Using discrete choice experiment (DCE) to study the employment preference of health personnel to choose jobs can provide scientific and efficient basis for the formulation of policies related to the attraction and retention of health personnel in rural and remote areas.

To review, summarize and generalize the researches on the application of DCE to the field of human resources for health, so as to provide reference for future research.

Web of Science, PubMed, CNKI, Wanfang and VIP were systematically searched for literature related to DCE application to the field of health human resources from February to April in 2022. The data of included studies such as study objectives, data analysis models, the choice item settings and results were extracted.

A total of 44 papers published from 2000 to 2020 were finally included. The study objects were mainly health workers in service and students; six studies were labeled and others were unlabeled in the included papers; the number of working attributes ranged from 4 to 8; the number of levels of working attributes was predominantly 2 to 4 (88.6%, 39/44). The teamwork of general practitioners and workload were used as working attributes in the studies in high-income countries; the most frequently selected working attribute in studies from low-and middle-income countries was housing (21 times), followed by essential equipment such as facilities and medications (19 times), and learning/training opportunities (16 times). Income (salary) was incorporated into the working attributes across the literature. Furthermore, we formulated a framework of working attributes containing four aspects: social aspects, including the hospital volume, social support/respect, identification (establishment or becoming a permanent employee) ; working aspects, including working location, working conditions, workload, working (management) atmosphere, teamwork, mentoring by supervisors (for primary health workers) ; career development aspects, including years of promotion, training (continuing education) opportunities, academic and research opportunities; life aspects, including income, housing, traffic, and children's education. Mixed Logit Model was the frequently used analytical model (19 times), followed by Conditional Logit Model (9 times), Generalized Multinomial Logit Model (3 times) .

The heterogeneity of research findings in this field is large, making it difficult to draw uniform conclusions. Moreover, the application of DCE in the field of human resources for health still needs to be promoted globally. Relevant studies are very limited and the evidence obtained needs to be confirmed by further research.