Given that exposing respondents to all 32 (2⁵) possible scenarios in a complete factorial design would be excessively demanding, we implemented a fractional factorial design. This design was constructed using manual catalog-based methods to ensure an optimal level balance (where each attribute level appears with equal frequency) and orthogonality (ensuring independence between levels of different attributes). Through this process, we generated 12 hypothetical binary choice tasks, each requiring a forced choice response. In each task, participants were asked to indicate their preferred consultation method between the two options presented (VC versus ICC). An opt-out option was not included because attendance at maternal and child health consultations is a universal entitlement and a practical necessity within Israel’s healthcare system; non-participation is therefore not a realistic choice for the target population.

Survey

The DCE survey was structured into three sections. The first section included DCE choice tasks and one dominant option set designed to assess the internal consistency of responses [32]; respondents who failed this check were excluded from the analysis. No additional internal validity tests were employed, though the questionnaire underwent cognitive interviews prior to administration to ensure clarity and comprehension of the choice tasks. The second section comprised sociodemographic questions, including age, number of children, religious stream affiliation (Lithuanian, Hasidic, or Sephardic), city of residence, education and income levels, self-reported level of modernity, and prior technology experience. The third section contained questions regarding various dimensions of trust in technology and healthcare system; and specific religious needs related to technology use, which are beyond the scope of this paper.

Following best-practice recommendations for DCE development, the survey underwent cognitive interviews with six UO women ,[34] followed by pilot testing with 15 participants. Based on participant feedback, several important cultural and methodological adaptations were implemented.

Regarding cultural-religious adaptations, all survey images were selected to meet community modesty standards, and religious terminology was incorporated naturally into the survey language (e.g., “kosher device” rather than “approved device”). To facilitate understanding, we used religiously familiar analogies, such as comparing the decision-making process to selecting an citrus (etrog) for Sukkot, which involves weighing multiple quality attributes [13].

Methodological refinements included using distinct colors for the two alternatives in each choice set to improve visual clarity, simplifying question wording based on comprehension feedback, expanding instructions to explain the trade-off concept, and revising the layout to reduce cognitive burden while maintaining validity. The questionnaire translated from Hebrew to English is attached in the Electronic Supplementary Material (ESM), and an example of a choice task from the survey is shown in Fig. 1. The study received approval from the University of Haifa Institutional Ethics Committeeת Approval No. 069/24, dated 9 September, 2024.

Fig. 1.

Example of a discrete choice experiment question in the questionnaire

Sample Size

According to established methods for calculating minimum required samples in DCE studies [34, 36], sample size depends on the number of scenarios (t), the number of alternatives (a), and the number of attributes including labeled attributes (c). With t = 12, a = 2, and c = 5 in the present study, the minimum sample size required was 104 participants. In practice, we collected data from a larger number of women to ensure an adequate sample size after excluding respondents who answered the dominant set incorrectly.

Survey Administration

Recruitment of UO Jewish women required culturally sensitive approaches. A research assistant who was member of the UO community was trained to facilitate survey distribution and completion. The survey was administered in paper format rather than electronically, in keeping with cultural preferences and limited Internet access among this population.

A purposive sampling approach with snowball elements was employed, targeting women across different geographic regions and UO religious streams (Lithuanian, Hasidic, and Sephardic) to ensure diverse representation of the study population. Participants were recruited in women’s gatherings in neighborhoods with high concentrations of UO residents across five cities in Israel. Eligibility criteria included women who self-identified as UO, had children aged under 18 years, and had visited a family health station at least once in the previous year. Research assistants explained the purpose of the study and asked the women to complete the questionnaires. They remained available to assist with questions while participants completed the surveys (taking approximately 18–25 minutes). Special attention was given to ensuring the linguistic and cultural appropriateness of survey administration, including the use of appropriate terminology and respect for modesty concerns.

