Patient Preferences for Out-of-Hospital Cardiac Arrest Care in South Africa: A Discrete Choice Experiment

Abstract

Objective:

This study examined the trade-offs low-resource setting community members were willing to make in regard to out-of-hospital cardiac arrest care using a discrete choice experiment survey.

Methods:

We administered a discrete choice experiment survey to a sample of community members 18 years or older across South Africa between April and May 2022. Participants were presented with 18 paired choice tasks comprised of 5 attributes (distance to closest adequate facility, provider of care, response time, chances of survival, and transport cost) and a range of 3 to 5 levels. We used mixed logit models to evaluate respondents’ preferences for selected attributes.

Results:

Analyses were based on 2228 responses and 40 104 choice tasks. Patients valued care with the shortest response time, delivered by the highest qualified individuals, which placed them within the shortest distance of an adequate facility, gave them the highest chance of survival, and costed the least. In addition, patients preferred care delivered by their family members over care delivered by the lay public. The highest mean willingness-to-pay for increased survival is 11 699 South African rand (ZAR), followed by distance to health facility (8108 ZAR), and response time (5678 ZAR), and the lowest for increasing specialization of provider (1287 ZAR).

Conclusions:

In low-resource settings, it may align with patients’ preference to include targeted resuscitation training for family members of individuals with high-risk for cardiac arrest as a part of out-of-hospital cardiac arrest intervention strategies.

Introduction

Out-of-hospital cardiac arrest (OHCA) is a leading cause of death worldwide.¹ OHCA survival rates are typically between 6% and 11% in high-income countries and survival with neurological functional recovery is estimated to be less than 10%.² The variation in OHCA survival is potentially driven by heterogeneity in socioeconomic, sociodemographic, and geographic factors.^3–6

Emergency care (EC) is the cornerstone of OHCA management that may improve survival rates. Unfortunately, in many low-resource settings (LRS)—including South Africa—EC is often delayed or not available, resulting in very low rates of recovery. Available studies indicate dramatically worse survival rates where OHCA chains of care are underdeveloped or absent.^7–10 In addition, many LRS experience low availability of post-resuscitation care, further worsening outcomes. Given resource limitations and lack of investment into the development of EC systems in LRS, resource allocations should focus on supporting effective interventions with wide reaching impacts and ameliorating concerning rates of morbidity and mortality. Examples of these interventions could be maternal, neonatal, and trauma emergency interventions.

Current resuscitation practice guidelines use elements of scientific evidence for OHCA decision making but little is known regarding community members’ preferences in care. Individuals’ psychological and sociocultural views on end of life, resuscitation, and preference on EC in LRS may differ from those held by individuals residing in high-income settings, where most guidelines are developed. As such, it is critical to explore further patient and provider expectations, to meet economic and structural challenges, and to address cultural differences in patient and provider expectations in LRS. Understanding patient preferences may assist in the formulation of feasible, acceptable, and appropriate policies for the setting.

Investigations into community views on death and dying, OHCA, and access to care in prehospital and African country settings have already begun (E. Theron et al, unpublished data, 2024),¹¹ and to the best of our knowledge, no systematic quantification of preference for OHCA in LRS exists yet. Taking into consideration the preference of individuals from relevant settings when formulating protocols and guidelines is one of the first steps to decolonizing global health and enhancing diversity, equity, and inclusion. This study contributes to the incorporation of preference of individuals from an LRS by leveraging existing findings and constructing a discrete choice experiment (DCE) survey to examine the trade-offs LRS community members are willing to make in regard to OHCA. We comprehensively assessed the strength of preferences among communities for attributes of OHCA management in an LRS and elicit how community members weigh the trade-offs between OHCA management options. The findings may assist policy makers and service providers in optimizing limited resources for health.

Methods

We administered a DCE survey to a sample of community members 18 years or older from across South Africa between April and May 2022. Responses to the survey were used to quantify the importance of attributes associated with OHCA care.

Survey Development

We determined the attributes and level in the DCE through extensive literature searches and in-depth interviews to identify the most important aspects of OHCA care delivery in LRS. First, a review of published articles was conducted to identify and assess all available data on OHCA treatment features in LRS and ensure that levels are plausible and clinically relevant to the setting.¹² All 5 attributes were identified via literature review. Second, we used preliminary results from a separate, qualitative study conducted in the Cape Town metropolitan area to understand perceptions on death and dying, OHCA, and access to care. Respondents of this qualitative study were asked to describe situations in which they confront death or dying. Topics of violence, fear, access, and a sense of futility emerged throughout these interviews. The results of the interviews did not differ from attributes identified in the literature review. However, themes derived from the interim narrative analysis of these results were used to refine the levels for provider of care of our survey.

