OCCUPATIONAL CHARACTERISTICS AND ECONOMIC ACTIVITIES OF HEALTH WORKERS IN THE QUARTERLY LABOUR FORCE SURVEY: 2008-2017 Aphiwe Dinga Student Number: 1589447 Supervisors: Professor Laetitia Rispel Dr Duane Blaauw A research report submitted to the School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements for the degree of Master of Science in the Field of Epidemiology and Biostatistics Johannesburg, South Africa 310 June 2024 2023 i DECLARATION I, Aphiwe NZ Dinga, declare that this research report is my original work. It is being submitted for the degree of Master of Science in Epidemiology and Biostatistics in the University of the Witwatersrand, Johannesburg. It has not been submitted before for any degree or examination at this or any other University. I am aware that plagiarism (the use of someone else’s work without their permission and/or without acknowledging the original source) is wrong. I confirm that the work submitted for assessment for the above degree is my own unaided work, except where I have explicitly indicated otherwise. I have read the sections on referencing and plagiarism in the Wits Plagiarism Policy (appendix 1), and I have followed the required conventions in referencing the thoughts and ideas of others. I understand that the University of the Witwatersrand may take disciplinary action against me, including suspension or permanent expulsion, if there is a belief that this is not my own unaided work or that I have failed to acknowledge the source of the ideas or words in my writing. I have included as an appendix a report from “Turnitin software indicating the level of similarity in my research report. Signature Date: 10 June 2024 ii DEDICATION To Bulelwa and Zola Dinga, thank you for believing in my dreams and the sacrifices you have made for me over the years. Words cannot express my gratitude for having you as my parents; half of my academic battles were already won because of your love and support. Thank you, GOD; with this, I glorify your name. iii ABSTRACT Background There is global emphasis on the importance of research and analyses of health labour markets. The latter is defined as dynamic systems consisting of the demand and supply of health workers, influenced by a country’s regulations and institutions. However, there is limited national data to inform a health labour market analysis. Aim The aim of the study was to analyse the demographic, occupational characteristics and the economic activities of health workers who were surveyed in the Quarterly Labour Force Survey (QLFS) from 1 January 2008 to 31 December 2017. Methodology This study was a cross-sectional secondary data analysis of the health workers captured in the QLFS, a household survey that is conducted every three months by Statistics South Africa. The survey focuses on the labour market activities of individuals aged 15 to 64 years who live in South Africa. The sample analysed for this study was all health workers surveyed in the QLFS during the study period. Both the South African Standard Classification of Occupations (SASCO) and the Standard Industry Classification (SIC) codes were used to extract data on all health occupations to ensure that the entire health workforce in the QLFS was included in the current study. To identify predictors of employment a multiple logistic regression was carried out. STATA ® 15 was used for the statistical analysis. Results The study sample comprised a total of 5 502 health workers. Nurses constituted the highest proportion of health workers in the survey (60.1%) while medical doctors and dentists represented 10.0%. Nurses were older than the other categories of health workers with a mean age of 43.6 years (SD±10.3), compared to the mean age of 41.8 (SD±10.8) for doctors, 38.6 (SD±10.4) for mid-level health workers and 37.8 (±10.8) for allied health workers. The majority (59.0%) of health workers were employed in the public sector, and in urban areas (83.8%). Only 4.6% of doctors and 7.0% of allied health workers were employed in rural areas. iv Overall, the study found that fewer than 1% of health workers reported more than one job during the 10-year period. The results of the logistic regression showed that the odds of employment were approximately two times higher for health workers between the ages of 36- 45 and 46-55 years old and 1.8 times higher for health workers between the ages of 26-35. There were 0.5 odds of employment for health workers aged 56-64 years compared to the reference age group of 18–25-year-olds. Females were less (0R=0.56) likely to be employed as compared to males. Compared to health workers in urban areas, those in rural areas were less (0.47) likely to be employed. Health workers were 0.53 times less likely to be employed outside the health industry as compared to being employed in the health industry. Conclusion Although the QLFS provides useful information on the health workforce in South Africa, the results highlight the need for investment in a robust human resources for health information system. Keywords: human resources for health; health workforce; health labour market; occupational characteristics, economic activities, South Africa v ACKNOWLEDGEMENTS Professor Laetitia Rispel and Dr Duane Blaauw, I have walked this difficult journey with you by my side; you were there every step of the way, guiding me and providing all the support I needed to complete this master's degree. I am grateful to God for the honour and privilege of having you as my supervisors. I appreciate your patience, even at times when I did not deserve it. Working with you has also helped me gain valuable experience in scientific research and strong character traits that will help me in my future endeavours. Thank you. My MSc (Epidemiology) research was nested in and funded by the South African Research Chairs Initiative (SARChI) entitled Research on the Health Workforce for Equity and Quality, held by Professor Rispel (National Research Foundation Grant # 102219). Additional funding was provided by the LESEDI Project, funded by Atlantic Philanthropies (Grant ID: 21408). The views expressed in this study are my own, not those of the funders. I want to thank my sisters, Kwandisa Gwija, Zenande Mkupa and Sinazo Dinga, for their unwavering love, encouragement, and support. I hope this will encourage you to study further. Many thanks to Associate Professor Peter Barron for the encouragement, career guidance, and advice and for guiding me in the right direction. Thank you to the SARChI team. Thank you to my lecturers, Professor Jonathan Levin, and Dr Michel Muteba. vi TABLE OF CONTENTS DECLARATION .......................................................................................................................................... i DEDICATION ........................................................................................................................................ ii ABSTRACT ............................................................................................................................................... iii ACKNOWLEDGEMENTS ........................................................................................................................... v TABLE OF CONTENTS ................................................................................ Error! Bookmark not defined. LIST OF TABLES ..................................................................................................................................... viii LIST OF FIGURES ..................................................................................................................................... ix LIST OF ABBREVIATIONS AND ACRONYMS ............................................................................................. x CHAPTER 1: INTRODUCTION ............................................................................................................. 11 1.1 Background ........................................................................................................................... 11 1.2 Literature review ................................................................................................................... 13 1.2.1 Definition of terms ........................................................................................................ 13 1.2.2 Approach to the study .................................................................................................. 14 1.2.3 Health labour market research ..................................................................................... 14 1.2.4 Health workforce and demographic characteristics ..................................................... 16 1.2.5 Occupational characteristics ......................................................................................... 19 1.2.6 Employment activities ................................................................................................... 19 1.3 Problem statement and study rationale ............................................................................... 21 1.4 Aim and objectives ................................................................................................................ 22 1.5 Outline of the research report .............................................................................................. 22 CHAPTER 2: METHODOLOGY ............................................................................................................ 23 2.1 Introduction .......................................................................................................................... 23 2.2 Background to the Quarterly Labour Force Survey .............................................................. 23 2.2.1 Design and sampling ..................................................................................................... 23 2.2.2 QLFS materials and methods ........................................................................................ 25 2.3 Description of the study ........................................................................................................ 26 2.3.1 Study population ........................................................................................................... 26 2.3.2 Data collection and preparation ................................................................................... 28 2.3.3 Study sample ................................................................................................................. 30 2.4 Variable description and data management ........................................................................ 31 2.5 Data analysis ......................................................................................................................... 33 2.5.1 Trends in sociodemographic characteristics trends of health workers. ....................... 33 2.5.2 Trends in occupational characteristics and economic activities of health workers. .... 33 vii 2.5.3 Demographics and occupational factors associated with employment status ............ 34 2.6 Ethical considerations ........................................................................................................... 34 CHAPTER 3: RESULTS ......................................................................................................................... 35 3.1 Introduction .......................................................................................................................... 35 3.2 Study participants ................................................................................................................. 35 3.3 Trends in socio-demographic characteristics of health workers .......................................... 36 3.3.1 Age of health workers ................................................................................................... 36 3.3.2 Gender .......................................................................................................................... 37 3.3.3 Population groups ......................................................................................................... 39 3.3.4 Geographical distribution ............................................................................................. 40 3.4 Trends in occupational characteristics and economic activities of health workers ............. 42 3.4.1 Health worker occupations ........................................................................................... 42 3.4.2 Employment status ....................................................................................................... 43 3.4.3 Employment industry .................................................................................................... 44 3.4.4 Employment sector ....................................................................................................... 46 3.4.5 Multiple job holding ...................................................................................................... 47 3.5 Demographic and occupational characteristics associated with employment status ......... 48 CHAPTER 4: DISCUSSION ................................................................................................................... 52 4.1 Introduction .......................................................................................................................... 52 4.2 Summary of key results ......................................................................................................... 52 4.3 Socio-demographic characteristics ....................................................................................... 53 4.4 Occupational characteristics and economic activities .......................................................... 54 4.5 Factors associated with the employment of health workers ............................................... 55 4.6 Limitations and Strengths of the Study................................................................................. 56 CHAPTER 5: CONCLUSION AND RECOMMENDATIONS ..................................................................... 