Several low- and middle-income countries (LMICs) experience massive deficits and uneven distribution of health care workers. This matter has been additionally exacerbated by the collapse of health care systems in LMICs and the global policy climate [1]. To attain the objective of universal health coverage, the supply of health care workers should be improved, alongside an effort to improve their distribution, accessibility, performance, and productivity [2]. However, such attempts are severely compromised by a lack of health care professionals motivated to practice in rural locations in LMICs [2]. To achieve the objective of universal health coverage in 2030, a total of 54 million health care workers are still needed, based on the recommendation by the World Health Organization of having 44.5 health care workers per 10,000 people [2]. According to the World Health Organization estimate, there was a global shortage of 17.4 million health care workers in 2013, of whom 2.6 million were physicians. Achieving the United Nations’ Sustainable Development Goal 3 of ensuring good health and well-being for all necessitates addressing the addressing the significant financial and policy implications of the dearth of health care professionals [2], particularly specialist physicians, in rural regions [1].

The availability of human resources (HR) is among the top crucial requirements for the effective operation of rural health services [3]. In India, the majority of health professionals are located in metropolitan areas, resulting in a severe scarcity of health care providers for the rural population [4]. Factors such as poor governance and underinvestment in the public health system in rural areas drive health workers to migrate to urban areas in search of better salary structures, good living conditions, proper working environments, better accessibility to use equipment and educational facilities for children, and greater transparency in HR policies [4-9]. International outmigration of health professionals also dramatically impacts the domestic health workforce’s availability [10]. Prior research has demonstrated that physicians raised and educated in cities were unwilling to take rural postings. The primary reasons attributed to their choice (an unwillingness to work in rural areas) were the limited scope of occupational growth opportunities, a lack of accountability, issues with access to the necessary medical resources needed to perform their jobs adequately, and the absence of appropriate work policies in the health system [11]. To improve the retention rates of specialist physicians in the rural health system, it is necessary to implement strategies that involve regular rotations between rural and urban job placements [11]. In addition, other factors such as intermittent electricity, poor road connectivity, an unreliable public transport system, and inadequate educational facilities also discourage specialist physicians from relocating to rural and remote regions, which is evident from the substantial number of vacant jobs [1,12-16]. A recent study revealed that specialist physicians were willing to relocate from rural community health centers (CHCs) to towns, cities, or district headquarters to live with their families [17].

In the Indian public health care sector, many sanctioned positions remain vacant, especially for specialist physicians at rural CHCs [3,4]. The country is grappling with a shortage of specialist physicians in rural CHCs, estimated at 79.5% according to the Rural Health Statistics Report 2021-2022 [3]. The Ministry of Health and Family Welfare has been facing persistent challenges in ensuring the availability of specialist physicians at most of the rural CHCs across the country. As of March 31, 2022, only 541 out of India’s 5480 operational rural CHCs had all 4 specialist physicians [3]. Throughout the past few decades, the Ministry of Health and Family Welfare and the Department of Health and Family Welfare (at the state level) have used a range of measures to retain staff in different Indian states [11]. These retention strategies have included mandatory rural postings, linking postgraduate program enrollment to rural positions, and offering specific financial incentives [11]. However, there is still a lack of information on acquiring a more comprehensive understanding of the evidence-based strategies required to resolve this issue further and design potential solutions.

The statistics on the vacancies of specialist physicians at rural CHCs are presented in Table 1.

The specialist physicians’ shortage profoundly impacts the distribution of rural health care services in the nation [3,4]. Their limited availability and uneven allocation in rural CHCs can lead to higher rates of newborn and maternal fatalities, contributing to a more significant burden of disease [17]. Thus, addressing vacancies in the public sector necessitates immediate policy focus [4] to guide essential policy measures to enhance the delivery and administration of health care services in rural areas.