Econometric Model

A preference function was estimated based on the relative change in preferences associated with the differences in attribute levels for each choice, assuming that, ceteris paribus, individuals choose the alternative that maximizes their utility [32, 37, 38]. According to random utility theory [36], true but not totally observable utility of the individuals can be estimated by generalized prediction models (RUT models), which incorporate uncertainty and can be described in Equation 1 below, where DV (Delta V) is the change in preference associated with moving from ICC to VC, α_ICC - α_VC is the constant in the empirical model, β represents the utility parameter of the attributes HOURS, WAIT, CHILD, and KNOW, and ε and μ are the unobservable error terms where ε is due to differences in respondents’ choices and μ is due to differences among respondents. The choice attributes model is:

$$\text{DV}={\text{V}}_{\text{ICC}}-{\text{V}}_{\text{VC}}=\left({\alpha }_{\text{ICC}}-{\alpha }_{\text{VC}}\right)+{\beta }_{\text{HOURS}}+{\beta }_{\text{WAIT}}+{\beta }_{\text{CHILD}}+{\beta }_{\text{KNOW}}+\epsilon +\mu$$ 1

Data Analysis

Nine participants provided incorrect responses to the dominant choice set (number seven), featuring one clearly superior alternative, indicating inconsistent decision-making patterns. These respondents were subsequently removed from the analytical sample; however, later robustness checks confirmed consistent results with and without these respondents.

To estimate the choice model and specify utility functions, we employed a mixed logit model, alternatively termed a random effect logit mode. This approach assumes independent and identically distributed error terms in Equation 1. Statistical analyses were conducted using R version 3.5.2 (R Foundation for Statistical Computing, Vienna, Austria) [39] with the lme4 package (glmer function; McMaster University, Hamilton, ON, Canada) [40]. The model incorporated all variables, including sociodemographic characteristics. All attributes were analyzed as variables with two levels each, using effects coding (−1, 0, 1) to estimate the impact of moving from one level to the other within each attribute.

To verify the stability of our findings, we performed multiple robustness checks examining: (1) potential heterogeneity across different subgroups within the UO population, comparing preference patterns among women from various ages, educations, and occupations and (2) the influence of excluding inconsistent respondents, contrasting the model estimates with and without the nine women who failed the dominant choice set validation.

Policy Analysis

We conducted a policy analysis by estimating the probability of VC adoption. Utility coefficients were transformed into probabilities using predefined attribute levels. The probability of VC uptake [Pc (VC)] was derived through an indirect utility analysis, applying Equation 2, whereby an individual (n) selects VC from a choice set containing j alternatives (j = 1, ..., J):

$$\text{Pc}\left(\text{VC}\right)=\text{exp}\left(\text{Vn}\right)/\mathrm{\Sigma }\text{j exp}\left(\text{Vjn}\right)$$ 2

Our analysis incorporated assumptions reflecting realistic scenarios for VC implementation in family health stations serving the UO community. These assumptions were validated by a group of senior policymakers from Israel’s Ministry of Health including co-investigators in this study (BM, SH), and were used to parameterize an Excel-based simulation model that applied the mixed logit coefficients to estimate uptake probabilities under each scenario using the logistic transformation of the linear combination of applicable coefficients. The baseline model assumes that VC, once introduced as an alternative to in-person consultations in family health stations, will be implemented with the following parameters: 70% of VC appointments available during regular clinic hours and 30% during after-hours; waiting time will be 15 minutes on average; 80% nurse familiarity; and child presence will be required in 70% of consultations. These parameters reflect current operational patterns and anticipated initial-phase VC implementation constraints in this population. A detailed rationale for each parameter is provided in the ESM. The results were stratified for two additional options: that VS will be implemented only for consultation in which child presence is not required, with a familiar nurse from a local family health station or with a nurse from a call center. To assess the range of potential VC adoption, we calculated uptake probabilities for both optimal and suboptimal scenarios based on women’s stated preferences. The best-case scenario incorporated the preferred levels across all attributes, while the worst-case scenario reflected the least preferred combination of attribute levels.

Results

UO Women’s Characteristics

A total of 178 women fully completed the DCE questionnaire. Response rates were estimated at 60–65% based on the number of women initially approached. The sample characteristics align with national statistics for UO women in Israel regarding education, employment, geographic distribution, and sectoral affiliation (Lithuanian, Hasidic, and Sephardic streams) [39]. Respondent characteristics and the distribution of choices between alternatives are displayed in Table 2, while model estimation results are presented in Table 3.

Table 2.