58 5.1 Conclusion ................................................................................ Error! Bookmark not defined. 5.2 Recommendations ................................................................... Error! Bookmark not defined. 5.3 Future Research .................................................................................................................... 61 REFERENCES .......................................................................................................................................... 62 APPENDICES .......................................................................................................................................... 67 Appendix 1: Plagiarism Declaration .................................................................................................. 67 Appendix 2: QLFS Questionnaire ...................................................................................................... 68 Appendix 3: Ethics Waiver ................................................................................................................ 85 Appendix 4: Turn it in report ............................................................................................................ 86 viii LIST OF TABLES Table 2.1: Illustration of QLFS rotation groups, 2008-2017 ................................................... 24 Table 2.2: Health occupations included in the study ............................................................... 27 Table 2.3: Occupations excluded from the study..................................................................... 28 Table 2.4: Sample of health workers and the overall QLFS sample ....................................... 30 Table 2.5: Data management process ...................................................................................... 31 Table 3.1: Mean (SD) and median (IQR) of ages by health worker category ......................... 37 Table 3.2: Age groups by health worker categories, 2008-2017 ............................................. 37 Table 3.3: Trends in occupational characteristics of health workers, 2008-2017 ................... 42 Table 3.4 :Trends in reported MJH among health workers, 2008-2017 .................................. 48 Table 3.5: Association between demographic variables and employment status .................... 49 Table 3.6: Association between occupational characteristics and employment status ............ 49 Table 3.7: Multiple logistic regression of predictors of employment in health workers ......... 51 ix LIST OF FIGURES Figure 2.1: Illustration of the design of the questionnaire used in the QLFS survey .............. 25 Figure 2.2:Collection of QLFS data and publication of results ............................................... 26 Figure 2.3: Data preparation process for analysis .................................................................... 29 Figure 3.1: Total number of health workers by occupation category, 2008-2017 ................... 35 Figure 3.2: Median age distribution among health workers, 2008-2017 ................................. 36 Figure 3.3: Gender distribution of health workers, 2008-2017 ............................................... 38 Figure 3.4: Break down of health worker categories by gender .............................................. 38 Figure 3.5: Trends in the population group of health workers from 2008-2017 ..................... 39 Figure 3.6: Breakdown of health workers’ categories by population group ............................ 40 Figure 3.7: Trends in the geographical distribution of health workers, 2008-2017 ................ 41 Figure 3.8: Breakdown of health worker categories by geography type ................................. 41 Figure 3.9: Trends in employment status of health workers 2008-2017 ................................. 43 Figure 3.10: Breakdown of health worker categories by employment status .......................... 44 Figure 3.11: Industry distribution of health workers, 2008-2017 ............................................ 45 Figure 3.12: Breakdown of health worker’s categories by industry ........................................ 45 Figure 3.13: Trends in the distribution of health workers by employment sector, 2008-2017 46 Figure 3.14: Breakdown of health workers by employment sector, 2008-2017 ...................... 47 https://worldhealthorg-my.sharepoint.com/personal/dingaa_who_int/Documents/Desktop/Dinga%20Aphiwe%20Final%20RR_31%20March%2023%20cleaned.docx#_Toc131508852 x LIST OF ABBREVIATIONS AND ACRONYMS DoH Department of Health GHWA Global Health Workforce Alliance HEEG HIC HLM Health Employment and Economic Growth High-Income Country Health Labour Market HREC Human Research Ethics Committee HRH Human Resources for Health HWF Health Workforce LMICs Low- and Middle-Income Countries MDGs MJH Millennium Development Goals Multiple Job Holding NDoH National Department of Health NHA National Health Act NHI NHS OECD OSD PERSAL National Health Insurance National Health Service Organisation for Economic Cooperation and Development Occupational Service Dispensation Personal and Salary System QLFS Quarterly Labour Force Survey RWOPS SANC Remunerative Work Outside of Public Service South African Nursing Council SARChI SASCO South African Research Chairs Initiative South African Standard Classification of Occupations SDGs SIC Sustainable Development Goals Standard Industry Classification StatsSA Statistics South Africa UHC Universal Health Coverage USA United States of America WHO World Health Organization https://www.google.co.za/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwjfmPv_htzOAhWfHsAKHZgdAwkQFggdMAA&url=http%3A%2F%2Fwww.health.gov.za%2F&usg=AFQjCNFPavQZZNOogaYiqOSC1mQ5MyCkvg 11 CHAPTER 1: INTRODUCTION 1.1 Background Human resources for health (HRH) are crucial for the effective functioning of any health system and achieving the global goal of universal health coverage (UHC), aimed at ensuring equitable access to essential health services and financial risk protection (1). HRH are also critical for resilient health systems that are able to respond to emergencies and natural disasters, and for population health improvements (2). The 2030 Global Strategy on HRH uses a labour market conceptual framework, with a health labour market defined as a “dynamic system composed of two distinct but closely related economic forces: the supply of health workers and the demand for such workers” (2, p.13,22). A key objective of the 2030 Global HRH Strategy is to align health workforce investment with population and health system needs by taking account of the behaviour of healthcare providers (who supply labour), the institutions that demand labour (employers), and how these health policy actors respond to legislation, regulation and health and education policies (2). In 2016, the High-Level Commission on Health Employment and Economic Growth (HEEG) recommended further research to analyse health labour markets (3). Despite the growing recognition of the importance of health labour markets, more research is needed in low-and