We hypothesized that priority healthcare service dimensions (or attributes) and associated characteristic (or attribute levels) are cost, anticipated survival, provider of care, and timing of care in the form of response time, and distance to the closest facility. A list of the full attributes and their levels selected for the survey is presented in Table 1. Prices published as part of the South African Road Accident Fund Medical Tariff informed the attribute levels for cost.¹³ We modeled variation in survival rates after scenarios of no interventions,¹⁴ receipt of cardiopulmonary resuscitation (CPR),¹⁵ and the use of an automated external defibrillator (AED).^16,17 Distance to facilities was derived from results published in Stassen et al.¹⁸

Table 1.

Attributes and levels.

Attribute	Levels
Distance to closest adequate facility	10 km, 60 km, and 360 km
Provider of care	Bystander, family member, religious leader, basic life support–trained individual, advanced life support–trained individual
Response time	7 minutes, 14 minutes, 21 minutes
Chances of survival	1 of 100, 8 of 100, 16 of 100
Transport cost*	2500 ZAR, 5000 ZAR, 10 000 ZAR

USD Indicates US dollar; ZAR, South African rand.

^*1 USD = 15.06 ZAR at the time of data collection.

Selected attributes were pretested first on a small non-random sample of prospective participants to evaluate suitability and the cognitive response in understanding. We conducted cognitive interviews—in which respondents completed the DCE and were probed throughout the exercise with questions related to the clarity of instructions, understanding, and relevance of choice sets—to identify and address any problems that might arise in the process of eliciting preferences. Respondents reported that attributes were easy to understand; as such, no changes were made to the attributes and associated levels after cognitive interview. However, we revised the survey structure to enhance ease of completing the survey. We included a survey progress indication bar to provide respondents with a progress report on their survey progress. We also clarified instructions, so respondents understood they are expected to make trade-offs in their choices. Choice sets were then reviewed and validated for content with our expert panel that included local prehospital policy makers and researchers. The final survey tool was pilot tested on a sample of 50 individuals before full survey launch.

Experimental Design

We used an unlabeled design method in which 18 paired choice sets were presented under 2 alternatives: option 1 or option 2. A sample choice set is presented in Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2024.101006. Each task comprised 5 attributes and a range of 3 to 5 levels: distance to closest adequate facility (10, 60, and 360 km), provider of care (bystander, family member, religious leader, basic life support [BLS]–trained individual, advanced life support–trained individual), response time (7,14, and 21 minutes), chances of survival (1%, 8%, 16%), and transport cost (2500, 5000, and 10 000 South African rand [ZAR]).

A brief description that explained the attributes was provided to respondents (Appendix 2 in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2024.101006). Each of the attributes had 3 levels except for provider of care that had 4 levels. The number of attributes and respective levels resulted in a full factorial design of (4³ × 1⁵) = 320 and alternatives of attributes (320 × 319) / 2 = 51 040. A fractional factorial design, which used a subset of the full set of possible designs to estimate the effects, was created using the dcreate Stata 13 software module (StataCorp LP, College Station, TX) to create the optimally d-efficient design within the given constraints. D-efficient design allows for orthogonality (no correlation between levels of different attributes), level balance (proportional inclusion of levels), and minimum balance and overlap; 18 choice set questions were generated from this design. The final DCE survey tool was incorporated into a broader survey instrument that included sociodemographic information questions. The final questionnaire was programmed into Qualtrics survey platform (Qualtrics, Provo, UT) following guidelines by Weber.¹⁹

Data Collection

Eligible participants were 18 years or older with an interest in the topic or previous experience with OHCA, particularly adult cardiac arrest of nontraumatic origin. Participants panel was drawn from a prerecruited group of respondents who were recruited via a double-opt-in online panel and have agreed to take part in market research projects. The participants for our study were sample by means of nonproportional quota sampling—for the purpose of including views from all religious and language groups in a specific community and from both males and females. Participants were invited to take our survey in exchange for a small compensation ($0.50) for their time. Subjects participated in the online survey via mobile phone, tablet, laptop, or desktop hosted on their platform.