Ultra-Orthodox women’s characteristics (N = 178^a)

Characteristic	n (%) or mean ± SD
Age (years), mean ± SD	30.04 ± 6.1
Number of children, mean ± SD	3.45 ± 2.1
Religious stream
Lithuanian	44 (25%)
Hasidic	53 (29%)
Sephardic	55 (30%)
Returnees to religion (Baalei Teshuva)	16 (9%)
Type of settlement
Ultra-Orthodox settlement	60 (34%)
Mixed settlement	47 (26%)
Secular settlement	67 (37%)
Education level
No certificate	7 (0.4%)
External exams, Saald exams	22 (12%)
Vocational training certificate	57 (32%)
Full matriculation certificate	32 (18%)
Academic degrees	40 (23%)
Employment status
Employed	117 (66%)
Self-employed	29 (16%)
Maternity leave	12 (0.7%)
Homemaker	9 (0.5%)
Work environment (if employed)
Mainly Ultra-Orthodox colleagues	56 (31%)
Mixed environment	72 (40%)
Mainly secular colleagues	43 (24%)
Spouse’s occupation
Kollel student	79 (44%)
Employed	42 (25%)
Kolle student and employed	49 (27%)
Family modernity level (1–10), mean ± SD	3.99 ± 2.3
Mobile phone ownership
No mobile phone	3 (0.02%)
Kosher non-smartphone	57 (32%)
Kosher/filtered smartphone	42 (26%)
Regular non-kosher smartphone	24 (13%)
Both kosher and non-kosher phones	27 (15%)
Internet use type
Regular Internet at home	20 (11%)
Regular Internet at work	18 (10%)
Kosher Internet with minimal blocking (Rimon)	28 (16%)
Kosher Internet with strict filtering (Etrog/Nativ)	86 (48%)
No Internet use	21 (12%)
Perception of Ultra-Orthodox society change
Becoming more open	126 (70%)
No change	29 (16%)
Becoming more Haredi	11 (0.06%)
Monthly household income (NIS)
Less than 2500	3 (1.7%)
2501–4000	7 (3.9%)
4001–6000	20 (11.2%)
6001–8000	26 (14.6%)
8001–10,000	33 (18.5%)
10,001–12,000	33 (18.5%)
12,001–15,000	28 (15.7%)
15,001–20,000	16 (9.0%)
More than 20,000	12 (6.7%)

NIS , SD standard deviation

^aPercentages are based on the number of respondents who answered each item and may not sum to 100% because of non-responses. Work environment is reported among employed and self-employed respondents only (n = 146). Mobile phone ownership percentages do not sum to 100% as participants could report owning more than one type of device

Table 3.

Estimation results for Ultra-

Attributes	RELM coefficient (SE) [95% CI]	RELM+SD coefficient (SE) [95% CI]	Rank order
Service attributes
Type of consultation (in-person vs VC)	0.1526 (0.1672) [−0.178, 0.483]	0.2869 (0.6041) [−0.907, 1.481]	–
Consultation timing (HOUR)	1.0976*** (0.1342) [0.834, 1.361]	1.1727*** (0.1405) [0.897, 1.448]	2
Queuing time before consultation (WAIT)	−1.5027*** (0.1337) [−1.765, −1.241]	− 1.6022*** (0.1413) [−1.879, −1.325]	1
Child’s presence at visit (CHILD)	−0.5066*** (0.1140) [−0.730, −0.283]	− 0.5371*** (0.1181) [−0.769, −0.305]	4
Familiarity with the nurse (KNOW)	0.5518*** (0.1204) [0.316, 0.788]	0.5846*** )0.1248) [0.340, 0.829]	3
Sociodemographic characteristics
Age	–	−0.0174* (0.0215) [−0.025, 0.059]	–
Number of children	–	−0.1099** (0.0828) [−0.272, 0.052]	–
Education level	–	0.0047 (0.0628) [−0.075, 0.171]	–
Employment status	–	−1.5873* (0.2918) [−1.159, −0.014]	–
Model fit statistics
Log-likelihood	−1456.84	−1429.44	–
Model comparison
Likelihood-ratio test, D (df)	–	54.8*** (14)	–

CI confidence interval, RELM random effects logit model, RELM+SE random effects logit model with socio-demographic variables, SD standard deviation, SE standard error, VC video consultation, *statistically significant at 0.05 level; **statistically significant at 0.01 level; ***statistically significant at 0.001 level