middle-income countries (LMICs), particularly in Africa (4). Furthermore, there needs to be more analyses of standard data sets on the labour market. Such labour market analysis can generate new knowledge on the characteristics and distribution of the health workforce in order to inform health policy decisions (3, 4, 5). South Africa has made strides in HRH since the end of apartheid, with numerous transformative laws, various governance structures, good fiscal space for public sector employment, and a robust health professional education system that is well-regulated (1, 6). However, some of the weaknesses include, “insufficient stewardship of HRH planning across the entire healthcare system; lack of a national integrated HRH information system; and inadequate information on overall HRH supply to address historical inequities between urban and rural areas and between the public and private health sectors” (7, p.13). The imperative to improve health workforce 12 planning and HRH information systems was reiterated in South Africa’s 2030 HRH Strategy (1) Notwithstanding South Africa’s sub-optimal HRH information system, routine datasets such as the population census, labour force surveys, and health professional registries could provide useful information on the health workforce (4, 8, 9). The Labour Force Survey of Statistics South Africa (StatsSA) is a quarterly household-based survey of the labour market activities of adult South Africans (10). Since 2008, the Quarterly Labour Force Survey (QLFS) has collected regular information on the employment, occupation, sector of work, remuneration, and job changes of the population (10). The surveys could be an important data source on the occupational characteristics of the health workforce (11). Understanding the South African health labour market is essential for the successful implementation of the proposed National Health Insurance (NHI) system, which is the country’s vehicle for UHC (1). This is because the health workforce is central to the realisation of the goal of UHC (1). In light of the gaps in health labour market research, the aim of this study was to analyse the demographic and occupational characteristics as well as the economic activities of health workers who were surveyed in the QLFS for the period from 1 January 2008 to 31 December 2017. This chapter provides the background and context of the Master of Science (MSc) research study. Section 1.2 contains the literature review, including the definition of terms, the importance of research on the health labour market, and the demographic and occupational characteristics of the health workforce. This is followed by the problem statement and the rationale for this research (section 1.3). The chapter concludes with the aims and objectives of the study (Section 1.4) and the structure and outline of the remainder of the research report (Section 1.5). 13 1.2 Literature review 1.2.1 Definition of terms Dual work Also known as moonlighting, or holding multiple jobs, is understood as having a second job (s) in addition to a primary full-time job (12). Economic activities Activities that contribute to the production of goods and services in a country. In this study, economic activities are work activities of an individual within an establishment/s, including dual work and types of work (13). Health worker/ Health workforce Health workers are defined as people whose job is to improve and protect the health of their communities (5). In this study a health worker is any individual who indicated that he/she had health training following completion of matric and delivered or previously delivered direct clinical or indirect healthcare services to sick or injured patients. The term includes professional health providers such as doctors and nurses, rehabilitation therapists, as well as auxiliary health workers and mid-level health workers. Labour force participation This is the proportion of a country’s working age population which is actively engaged in the labour market through working or actively looking for work (13) Unemployment Individuals aged 15–64 years who did not have a job, have actively looked for work or tried to start a business in the four weeks preceding the OLFS interview, were available for work or had a job or business to start at a definite date in the future (13). 14 1.2.2 Approach to the literature review The literature review was informed by the study objectives (Section 1.4). Google Scholar was the primary search engine used to search for literature. The following databases were also used: PubMed, ScienceDirect and Web of Science. The following key words and medical subject headings (MeSH) were used in different combinations during the literature search: Africa; demographics; dual work; economic activities; gender; geographic distribution; health labour market; health workforce; health worker; human resources for health; low-and-middle income countries; quarterly labour force survey; occupational characteristics; South Africa. A grey literature search was also conducted. An effort was made to only include results from studies which used routine data such as labour force surveys. The literature review foregrounds studies published between the beginning of the study period (2008) until 2023. 1.2.3 Health labour market research Health labour markets are influenced by various factors, including globalisation, public sector reforms, population health needs, the demand for health services and the supply and governance of health workers (5, 14, 15). The 2006 World Health Report cautioned that expanding labour markets had “intensified health professional concentration in urban areas and accelerated international migration from the poorest to the wealthiest countries” (16, p.143). The characteristics of the resultant health workforce crisis included health worker shortages, inappropriate skill mixes, and health service coverage gaps (16, 17). Although health labour markets were not singled out as a priority, the World Health Organization (WHO) 2006 report recommended funding for priority human resources research (16). In the decade between 2006 and 2016, the importance of research on health labour markets received increased attention from health policy makers and scholars (4, 18, 19, 20, 21). In 2013, McPake and colleagues (22) argued for the application of labour economic frameworks to analyses of the health workforce crisis, especially in LMICs. Similarly, Soucat and Scheffler (23) proposed a labour market conceptual framework and regional and country data to examine the HRH crisis in Africa. In 2016, WHO released the 2030 Global HRH Strategy, in recognition of the criticality of the health workforce to the achievement of the Sustainable Development Goals (SDGs) (2). The Global HRH Strategy emphasised the centrality of health labour market 15 analysis to the formulation of national health workforce strategies and to the mobilisation of resources to implement these strategies, subsequently echoed in the 2016 HEEG Report of the High-level Commission (2, 3). This is because