Minimum sample size was calculated based on the number of choice tasks (t), number of alternatives (a), and the largest product of levels for any 2 attributes (c) using the equation N > 500c / (t × a).²⁰ However, we sought to recruit a minimum of 150 respondents per province for adequate powering of subanalyses leading to a minimum sample of n = 1125.

Data were collected between April 25, 2022, and May 19, 2022, using Qualtrics software (Qualtrics, Provo, UT). The survey included sociodemographic characteristics and asked respondents to imagine a scenario in which they were experiencing an OHCA and had to choose between the services delivered (option 1 or option 2). Respondents were informed that there were no right or wrong answers and that they were free to end the experiment at any time.

Ethics Approval

Ethics approval for the study was provided by the University of California, San Francisco Institutional Review Board (IRB #21–34071) and the University of Cape Town Human Research Ethics Committee (HREC #564/2021).

Analysis

Mixed logit choice models were used for analysis and evaluation of variation in respondents’ preferences for selected attributes based on sociodemographic characteristics. Responses were imported and analyzed using Stata 17 (StataCorp LP, College Station, TX). Responses completed in less than 2 minutes were considered to be instances of straightlining and excluded from our analyses. Basic descriptive statistics were compiled on survey participants. A mixed effects logit regression model was used to identify the trade-offs between option choices. All levels within attributes were defined using binary coding and results were presented as odds ratios. Each participant’s choice between the pair presented in the task was treated as a single observation and included in the model a binary dependent variable where a “1” represented the option being chosen and “0” was an option not chosen. The dependent variable was set as probability of choosing one alternative over another, and the independent variables were characteristics of the surveyed population including gender, age, and religion. The command cmmixlogit in Stata was used to estimate the relationship between characteristics of the population and effect on the probability of choosing an alternative.

Subgroup analyses were conducted for the effect of participant characteristics on likelihood of selection of attributes and levels. Groups were defined by various demographic and socioeconomic characteristics, including (1) religion, (2) enrollment in private medical aid, (3) employment status, and (4) residence location. A log-likelihood ratio tested whether likelihood of selecting service differs across values within each subgroup.

We estimated marginal willingness-to-pay (WTP) as the marginal rate of substitution between the given attribute and cost. Mixed logit models in the WTP space were estimated using the mixlogitwtp command in Stata. We assume that the WTPs for the selected attribute randomly vary between individuals but remain constant across choice tasks, estimating our model with interpersonal heterogeneity and no intrapersonal heterogeneity.

We did not limit our design to restrict implausible scenarios. Options were randomly generated using software; therefore, it is possible that some scenarios were implausible, such as having a response time of 21 minutes and the probability of survival of 16% or a distance to the closest facility of 360 km and a cost of transport of only 2500 ZAR. Although the survey occasionally presented incongruity among the response time, time to hospital, and percentage of survival, in using an unlabeled approach, responses still allow for us to understand the variation in preference between attribute levels.

Limitations

Survey samples recruited from online platforms could have selection bias with responses deriving from only those with internet access and who are part of the platform research lists. Furthermore, online engagement often varies based on age and income.^21,22 For this reason our sample is younger, more highly educated, computer literate, and more likely to come from a metropolitan center than the average South African.²³ Compared with previous census, data our sample are not completely representative provincially; for example, only 1.76% of our respondents were from Northern Cape Province whereas the province represented 2.2% of the South African population in 2020. Nevertheless, online panels are a cost-effective mechanism for survey recruitment and facilitate large-scale sampling that is relatively representative.

Further research is needed to better understand the cost-effectiveness and financial feasibility of the services identified here as preferential by communities in South Africa. Additional DCEs could also help policy makers better understand the minutia of the trade-offs of these choices and further develop interventions that are acceptable to the community.

Results

A total of 2228 community members from across South Africa participated in our survey, completing a total of 40 104 choice tasks. Details of the respondents’ characteristics are presented in Table 2.

Table 2.

Demographic characteristics of survey respondents.