Discrete choice experiments (DCEs) were first used in market research, transportation, and environmental economics and later applied to investigate outcomes of interest in health care [18-22]. The DCE methodology identifies health workers’ preferences for a job based on its attributes. It is valuable for policy makers seeking to examine the most effective combinations of incentives or policy choices to promote health workers’ employment in underserved regions, particularly rural regions [9,23,24].

This protocol provides a detailed research method for a DCE, which will be used to quantify specialist physicians’ preferences for rural job incentives in Rajasthan. Rajasthan is among India’s largest states, considering its geography and cultural diversity [25], and it is currently facing a massive shortage of specialist physicians at rural CHCs [3].

The study will investigate the motivational job attributes valued by final-year postgraduate residents, who exhibit a higher propensity to quit their occupations and stay for shorter durations in rural areas, and specialist physicians, given the huge number of vacancies in the public health system. This will help policy makers understand the behavioral choices influencing and shaping specialist physicians’ decisions to accept rural postings. The study will specifically address the following research objectives:

The study research method used an exploratory sequential (qualitative+quantitative) mixed methods research design [26] to assess the incentive preferences and motivations of final-year postgraduate medical residents and specialist physicians to work in rural CHCs in Rajasthan. The survey tool for the study was constructed in 2 phases: qualitative and quantitative. The research design for the study protocol is provided in Figure 1.

DCEs are a precise and quantifiable method for gathering choices [9,27]. They are rooted in the theory formulated by Lancaster [28], which posits that commodities and services may be characterized by their fundamental properties, and the merit they hold for an individual is decided by a blend of these features. This approach involves providing participants with various choice profiles resembling hypothetical job alternatives and asking them to make selections. Every profile has multiple attributes that describe the job alternatives, and each attribute has multiple potential levels. Job descriptions are commonly merged to form decision sets, where participants are directed to pick the most appealing work profile. The analysis of participants’ selections among several alternatives allows for determining the relative significance of these attributes [22].

The DCE method is theoretically based on random utility theory, where individuals are reasonable decision makers with perfect discrimination ability to maximize utility relative to their choices [9,24,29].

The equation presented after this paragraph shows how the perceived value of an option (i) in a set of choices (C_n) for an individual (n) is broken down into 2 parts. First, there is a systematic component, which depends on the attributes of the options V (X_in, β). Second, there is a random component (ε_in), representing variations in preference that are not accounted for. This approach facilitates the incorporation of variations in desires across individuals, which can impact their decision-making.

U_in = V (X_in, β) + ε_in (1)

In the DCE method, it is assumed that an individual (n) will pick option (i) only if it provides the highest utility compared to all other options in the set (C_n). This logical decision-making process ensures that the chosen option truly offers the most value to the individual:

(2)

where y_in is a choice indicator equal to 1 if alternative i is chosen and 0 otherwise.

Rajasthan, located in India’s western region, is one of the country’s largest states in terms of geographic area. It spans 342,239 square kilometers, accounting for 10.4% of the country’s total land area [25]. It is home to 68.6 million people, representing 5.7% of India’s population. The state comprises 33 districts and 44,672 villages, with 75.13% of the population residing in villages [25,30]. Rajasthan faces an HR scarcity, with a shortage of 1939 specialist physicians for the 616 functioning CHCs [3]. In 2019, Rajasthan’s health system had only 22.02% (25,056/113,742) of the required physicians (both general and specialist) [31]. Hence, the scarcity of specialist physicians, particularly in the rural regions of Rajasthan, is a significant issue that hampers the efficient provision of health care services in such areas [32].

To develop the DCE for the study, published recommendations in the literature were followed, including identifying criteria for acceptable research methodologies [33-36], implementing the suggestions on constructing the study’s design [36-39], and adhering to guidelines on the appropriate statistical methods [40].

One crucial aspect of building a DCE is determining the subject matter’s pertinent qualities. Attributes are usually identified by collecting primary and secondary data to customize the DCE for the specific study environment [34].

The research obtained approval from the institutional ethics committee of the Indian Institute of Public Health Gandhinagar (TRC/2021-22/14). Administrative permission was granted by the Secretary of the Department of Medical, Health & Family Welfare and the Department of Medical Education, government of Rajasthan.