such health labour market analysis can generate new knowledge on the drivers and policy levers that influence health workforce production, deployment, retention, and performance, thus contributing to the practical design of health policies and appropriate interventions (3). In addition, the value of using routine data in health labour market analyses has been illustrated by joint WHO and World Bank teams (5). Notwithstanding variations in the completeness and the quality of data, the 2017 estimates identified a growing HRH demand, with more than 18 million additional health workers needed by 2030 to achieve the SDGs (5). The modelling estimates suggest that health workers' demand and supply will continue to fall short of population health needs in LMICs (5). Notably, the joint WHO and World Bank team underscored the need for quality HRH data and evidence and recommended that all countries invest in "analytical capacity for HRH and health system data'' based on agreed-upon standards (5, p.21). High-income countries continue to have superior data quality and analytical capacity for research on health labour markets compared to many LMICs, yet the evidence base on HRH labour markets in high-income Organization for Economic Co-operation and Development (OECD) countries does not provide a clear picture of the expected future HRH situation in these countries (21). Nonetheless, in other diverse country settings, there is an encouraging increase in the research and analyses of HRH requirements and labour market dynamics (23, 24, 25, 26). Similarly, the 2013 study on the labour market for health workers in Africa provides valuable information (27). The author highlighted the importance of context for each of the 53 African countries studied and recommended a country's need to select a combination of relevant health measures and contextual factors to estimate HRH requirements (27). In South Africa, the Ministerial HRH Task Team adopted a labour market approach and three different modelling estimates to develop the 2030 HRH Strategy (1). The modelling indicates a need for more skilled health professionals in South Africa (1). One of the five recommendations focused on institutionalising data-driven and research-informed health workforce policy, planning, management, and investment (1, p.42). This recommendation highlights the importance of analysing routine datasets such as the QLFS conducted by Statistics South Africa to add information on South Africa's health labour market dynamics. 16 Although Festus, Kasongo, Moses and Yu (6) examined the changes in the South African labour market from 1995- 2015 and the employment trends in the general South African population, no published studies could be found that have focused on an analysis of health workers in the QLFS. 1.2.4 Health workforce and demographic characteristics Age analysis of health workers Demographic trends directly impact labour markets, notably labour supply, productivity, and demand (5). An ageing population due to demographic and epidemiological changes is expected to increase the demand for health services and, therefore, the demand for HRH, and is expected to reduce the supply of health workers (28). The age structure of the health workforce also has important implications for policy and management, such as workability, training opportunities, and occupational health and safety (29). The combination of an ageing population and an ageing health workforce has been termed a "double whammy" while representing an opportunity for innovation and transformation (29, p.1). In 2020, Szabo and colleagues (30) highlighted the dearth of research on the notion of health worker demography. This is because demography on age-trends could have direct implications for labour markets, notably labour supply, productivity, and demand (28). Szabo and colleagues (30) proposed a conceptual framework that focuses on key types of health workforce entry (e.g., education, migration) and exit (e.g., dismissal, retirement, death), the nature of which can influence the age and gender profile of the workforce. Unsurprisingly, the authors underscore the need for robust data for each health worker category (e.g., doctors, nurses), their age or age group, and disaggregation by employment sector and geographical distribution (30). Studies have found that the age distribution of health workers varies across regions and even among countries within a region. In Finland in 2015, the largest age group among nurses and midwives was 35 years to 44 years (30). Similar variations in the age distribution of the health workforce have been reported in LMICs. In China, a study to examine the trends, composition and distribution of the nurse workforce using national public datasets from 2003 to 2018 found that most nurses (60.3%) were under the age of 35 years (31). A Cameroon study found that 17 36.1% of the health workforce was between the ages of 31 years and 40 years (32), while a study in Sudan reported that only 20% of health workers were between 50 years and 60 years of age (33). An analysis of the HRH labour market in Zambia found that most nurses were under 35 years (34). The South African Nursing Council (SANC) Statistics indicated that only 28.0% of registered nurses were less than 30 years of age, with those in the 50 years and older age group accounting for 47.0% of professional nurses (35). As nurses make up a large part of the healthcare workforce in South Africa, this ageing nursing workforce has potentially negative implications for healthcare delivery, exacerbating the widespread reported shortages (35). In contrast, using the database of the Health Professions Council of South Africa (HPCSA), Tiwari and colleagues (36) found that the majority of registered male doctors (21.9%) are between the ages 35 years and 44 years, whereas the majority of the registered female doctors (38.8%) are between the ages 30 years and 39 years. Gender analysis of health workers Gender biases in health systems and workforce policies have created systemic inefficiencies regardless of the country's income level (37). Evidence suggests that structural and relational factors impact women's engagement in the labour market in general and in the health sector (3, 4, 5, 30, 37). Globally, women comprise 75.0% of healthcare workers, yet they face significant work-related inequities in compensation, working conditions, and health outcomes (37). Data on 16 high-income countries for 2007 show that women make up the majority of managers in the health and social sector but were underrepresented when adjusting for their share of total employment (5). Nonetheless, an analysis of gender equity in the health workforce of 104 countries demonstrated progress and changing trends (38). The study found that the nursing and midwifery professions were dominated by females, whereas most health workers in other professions such as doctors, dentists, and pharmacists, were males (38). The study also found that women’s representation in the health sector has increased over time, particularly among the higher wage health care occupations (i.e. physicians, dentists and pharmacists) (38). 