N = 2228	n (%)
Gender
Male	842 (37.79)
Female	1384 (62.12)
Other	2 (0.09)
Age (in years)
18–24	529 (23.74)
25–29	514 (23.07)
30–39	729 (32.72)
40–49	289 (12.97)
50–59	110 (4.94)
60 +	57 (2.56)
Religious
No	373 (6.74)
Yes	1855 (83.26)
African traditional religions	78 (4.20)
Christianity	1695 (91.37)
Hinduism	21 (1.13)
Islam	41 (2.21)
Judaism	3 (0.16)
Other	17 (0.92)
Type of residence area
Urban area	1192 (53.50)
Urban (suburban) cluster	861 (38.64)
Rural	175 (7.85)
Province
Eastern Cape	198 (8.89)
Free State	97 (4.35)
Gauteng	783 (35.14)
KwaZulu-Natal	357 (16.02)
Limpopo	176 (7.90)
Mpumalanga	146 (6.55)
North West	125 (5.61)
Northern Cape	40 (1.80)
Western Cape	306 (13.73)
Private medical aid
Yes	1050 (47.13)
No	1178 (52.87)
Employed
No	631 (28.32)
Yes	1597 (71.68)
Marital status
Single	1449 (65.04)
Married	681 (30.57)
Divorced	78 (3.50)
Widowed	20 (0.90)
Personal OHCA experience
Yes	217 (9.74)
No	2011 (90.26)
Family OHCA experience
Yes	1040 (46.61)
No	1188 (53.32)

Age of respondents ranged from 18 to 85 years with a mean of 32 (SD 10.31; interquartile range 25–38); 80% of respondents were younger than 40 years and only 20% of our sample were older than 40 years; 62% of respondents were female and most respondents were religious (83%) of whom most practiced Christianity (91%). Most of our respondents were single (65%) and employed (72%). Only 8% of respondents lived in a rural area.

Notably, 10% reported having a personal OHCA experience and 47% reported having a close family member experience an OHCA. A large proportion of respondents (47%) reported being on a medical aid scheme (ie, private medical insurance in which monthly premiums are paid) rather than the free public healthcare system provided by the South African government.

Main Effects Model

Results of our primary mixed logit choice model analysis on attributes and levels of OHCA care are presented in Table 3. All attribute levels, except care delivered by religious leaders under the provider attribute, were found to be statistically significant.

Table 3.

Mixed logit choice model analysis on attributes and levels of OHCA care.

Variables	Odds ratio
Base (10 000 km bystander, 7 minutes, 1 of 100, 2500 ZAR)
60 km	0.669* (0.014)*
360 km	0.360* (0.011)
Family member	1.404* (0.040)
Religious leader	0.980 (0.027)
BLS provider	1.398* (0.040)
ALS provider	1.488* (0.042)
14 minutes	0.730* (0.015)
28 minutes	0.472* (0.012)
8 of 100	2.323* (0.064)
16 of 100	3.938* (0.151)
5000 ZAR	0.840* (0.018)
10 000 ZAR	0.626* (0.017)
Observations	79 834

Note. Standard errors in parentheses.

ALS indicates advanced life support; BLS, basic life support; OHCA, out-of-hospital cardiac arrest; ZAR, South African rand.

^*P < .01.

The levels most preferred by patients (comparing with other levels in the same attribute) were care delivered by advanced life support–trained providers, who respond to scene in 7 minutes, where the closest adequate facility is 10 km away, the patient has a 16% chance of survival, and the cost of transport is 2500 ZAR. Preference orderings for all levels under each attribute were logically consistent; for example, respondents preferred shorter response times to longer response times, closer facilities to further facilities, and higher probabilities of survival rather than lower.

The preference ordering for the provider attribute was advanced life support–trained provider (1.49, CI 1.41–1.57), family member (1.40, CI 1.33–1.49), BLS-trained individual (1.40, CI 1.32–1.48), bystander (base), and then religious leader (0.98, CI 0.93–1.03). Respondents were 40.4% more likely to choose scenarios where family members provided resuscitation care rather than bystanders. This strength of preference was also slightly stronger than preference for scenarios of care delivered by BLS-trained providers (39.8%). Respondents were only slightly (2%) less likely to seek care provided by a religious leader. Respondents exhibited the strongest preference for probability of survival and were 2.32 times more likely to select 8 of 100 survival and 3.91 times more likely to select 16 of 100 survival. Cost seemed to be less crucial to participants. Respondents were 16% less likely to select options with twice the cost (5000 ZAR) and 37.4% less likely to select option that is 4 times the costs (10 000 ZAR).