^9 women, choice set (no. 7 set) with one superior option; therefore, their choices were considered not to be consistent observations and thus were excluded from the analysis

Estimation Results

Ultra-Orthodox women exhibited an non-significant preference for shifting from ICCs to VCs (α1 = 0.15, P,.47). All attributes considered in the experiment were important for UO women when choosing between ICCs and VCs. The attribute “waiting time before the consultation” was judged the most important attribute for UO women (β = −1.5027, P,.001), the shorter the time until the next available appointment, the greater the chance they would choose this type of consultation as their preferred option. Women chose “timing of consultation” as a second important attribute for consultation choice (β = 1.0976, P,.001). “Familiarity with nurse” ranked as the third important attribute, and proved to be a significantly important parameter for UO women (β = 0.5518, P,.001), who preferred to be familiar with the nurse. Whether child presence was required emerged as the fourth most significant factor (β = −0.5066, p < 0.001), indicating that women preferred VCs that did not require the child to be present. When sociodemographic characteristics were incorporated into the model, three additional factors were significantly associated with VC preference: older age (β = −0.0174, p < 0.05), a greater number of children (β = −0.1099, p < 0.01), and current employment (β = −0.5873, p < 0.05), all of which were negatively associated with preference for VCs.

For parsimony, Table 4 presents only the interaction effects between attributes and demographic characteristics. Examining age-related heterogeneity, older UO women demonstrated a significantly stronger negative response to consultations requiring a child’s presence (β = −0.065, p < 0.001) [Model 1]. Higher educated women exhibited a reduced preference for both child-accompanied consultations (β = −0.163, p < 0.05) and consultations with familiar nurses (β = −0.175, p < 0.05) [Model 2].

Table 4.

Assessment of preference heterogeneity through interaction-effects models

Attribute interactions	Age (Model 1)		Education (Model 2)
	β (SE)	P-value	β (SE)	P-value
Extended consultation hours
Available	0.003 (0.023)	0.904	−0.062 (0.091)	0.494
Wait time
40 minutes	0.023 (0.023)	0.325	0.006 (0.090)	0.944
Child presence required
Required	−0.065* (0.019)**	<0.001	−0.163 (0.078)*	0.036
Provider familiarity
Unfamiliar provider	−0.014 (0.021)	0.488	−0.175 (0.082)*	0.032
Demographic variable	0.056 (0.033)	0.086	0.274* (0.115)	0.017
Model fit statistics
Log likelihood	−968.32		−971.15
Participants	145		145
Observations	1724		1724

SE standard error, *p < 0.05; **p < 0.01; ***p < 0.001

Age: continuous variable (years); education: per additional category of education level, where the categories were no certificate, external exams and Sald exams, vocational training certificate, full matriculation certificate, and academic degrees

Policy Analysis

Table 5 presents the probability of VC adoption under various implementation scenarios. The utility model indicated that about 58% of ICCs could shift to video format under baseline conditions. Ultra-Orthodox women demonstrated significantly higher VC preference when consultations met specific criteria aligned with their needs and constraints.

Table 5.

Probability of Ultra-Orthodox women uptake

	Probability of uptake	Option 1, VC for no child visits with familiar nurse	Option 2, VC no child visits with call center	Best scenario	Worst scenario
Likelihood of choice of VC instead of ICC	58%	69%	55%	78%	22%

ICC in-clinic consultation, VC video consultation

Attribute-level assumptions: α = VC, β1 = evening consultation at 30%, β2 = 15 minutes, β3 = with child in 70% of visits, β4 = 80% familiarity with the nurse; women’s uptake stratified for regular vs evening hours of consultation: option 1: β1 = 70% regular hours, β2 = 15-minute wait, β3 = consultation where no child presence required; β4 = familiar nurse only; option 2: β1 = 70% regular hours, β2 = 15-minute wait, β3 = consultation where no child presence required; β4 = call center nurse

The probability of VC adoption increased substantially when appointments were scheduled mostly during evening hours and did not require child presence—circumstances that eliminate two major barriers for this population. When these conditions were combined with consultation by a familiar nurse, the predicted uptake reached 69%. However, when consultations were conducted by an unfamiliar nurse from a call center, the adoption probability decreased to 55%, underscoring the importance of relational continuity in this community. Simulation analyses examined best-case and worst-case implementation scenarios to establish the range of potential adoption rates. Under optimal conditions—evening appointments, no child presence required, and familiar nurse provision—adoption rates reached 78%. Conversely, worst-case scenarios featuring unfavorable combinations of these attributes yielded a minimal uptake at 22%.