18 In the Organization for Economic Cooperation and Development (OECD) countries, the proportion of female physicians increased by 13.0% between 2000 and 2017 (38). Furthermore, the majority of health workers in higher wage health occupations (i.e. physicians, dentists and pharmacists) under the age of 40 are female (38). Interestingly, there is also an increasing share of male workers in younger age bands in nursing and midwifery (38). However, females continue to dominate the nursing and midwifery professions, whereas men continue to dominate in the medical, dental and pharmacy professions (37). Context remains important regarding gender equity. South Africa’s 2030 HRH Strategy underscores the importance of gender-transformative policies (1). Although Tiwari et al (36), found that 59.4% of the medical doctors on the HPCSA register were men, South Africa's HRH Strategy modelling suggests an average projected increase of 259% in female doctors within the surgical specialties by 2030 (1). Geographical distribution of health workers The urban-rural maldistribution of healthcare workers is a universal challenge, particularly in LMICs (39). The maldistribution between urban and rural areas is exacerbated by the inequities between the public and private health sectors (2). The reasons for the geographical maldistribution are complex, and include lack of explicit redistribution policies, health professional education, tertiary hospitals in urban areas, personal preferences of health workers, resource constraints, poor infrastructure in rural areas, and low employers demand (2, 39, 40, 41, 42). Analysing the urban-rural distribution of health workers provides insights into targeted policies to address the maldistribution and the shortages (2, 40, 43). In India, researchers have found a significant disproportion between the urban and rural distribution of health workers across all provinces (44). A study in Cameroon found evidence of health workforce imbalances between the capital city, Yaoundé and the poorest and under-developed regions (45). Similarly, in Sudan the study also reported a maldistribution of the health workforce between urban and rural areas, with the urban northern state having 4.6 times more health workers than the rural South Dafar state (45). In South Africa, the NHI white paper emphasises the disparities between the urban and rural health workforce (1). The maldistribution of health workers hampers access to health care for people in rural areas and impacts mortality rates (1). 19 1.2.5 Occupational characteristics Globally, the health sector is the largest employer of workers, accounting for a large portion of national budgets (5). The majority of health workers fall into four critical healthcare occupations: doctors, dentists, nurses, and midwives (5). Although countries differ in planning and implementing human resource policies and strategies, the starting point was to collect data on health workers' occupational characteristics, especially their categories and skills mix (46). Skills mix refers to the mix of staff or the distribution of different health professional categories in the health sector (47). Researchers have investigated the skill mix in the health workforce and found that globally, nurses make up the majority of the health workforce, with sizable differences between doctors and nurses across countries (46, 47, 48, 49). Similarly, an Australian study reported that over half of their health workforce was dominated by nurses, with medical doctors being the second largest (14). In LMICs such as Bangladesh, severe shortages of nurses were reported, as well as inadequate numbers of physicians (48). In contrast, a labour force survey in India found that the majority of the health workers comprised nurses and midwives and pharmacists (50). The public health sector in South Africa is predominantly nurse-driven, with nurses making up more than half of healthcare providers (1). However, there were wide variations in the skills mix across the nine provinces, with larger numbers of doctors in the urban provinces of Gauteng, the Western Cape and KwaZulu-Natal, and a greater reliance on community health workers in the rural provinces (1). 1.2.6 Employment activities Employment status The employment status of healthcare workers influences their availability and many other factors in the healthcare system (3, 51). Traditionally, the health care systems of most countries offered a standard employment contract featuring permanent full-time work with one employer (2). However, the permanent full-time employment contract lost its dominance with the realisation that employees prefer flexible contracts, whereas employers prefer to offer fewer employee benefits (19). In South Africa, for example, the employment status of many healthcare workers has changed, with the emergence of non-standard forms of employment such as agency nursing (52). One study has found that some healthcare workers prefer casual 20 employment because of better quality of life, reduced work responsibilities, limited workplace politics, and work-life balance with reduced job stress and better remuneration (52). In contrast, permanent healthcare workers have reported the advantages of paid leave for holidays and illness, retirement annuity, workplace coherence, access to career training and progression and job security that (52). Employment sector There is a dearth of studies that have examined healthcare workers' employment by sector which would assist in understanding employment preferences and occupational characteristics (11). More than half of health workforce in India was reported to be in the private sector (53). Ngah et al. (32) reported that about two-thirds of health workers in Cameroon were employed in the public sector. The remaining workforce was in the private sector, including non-profit organisations (32). In South Africa, the government uses the Personal and Salary System (PERSAL) to record and pay all health workers employed in the public sector, but this system does not exist in the private health sector (1). However, an analysis of general practitioner and specialist numbers by ECONEX reported that approximately 40.0% of general practitioners and nurses work in the private sector, caring for approximately 17.0% of the population (54). In general, while many South African healthcare workers continue to work in the public health sector, it has been reported that they hold negative perceptions about their working conditions and environment, which often influences their decision to leave the sector (1). Understanding the employment sector where healthcare workers are located will contribute to new knowledge about their occupational characteristics and economic activities (1). Multiple Job holding Another factor has been multiple job holding (MJH), a practice where an individual has a primary full-time job with at least one extra paid job (12). Labour economists have also used the labour supply economic theory to explain multiple job holding as an additional source of income (12, 55, 56). Researchers have investigated dual practice among medical doctors and given various reasons for having more than one job, including improving their skills and knowledge (12, 55, 56, 57). 