We ran the main effects model and WTP analyses on a subgroup of respondents aged 40 years and older (n = 463) to better understand preferences of older populations who are more likely to suffer a myocardial infarction and subsequently OHCA. Results were consistent with our main analyses and presented in the Supplemental Materials found at https://doi.org/10.1016/j.vhri.2024.101006 (Appendices Tables 1 and 2).

Subanalyses

We tested the effect of the following patient characteristics on likelihood of preference: enrollment with private medical insurance, religious status, employment status, personal experience with OHCA, family experience with OHCA, and geographic area (urban vs rural). The results of the additional models are presented in Appendix 3 in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2024.101006.

Respondents who were not a part of a medical aid scheme had a stronger preference for lower costs of care than patients on private medical aid schemes; they were 18.2% versus 13.5% less likely to select scenarios that were twice the cost and 41.9% versus 31.9% less likely to select scenarios that were 4 times the cost. There was little variation between the preferences of respondents who identified as religious and those who identified as nonreligious, except respondents who identified as religious were 3% more likely (vs 26% less likely) to prefer care by a religious leader.

Rural responders had a slightly lower preference for decreased response times than urban respondents (23.7% vs 26.3% less likely to select scenarios with twice the response time and 51.1% vs 52.3% less likely to select scenarios with 4 times as long of response time). Having had OHCA experience was associated with a weaker preference for reduced distances to the closest adequate facility (21.7% vs 34.4% less likely to select scenarios with twice the distance; 55.8% vs 64.9% less likely to select scenarios with 4 times the distance) and reduced response times relative to respondents who had not experienced an OHCA (18.6% vs 28.0%).

WTP

The mean and SD of WTP measures derived from the model are presented in Table 4. The highest mean WTP for increased survival is 11 700 ZAR, followed by distance to health facility (8108 ZAR) and response time (5679 ZAR) and the lowest for increasing specialization of provider (1287 ZAR). Each of these represents between 0.4% and 4% of the average annual income in South Africa (312 384 ZAR per year).²⁴ In the case of distance to health facility and response time, there is a negative value of marginal WTP; this indicates that an increase in distance or response time is less preferred. Therefore, respondents would need to have a reduction in price to compensate for the downgrade in preference. WTP to reduce response time in half is 5679 ZAR, whereas respondents were willing to pay 11700 ZAR for an 8% increase in probability of survival. In cases where marginal utility of change in attribute depends on the level of the attribute, such as the provider of care, respondents had the lowest marginal WTP for more highly trained providers.

Table 4.

Selected estimates of marginal WTP (in ZAR) for attributes with assumed linearity.

Attribute	Estimate (95% CI)	SD
Chances of survival	11 699.93 (10 742.83–12 657.03)	10 584.51
Distance to closest adequate facility	−8108.18 (−8759.85 to −7456.51)	5713.93
Response time	−5678.59 (−6159.92 to −5197.26)	3139.25
Provider of care	1287.01 (1106.53–1467.49)	1576.25

Note. Data presented as estimates (95% CI).

WTP indicates willingness-to-pay; ZAR, South African rand.

The SDs of the WTP measures are generally high, indicating a substantial amount of heterogeneity in the respondents’ preferences. The correlation with other attributes associated with survival (distance and time) may reduce the magnitude of the effect associated with survival.

Subanalysis

The mean and SD of WTP measures based on respondent enrollment in private medical aid scheme are presented in Table 5. For those with private medical aid, the mean WTP was higher for all attributes relative to patients not enrolled in a private medical aid.

Table 5.

Selected estimates of marginal WTP (in ZAR) for attributes with assumed linearity by medical scheme enrollment.

Attribute	Estimate (95% CI)	SD	Estimate (95% CI)	SD
	Patients without medical aid		Patients with medical aid
Chances of survival	9461.19 (8411.98–10 510.39)	9345.16	15 365.59 (13 178.75–17 552.43)	14 402.19
Distance to closest adequate facility	−6692.41 (−7408.41 to −5976.40)	5030.65	−10 667.95 (−12 140.31 to −9195.60)	7442.60
Response time	−4727.84 (−5243.02 to −4212.67)	2205.83	−7469.63 (−8576.39 to −6362.89)	4966.36
Provider of care	1042.43 (831.54–1255.31)	1143.38	1631.53 (1250.13–2012.92)	2788.55

WTP indicates willingness-to-pay; ZAR, South African rand.