Discussion

This study represents the first investigation of VC preferences among UO women accessing maternal and child health services through family health stations. Beyond contributing empirical evidence about VC preferences in this population, the study demonstrates how DCE methodology can be adapted when working with cultural and religious minority populations. The methodological approach and findings have broader implications for understanding technology-mediated healthcare delivery among diverse faith-based communities worldwide, where traditional values intersect with modern healthcare innovations. The findings have implications not only for VC implementation in UO communities but also for understanding how healthcare preferences are embedded within distinct cultural frameworks that shape healthcare-seeking behaviors in ways that differ substantially from mainstream populations, especially related to technology usage.

Before discussing our findings, it is important to address a notable non-finding that points to fundamental differences in healthcare access preferences between UO women and the general population. Two conceptually distinct waiting-time attributes are relevant in this context: time until appointment, referring to the number of days until the next available slot, and queuing time before consultation, referring to the waiting time in the clinic immediately before the visit begins. Time until appointment has consistently been identified as a key driver of telemedicine preferences in international studies [41, 42], and it emerged as the most important attribute in a DCE conducted among the general Israeli population [13]. Remarkably, this attribute was not significant in our UO sample and was therefore not included in the present DCE. This absence likely reflects a distinctive pattern of healthcare access in UO communities. Whereas medical encounters in the general population are typically organized through scheduled appointments, reflecting the bureaucratic coordination of time characteristic of modern healthcare systems designed to manage large numbers of otherwise anonymous patients, UO women frequently arrive at primary care settings without prior scheduling. This walk-in culture is rooted in the collectivist nature of UO society, where informal access to services is normalized [43], consistent with Tönnies concept of Gemeinschaft, in which social relationships and shared norms supersede formal institutional procedures [44].

What did emerge as critical were attributes directly related to time management within visits themselves: waiting time in line and evening consultation hours. The strong preferences for a reduced waiting time and evening service’ availability reflect the unique circumstances of UO women, who typically work full-time as primary breadwinners, manage large households, and bear near-exclusive responsibility for family health management [45]. Research demonstrates that household and childcare activities in UO families intensify during evening hours—“life begins at night” when children return from educational institutions and family routines commence [12, 46]. Only after children are settled for bed can mothers address healthcare needs. Similar patterns of competing domestic and healthcare responsibilities have been documented among other religious minority women globally, where traditional gender roles and large family structures create distinct healthcare access barriers that differ from those identified in majority populations [9, 10]. The preference for evening appointments is particularly significant given that currently no family health stations offer evening services, suggesting a substantial unmet need. Video consultations could provide a practical solution, allowing nurses to offer remote evening services without requiring physical facility operation during these hours.

Familiarity with the healthcare provider emerged as the third most important attribute. The preference for relational continuity is well documented in the primary care literature [47, 48]. In the UO context, this preference reflects not merely personal preference but cultural values around tzniut (modesty) and the importance of trust relationships in communities where healthcare providers are often neighbors and community members. The importance of provider familiarity suggests that VC implementation models featuring call center nurses unknown to patients would face significant acceptance barriers, whereas models maintaining continuity with family health station nurses would be more readily adopted.

The requirement for child presence during consultation emerged as the last ranked significant factor, with women preferring consultations that do not require bringing a child. Given large family sizes and limited access to external childcare, VCs that eliminate the need to transport children could substantially reduce the logistical burden. Additionally, many UO parents express concerns about exposing children to technology, which remains viewed as taboo in many religious circles and is perceived as potentially compromising children’s spiritual and religious protection. This cultural consideration adds another dimension to the preference for consultations without child presence, beyond the practical logistical challenges.

Ultra-Orthodox women exhibited a non-significant overall preference for shifting from ICCs to VCs, suggesting that women view both modalities as viable options depending on specific service characteristics. This neutral stance may reflect competing considerations: a VC offers practical benefits such as reduced travel and childcare flexibility, while ICC visits provide valued social interaction and preserve the comfort of traditional healthcare access without requiring personal technology use, which can be stigmatized and is not always rabbinically sanctioned in this community.