21 In Cameroon, Ngah et al. (32) found that many public sector health workers have part-time jobs in the private sector to compensate for their low salaries, resulting in high levels of absenteeism, reduced performance, and reduce productivity in the public health sector (32). In South Africa, the Public Service Act makes provision for Remuneration Work Outside the Public Service (RWOPS) based on certain conditions set by the employer (1). A cross-sectional study of nurses reported that agency nursing and dual practice were common in the South African health system (58). Multiple job holding has severe consequences for the health system, including exhaustion, paying less attention to work while on duty, taking sick leave when not sick and conflicting schedules between their primary and secondary jobs, thus undermining the efficiency of the South African health system (58). 1.3 Problem statement and study rationale This MSc research study was informed by three problems. Firstly, despite an encouraging increase in research on the health labour market, there are still knowledge gaps on the South Africa’s health labour market. Secondly, the QLFS datasets from Statistics South Africa are available, but there has been insufficient analysis of this data source to examine the demographics, occupational characteristics, and economic activities of health workers, as well as the factors associated with the employment status of health workers. Lastly, the current HRH strategy reiterates the need for a dedicated intelligence unit in the National Department of Health to steer planning and the development of a robust HRH information system (1). However, it is unclear whether the QLFS data could contribute to the gaps on HRH information in South Africa. Hence, the rationale for this study was to examine the occupational characteristics and economic activities of health workers surveyed in the QLFS. Additionally, the study analysed trends over a ten-year period from the QLFS’s inception in 2008 to 2017. The study also set out to determine whether the data on health workers in the QLFS can be used to assist with HRH analysis and contribute to planning. Lastly, the study also contributes to the generation of new knowledge on the South African health labour market. 22 1.4 Aim and objectives The aim of the study was to analyse the demographic and occupational characteristics as well as the economic activities of health workers who were surveyed in the QLFS from 1 January 2008 to 31 December 2017. The specific objectives of the study were to: 1. Describe the trends in demographic characteristics of health workers who were surveyed in the QLFS from 1 January 2008 to 31 December 2017. 2. Describe the trends in occupational characteristics and economic activities of health workers who were surveyed in the QLFS from 1 January 2008 to 31 December 2017. 3. Evaluate the demographic and occupational factors associated with employment status of the health workers who were surveyed by the QLFS using data from the fourth quarter of 2017. 1.5 Outline of the research report The remainder of this research report is organised as follows: Chapter 2 focuses on the study methods, including a detailed description of the primary study, namely the QLFS, and the methodological approach used to achieve the study objectives. Chapter 3 describes the study results and the trends for the selected period. Chapter 4 discusses the study findings, guided by the study objectives and the review of the relevant literature. Chapter 5 presents the study conclusion and recommendations. 23 CHAPTER 2: METHODOLOGY 2.1 Introduction This chapter describes the methodology used to meet the objectives of this study, as outlined in Chapter 1. Section 2.2 presents an overview of the preliminary study, namely the Quarterly Labour Force Survey (QLFS) from which the data for this study was drawn. Section 2.3 presents a detailed description of the methods of the current study, which was an analysis of the QLFS data on health workers. Section 2.4 summarises data management, while Section 2.5 describes the data analysis in this study. The chapter concludes with ethical considerations. 2.2 Background to the Quarterly Labour Force Survey 2.2.1 Design and sampling The QLFS is a household survey that Statistics South Africa conducts every three months (or every quarter). Although the original Labour Force Survey (LFS) started in 2000, the survey in its current form was launched in 2008. The purpose of the QLFS survey is to collect data on the labour market activity of individuals aged 15 years or older who live in South Africa (14). The QLFS is a longitudinal survey with a rotating panel design. The sampling was done in a stratified two-stage design. The first stage used probability proportional to size sampling of the Primary Sampling Unit (PSU), followed by systematic random sampling of the dwelling units in the second stage. The current sample size of the QLFS is 3 080 PSUs, divided equally into four rotation groups. The rotating panel design is shown in Table 2.1, with each shape representing a rotation group (or panel) and each colour representing a different year. The QLFS base year was 2008, with the four original sampled rotation groups (panels). The rotation of groups (panels) commenced in the first quarter of 2009 when 25% of the 2008 sample was replaced with a new rotation group. Hence, 25% of the sample is replaced every quarter. The sampling meant that every fourth quarter had a unique group of households. 24 Table 2.1: Illustration of QLFS rotation groups, 2008-2017 Q1 Q2 Q3 Q4 2008 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2009 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2010 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2011 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2012 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2013 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2014 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2015 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2016 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ 2017 ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ ● ■ ▲ ★ Source: Adapted from Statistics South Africa [16] Legend: each shape represents a rotation group, and each colour represents a specific year. Rotation group Shape Year Colour 1 ● 2008 2 ■ 2009 3 ▲ 2010 4 ★ 2011 The QLFS used the South African Standard Classification of Occupations (SASCO) to classify all occupations (such as pharmacists, nurses, and doctors) and used the real-time management system to code all occupations into four digits which range from 1110-9333. These occupations were also grouped by the industry they belong to using the Standard Industry Classification (such as Manufacturing, Health, and Mining). The SIC provides a national framework and allows for international occupational comparability. 