Discussion

Using a DCE, we explored preferences for OHCA care attributes among South African adults. We elicited and quantified these preferences and calculated respondents’ WTP for different attributes. The results derived from this study have several important implications that inform context-specific, culturally sensitive policy interventions that are understood and supported by community members. In particular, the findings may help guide decisions about when and how to invest in services that are acceptable to populations in South Africa and provide evidence for creating applicable OHCA management protocols to support the structure for an LRS.

Our findings indicated that the preferences of individuals in this LRS are in general congruent with preferences documented in high-income settings. We identified that patients prefer care with the shortest response time, delivered by the highest qualified individuals, which places them within the shortest distance of an adequate facility, gives them the highest chance of survival, and costs the least. However, our results also suggest that patients prefer care delivered by their family members over care delivered by the lay public in this setting.

In an LRS, where there is a fragmentation of trained providers, the ability to implement early defibrillation may be limited.^25,26 The possibility of public access to defibrillation has been expanded as a result of AEDs. Although bystander CPR programs have been extensively explored, it seems older individuals are less likely to CPR trained.²⁷ Our study indicates that possible redirection of investments from training for general laypersons toward targeted training of families of at-risk patients might be better suited to some communities in LRS. Furthermore, we know that OHCA incidents occur predominantly at home²⁸ and placement of AEDs in homes for use by family members has already been shown to have high acceptance.²⁹ Our findings suggest that an analysis to inform optimal defibrillator placement and how placement of AEDs in high-risk homes may contribute to health and policy goals in LRS is needed.

Surprisingly, nearly half of respondents reported having a close family member experience an OHCA; however, this may be due to respondents’ confusion between the terms heart attacks and OHCA. In addition, almost half of respondents reported being on a private medical insurance. This finding is significantly higher than reported statistics for South Africa in which only 15.2% of the population are served primarily through the private sector.³⁰ This pattern may be because the respondents included in our panel are more highly educated and possibly wealthier than average South Africans. However, 2 of the best represented provinces in our study have the highest percentage of individuals who are members of medical aid schemes (private health insurance), Western Cape Province (25.1%) and Gauteng Province (21.2%),³⁰ and larger proportion of individuals in metros were members of medical aid schemes (24%).²⁴

There are serious potential ethical and financial challenges to conducting laypersons OHCA training in LRS³¹—with evidence from high-resourced settings indicate that AED placement in homes of high-risk patients can have an incremental cost-effectiveness ratio of $87 569 per quality-adjusted life year gained, making it likely unaffordable most LRS.³² Using scarce resources where other interventions could have a greater impact on the general population or the costs families might assume after resuscitation may be particularly worrisome in LRS. Therefore, a better understanding of the cost-effectiveness of any such programs and a deeper understanding of the motivations of family members in LRS could help formulate targeted CPR training to meet citizens and family’s needs.

With every 1-minute decrease in response time to OHCA, the odds of survival may increase by 24%³³ and favorable neurological outcome by 13%.³⁴ Our study found that respondents are willing to pay the most for marginal gains in survival and response time. This finding indicates that patients could experience potential utility for shortened response time that extends beyond the capacity of this attribute to increase rate of survival. Additional research into reducing response times and mechanisms that can facilitate meeting nationally set response times in South Africa is direly required.³⁵

Our survey attempted to illustrate the considerable medical care costs that correspond to relatively small gains in OHCA survival in this the setting. Most urgent care delivered in the South Africa setting is done so free of charge to patients via the national government health services. Notably, 75% of South Africans do not use care outside of the public sector and are unlikely to need to consider cost in the care of their services.³⁰ To public sector patients the costs in our survey may seem more conceptual rather than pragmatic. However, our findings indicate that respondents who are members of private medical aid schemes have a higher WTP for improvements in attributes of care relative to those who only use the public health system. For populations where payment for healthcare services might be rare, this may bias having very low consideration for a cost attribute. Nonetheless, our results may be useful for private medical aid and ambulance providers who serve individuals with private medical insurance or in settings where patients predominantly pay for services out of pocket. Future analyses may consider exploring WTP response variation based on whether respondents consider costs as out-of-pocket expenses or covered benefits to provide additional insight into cost considerations.

Conclusions

In this study, we used a DCE to explore preferences in OHCA care by communities in an LRS. The study findings suggest that community members are willing to pay the most for marginal gains in probability of survival and shortened response times. The results further underline the potential for targeted training and AED placement among family members of high-risk individuals.