Due to the COVID-19 pandemic, participants recruited for the initial phase received the consent form and information sheet on the web before the qualitative interviews. They provided verbal consent for participation and recording during this phase. Telephone interviews were conducted attentively, with participants informed of their right to end the conversation at any time. For the second phase, involving the quantitative survey, participants provided written informed consent. Distinct codes were used for participant identification to maintain confidentiality. The study adhered to ethics guidelines and ensured the confidentiality of all participants.

The collection of qualitative data occurred between September and December 2021, followed by the quantitative data collection phase conducted between May and August 2022, using diverse recruitment methods. The selected attributes (and the number of associated levels) derived from the qualitative phase of the study included workplace infrastructure (2 levels); salary, including rural retention bonus (3 levels); staffing levels and workload (2 levels); residential facilities (3 levels); workplace location (2 levels); and transfer and promotion policies (2 levels; Table 2).

Using a blocked orthogonal fractional factorial design, attributes and associated levels derived from the qualitative phase of the study resulted in a construction of 90 unique choice sets partitioned into 6 blocks, each containing 15 (17%) distinct choice sets. A sample of a choice set is provided in Textbox 1.

For the final data analysis, the sample comprised 156 individuals, categorized into 4 specialist physicians’ groups: medicine (n=58, 37.2%), surgery (n=34, 21.8%), pediatrics (n=32, 20.5%), and OB-GYN (n=32, 20.5%). Statistical analysis of the data has been initiated, and the findings are expected to be disseminated in January 2025.

A preliminary investigation was performed using a conditional logit regression model, as mentioned in the existing literature on model specification [33,69]. The analysis was performed using Stata software (version 17.0) [64]. The results were analyzed using the beta coefficients and their related 95% CIs. A significance level of 5% was used. The model demonstrated a suitable fit for DCE (pseudo-R²=0.09) [23,47]. The findings indicated that all job attributes except advanced workplace infrastructure and substandard residential facilities significantly explained the preferences for rural jobs (P<.05). The results aligned with the preexisting preference expectations, demonstrating the theoretical validity of the design (Multimedia Appendix 2).

The findings indicated that 71% (30/42) of the participants consistently selected repeat-choice tasks, suggesting satisfactory reliability throughout the pilot survey (Multimedia Appendix 3). It was further found that the design exhibits version consistency because there was no remarkable variation between the blocks in selecting alternatives for the fixed-choice tasks. Furthermore, the study revealed that 30 (71%) of the 42 participants made rational decisions by choosing the dominant alternative. As a result, the design’s within-set monotonicity was confirmed. The task nonattendance rate was 29% (12/42), which is within an acceptable range for a DCE. Of the 42 participants, 30 (71%) based their selections on the consistency principles proposed by Sen [56], suggesting satisfactory decision reliability. Hence, the data from 29% (12/42) of the participants who were found to have not attended to the task were excluded from the final analysis due to inadequate information. Most of the participants concurred that the design was understandable (30/42, 71%) and that the questionnaire was easy to respond to (30/42, 71%).

Investing in increasing the number and accessibility of the health workforce is currently the most crucial area in India requiring policy attention [4]. The protocol presented herein seeks to bridge this knowledge gap by addressing the significant shortage of specialist physicians, a critical segment of the health care workforce, informing future policy interventions in Rajasthan, one of India’s largest states. This study uses a novel approach—DCE—to generate valuable, actionable, and policy-relevant data to improve the retention of specialist physicians in rural regions. Prior research on health HR in India has predominantly focused on general physicians, nurses, and midwives, while neglecting the importance of specialist physicians [5,7,70-72].

This study provides a thorough approach for performing a DCE and analyzes the findings of an initial survey conducted to assess the effectiveness of the study design before its complete adoption.

The pilot tests present a comprehensive overview of the essential procedures for determining job attributes and associated levels, generating choice sets, defining the utility model, selecting labeled and unlabeled choices, and implementing the design.