Our heterogeneity analysis revealed that preferences varied across demographic subgroups. Older women demonstrated significantly stronger negative responses to consultations requiring child presence, suggesting that as women age and manage more children simultaneously, the burden becomes increasingly prohibitive. Education level emerged as a significant moderator—higher educated women exhibited a reduced preference for both child-accompanied consultations and consultations with familiar nurses, possibly reflecting greater comfort navigating healthcare systems and communicating with unfamiliar providers. These findings suggest that VC implementation strategies should emphasize maintaining nurse continuity particularly for less educated women.

The substantial preference for familiar nurses has important implications for implementation models. Rather than centralized call center approaches common in some telemedicine systems, our findings suggest that VC services for UO women should maintain connections to local affiliated family health stations, with consultations conducted by the same nurse women would see in person. This “extended clinic” model appears most aligned with this population’s preferences and cultural values. The requirement for rabbinically approved “kosher” devices represents an additional implementation consideration requiring collaboration with religious authorities.

This study has several limitations. First, the sample size of 178 women, while adequate for a DCE analysis, limits extensive subgroup analyses. Second, willingness to pay was not included as an attribute because family health station services are provided free of charge through Israel’s public healthcare system, making cost considerations irrelevant to women’s actual consultation choices. Moreover, VCs have been implemented in Israel’s public healthcare system for over 7 years across all health maintenance organizations and some hospitals departments, provided at no cost to patients; thus, it is unlikely that future implementation for the UO population would involve payment. Future research should consider travel and opportunity costs. Finally, the study examined non-urgent consultations; preferences may differ substantially for urgent care situations.

Future research should examine preferences for other telemedicine modalities beyond synchronous VCs, such as asynchronous messaging or artificial intelligence-assisted triage tools, which may offer different accessibility advantages for UO communities. Longitudinal studies are needed to explore how preferences for digital health services may evolve as telemedicine becomes more integrated into the daily healthcare experience of UO families. Research should also examine VC preferences in other healthcare contexts relevant to this population, such as mental health services or specialist referrals, where cultural sensitivities may similarly shape technology adoption.

Conclusions

This study provides the first comprehensive examination of UO women’s preferences for VCs in maternal and child health services. While women showed no strong overall preference for either modality, specific service characteristics, particularly waiting time, appointment timing, nurse familiarity, and child presence requirement, significantly influenced their choices. The findings demonstrate that culturally appropriate VC implementation is feasible for this population, provided that services are designed to reflect their preferences and constraints. The absence of appointment scheduling as a significant factor, in contrast to general population studies, reveals fundamental differences in healthcare preferences between collectivist and individualist cultural frameworks. Implementation plans should prioritize evening appointment availability, minimize child presence requirements, maintain nurse continuity, and reduce waiting times to optimize adoption. By understanding and addressing the unique preferences of UO women, healthcare systems can design telemedicine services that enhance accessibility while respecting cultural values and practical constraints, offering a model for adapting digital health innovations to serve cultural minority populations.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open access funding provided by University of Haifa. This research was funded by The Israel National Institute for Health Policy Research (Grant number 2023104). Its publication is not contingent upon the sponsor’s approval. The authors declare that no funds, grants, or other support were received during the preparation of this article.

Declarations

Conflicts of Interest

Irit Chudner, Anat Drach-Zahavy, Tamar Almor, Batia Madjar, Leah Gelman, and Sonia Habib have no conflicts of interest that are directly relevant to the content of this article.

Ethics Approval

All procedures followed were in accordance with the ethical approval for the study granted by the Haifa University Ethics Committee.

Consent to Participate

Informed consent was obtained from all responders included in the study.

Consent for Publication

Not applicable.

Availability of Data and Material

Not applicable.

Code Availability

Not applicable.

Authors’ Contributions

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by IC, AD-Z, and LG. BM and SH contributed to conclusions and recommendations development and facilitated access to key stakeholders and rabbinical authorities in this closed community, enabling the recruitment of participants. The first draft of the manuscript was written by IC, and all authors commented on previous versions of the manuscript. All authors read and approved of the final manuscript.