背景:健康劳动力预测模型是强大的医疗保健系统的组成部分。本研究旨在回顾卫生人力预测模型的方法和方法的最新进展,并提出一套良好实践报告指南。
方法:我们通过搜索医学和社会科学数据库进行了系统综述,包括PubMed,EMBASE,Scopus,还有EconLit,涵盖2010年至2023年期间。纳入标准包括预测卫生人力需求和供应的研究。PROSPERO注册:CRD42023407858。
结果:我们的综述确定了40项相关研究,包括39个单一国家分析(在澳大利亚,加拿大,德国,加纳,几内亚,爱尔兰,牙买加,Japan,哈萨克斯坦,韩国,莱索托,马拉维,新西兰,葡萄牙,沙特阿拉伯,塞尔维亚,新加坡,西班牙,泰国,英国,美国),和一项多国分析(在32个经合组织国家)。最近的研究越来越多地在卫生劳动力建模中采用复杂的系统方法,结合需求,供应,和供需缺口分析。该综述确定了最近文献中常用的至少八种不同类型的卫生劳动力预测模型:人口与提供者比率模型(n=7),利用模型(n=10),基于需求的模型(n=25),技能混合模型(n=5),存量与流量模型(n=40),基于代理的仿真模型(n=3),系统动态模型(n=7),和预算模型(n=5)。每个模型都有独特的假设,优势,和限制,从业者经常结合这些模型。此外,我们发现卫生劳动力预测模型中使用了七种统计方法:算术计算,优化,时间序列分析,计量经济学回归模型,微观模拟,基于队列的模拟,和反馈因果循环分析。劳动力预测通常依赖于不完美的数据,在地方一级粒度有限。现有的研究在报告其方法时缺乏标准化。作为回应,我们为卫生人力预测模型提出了一个良好的实践报告指南,旨在适应各种模型类型,新兴方法,并增加利用先进的统计技术来解决不确定性和数据需求。
结论:这项研究强调了动态,多专业,以团队为基础,精细化需求,供应,以及由强大的卫生劳动力数据智能支持的预算影响分析。建议的最佳实践报告指南旨在帮助在同行评审期刊上发表卫生人力研究的研究人员。然而,预计这些报告标准将证明对分析师在设计自己的分析时很有价值,鼓励对卫生人力预测建模采取更全面和透明的方法。
BACKGROUND: Health workforce projection models are integral components of a robust healthcare system. This research aims to review recent advancements in methodology and approaches for health workforce projection models and proposes a set of good practice reporting
guidelines.
METHODS: We conducted a systematic review by searching medical and social science databases, including PubMed, EMBASE, Scopus, and EconLit, covering the period from 2010 to 2023. The inclusion criteria encompassed studies projecting the demand for and supply of the health workforce. PROSPERO registration: CRD 42023407858.
RESULTS: Our review identified 40 relevant studies, including 39 single countries analysis (in Australia, Canada, Germany, Ghana, Guinea, Ireland, Jamaica, Japan, Kazakhstan, Korea, Lesotho, Malawi, New Zealand, Portugal, Saudi Arabia, Serbia, Singapore, Spain, Thailand, UK, United States), and one multiple country analysis (in 32 OECD countries). Recent studies have increasingly embraced a complex systems approach in health workforce modelling, incorporating demand, supply, and demand-supply gap analyses. The review identified at least eight distinct types of health workforce projection models commonly used in recent literature: population-to-provider ratio models (n = 7), utilization models (n = 10), needs-based models (n = 25), skill-mixed models (n = 5), stock-and-flow models (n = 40), agent-based simulation models (n = 3), system dynamic models (n = 7), and budgetary models (n = 5). Each model has unique assumptions, strengths, and limitations, with practitioners often combining these models. Furthermore, we found seven statistical approaches used in health workforce projection models: arithmetic calculation, optimization, time-series analysis, econometrics regression modelling, microsimulation, cohort-based simulation, and feedback causal loop analysis. Workforce projection often relies on imperfect data with limited granularity at the local level. Existing studies lack standardization in reporting their methods. In response, we propose a good practice reporting
guideline for health workforce projection models designed to accommodate various model types, emerging methodologies, and increased utilization of advanced statistical techniques to address uncertainties and data requirements.
CONCLUSIONS: This study underscores the significance of dynamic, multi-professional, team-based, refined demand, supply, and budget impact analyses supported by robust health workforce data intelligence. The suggested best-practice reporting
guidelines aim to assist researchers who publish health workforce studies in peer-reviewed journals. Nevertheless, it is expected that these reporting standards will prove valuable for analysts when designing their own analysis, encouraging a more comprehensive and transparent approach to health workforce projection modelling.