This study used a diverse array of methodologies to assess the efficiency of the design. Attributes were identified throughout the qualitative phase of the study through a meticulous iterative procedure involving in-depth interviews. However, the qualitative analysis revealed unique characteristics and levels specific to the location [73,74], which may add bias to the findings when examined in the broader context. With the incorporation of various attributes and associated levels, several experimental designs were created using an orthogonal fractional factorial design to create choice sets that guaranteed orthogonality, balance, and little overlap between attribute levels. Generic label options were used to create choice sets. Although specific labeling more accurately reflects individual preferences and actual choices, generic labeling is preferable for analyzing the trade-offs among multiple characteristics [38].

The direction of the coefficients validates the theoretical validity, demonstrating that the instrument effectively assesses participants’ choices [47,74] and that the attributes have been assigned appropriate levels, allowing for effective trade-offs [23]. Furthermore, the outcomes of the reliability and validity evaluations of the design indicate that it exhibits response efficiency [47].

By conducting a DCE, this study will serve as a cornerstone in providing a systematic and comprehensive understanding of several job characteristics, individual traits, and the importance of work incentives for specialist physicians in rural areas. This study’s findings will inform and shape judicious policy measures concerning implementing attractive work incentive schemes to attract and retain specialist physicians in rural areas. This will allow policy makers and health care administrators to refine their strategies for managing attrition and talent migration, thereby facilitating the development of a policy evaluation framework and planning health care HR for the state.

The primary advantage of a DCE involves its experimental nature, which allows for recording participants’ preferences under controlled experimental conditions, which helps to gain insight into how attribute changes affect individuals’ choices. Second, the job attributes and their corresponding levels were produced by performing comprehensive and precise objective evaluations, enhancing the design quality and result validity. The study identifies a wide variety of job characteristics that can be modified and adjusted to suit the preferences of specialist physicians in different national and international scenarios. Furthermore, pilot research was conducted using established frameworks to develop a thorough strategy for designing and assessing the quality of the DCE instrument.

While DCEs offer benefits, their implementation presents numerous challenges. At first, the study provided participants with hypothetical work options developed using qualitative research to collect relevant attributes. By using experimentally determined hypothetical alternatives during qualitative interviews, it became feasible to examine the participants’ “expressed preferences” rather than their “revealed preferences.” Moreover, the primary constraints of DCE methodologies stem from the cognitive challenge posed by the intricate assortment of options presented in bundles that encompass numerous attributes and levels. This can present several intricate trade-off questions to optimize statistical efficiency. Furthermore, the DCE’s distinct emphasis on specialist physicians guarantees that the results are particularly relevant to this group.

The results of this study will be shared through various platforms, such as scholarly journals that undergo peer review, conferences, and policy briefs. A concise summary of the findings will be made available to the participants on demand.

Further research is necessary to fully understand the broader context in which specialist physicians operate and the influence of the sociopolitical environment on their professional independence within the rural health care system to improve the retention and job satisfaction of rural specialist physicians. It is essential and equally important to further investigate and examine the factors that influence the uneven distribution and scarcity of female specialist physicians. To optimize the allocation of scarce monetary resources and conduct thorough assessments of policy initiatives, it is necessary to consider their costs, the accuracy of cost information, funding sources, and long-term sustainability. To comprehend the expenses linked to the policy intervention, a financial assessment of all resources used in executing the intervention is necessary.

This study offers a comprehensive process for building the DCE instrument and presenting the findings of instrument pilot testing. This study also demonstrated a unified and thorough framework that may be applied in future research endeavors. This study will be helpful in investigating the job factors needed to understand this issue thoroughly. Moreover, the study findings will aid in designing rural intervention packages, essential in promoting relevant policy measures to address the high turnover rates of specialist physicians in Rajasthan. In addition, it will offer valuable evidence to guide future policy interventions globally, facilitating the creation of incentive packages aimed at attracting and retaining specialist physicians in rural and underserved areas of LMICs.