25 2.2.2 QLFS materials and methods The QLFS survey involved interviews with household members using a structured questionnaire (see Appendix 2). The questionnaire is divided into five sections (illustrated in Figure 2.1) with various skip patterns depending on the responses to questions on economic activities. Figure 2.1: Illustration of the design of the questionnaire used in the QLFS survey Source: Adapted from Statistics South Africa [13] 26 The QLFS questionnaires were completed by trained field staff and captured using an electronic tool called the Real-Time Management System. As indicated, data collection happened every quarter, and the results were released in the next quarter while simultaneously starting the data collection for that quarter (Figure 2.2). Figure 2.2:Collection of QLFS data and publication of results Source: Statistics South Africa (13) 2.3 Description of the MSc study This study was a secondary data analysis of the QLFS conducted by Statistics South Africa, focusing on the ten years from 1 January 2008 to 31 December 2017. 2.3.1 Study population The study population consisted of all health workers who were surveyed in the QLFS during the study period. Both the South African Standard Classification of Occupations (SASCO) and the Standard Industry Classification (SIC) codes were used to extract data on all health occupations to ensure that the entire health workforce is included. The inclusion criteria were any individual who: (1) had formal health training; (2) and delivered or previously delivered direct clinical or indirect healthcare services to sick or injured patients, or (3) formed part of the formal healthcare system defined as public and private health sectors in South Africa. Table 2.2 shows health occupations included for the final analysis and how they were grouped into different health worker categories. 27 Table 2.2: Health occupations included in the study SASCO codes Health Occupations Grouped categories 2221 Medical practitioners, physicians 1. Doctors/Dentists 2222 Dentists (general), Dental Specialists 2230 Nursing and midwifery professionals 2. Nurses 3231 Nursing associate professionals, Nurses 3232 Midwifery associate professionals 3228 Pharmaceutical assistants 3. Mid-level health workers 5132 Paramedics and nursing aids 3228 Pharmacists’ assistants 3220 Optometrist assistants 3221 Medical assistants 3225 Dental assistants 3223 Dieticians and nutritionists 4. Allied health workers 3224 Optometrists and opticians 3226 Physiotherapists and related associate 2224 Pharmacists 3229 Modern health associate professionals (e.g., speech therapists) As shown in Table 2.3, ten health occupations were excluded from the analysis because they did not meet the criteria of a health worker in this study or did not form part of the formal health care system. Psychologists were excluded because there needs to be a clear distinction in the QLFS between clinical psychologists and other psychologists. In addition, psychologists are grouped with psychometricians and psych technicians who are unrelated to health. 28 Table 2.3: Occupations excluded from the study SASCO codes Health Occupations 3241 Traditional healers 3242 Faith healers 3222 Sanitarians 5139 Personal care workers, e.g., child-minders 5133 Home-based care workers 2445 Psychologists/Psychometricians/ 2446 Social work professionals 3460 Social work associates 3152 Safety health inspectors 1110 Legislators 2.3.2 Data collection and preparation Statistics South Africa publishes the QLFS data on their official website, and the datasets can be downloaded freely. The de-identified datasets downloaded from the NESSTAR Quarterly Labour Force Survey database were available on (nesstar.statssa.gov.za:8282/webview/) for the study period. Although 40 datasets that correspond to the quarters of the ten-year study period were available and initially downloaded, only the quarter four data of each year were used. This is because the fourth quarter represents a unique group of households and individuals due to the rotating panel nature of the QLFS (illustrated in Table 2.1 above). This avoided the double counting of individuals and ensured that the health workers in each year were unique. The 10 datasets were cleaned and inspected individually for consistency and completeness. This was done by ensuring that all selected variables were named and coded the same in all 10 datasets before appending them. Figure 2.3 shows the steps followed in preparing the final dataset for statistical analysis and is further explained in detail below. http://nesstar.statssa.gov.za:8282/webview/ 29 Figure 2.3: Data preparation process for analysis ● Step 1 entailed downloading 40 QLFS datasets from the StatsSA website, inspecting the data to verify the rotating panel (overlap of participants) and selecting data from the fourth quarter of each year only. At the end of this process, ten datasets were selected. ● Step 2 entailed identifying key variables, checking duplicates, cleaning, and ensuring all variables were named and coded appropriately in each of the ten datasets before appending. ● Step 3 entailed appending all ten datasets making a final dataset that contained 813 861 observations. ● Step 4 entailed data inspection and rechecking of duplicates in the appended dataset using UQNO, which is an 18-digit household ID, as the unique identifier. ● Step 5 entailed selecting 14 key variables that were kept for the analysis of this study. The 14 key variables selected are presented in the data management section below in Table 2. 5. ● Step 6 entailed extracting observations of all health occupations, which were 9919. Step 1: selecting 4th quarters from 2008-2017 Step 2: data inspection and cleaning of critical variables in each dataset. 0 duplicates found Step 3: appending ten dataset sets into 1 dataset. 813 861 observations Step 4: data inspection and rechecking of duplicates in the appended dataset. 0 duplicates found Step 5: Select key variables for analysis. 16 variables selected. Step 6: Extract all health occupations. 9919 health occupations Step 7: excluded 4263 that did not meet the criteria., leaving: 5656 health occupations Step 8: excluded 154 health workers from the eligible sample that were not between 18-64 years old. Step 9: Final study sample: 5502 health workers 30 ● Step 7 entailed excluding observations of 4263 health occupations that did not meet the definition of health workers in this study, e.g., traditional healers and environmental health practitioners. ● Step 8 excluded 154 observations of participants younger than 18 (2) and older than 64 (152) from the dataset. ● Step 9, the remaining eligible health workers were 5502. 2.3.3 Study sample The entire sample of relevant health workers was used in the analysis (N=5502); hence no further sampling was done. In the study's final eligible sample of health workers, 461 (8.37%) were unemployed, and 5 041 (91.62%) were employed health workers. Table 2.4 compares the total sample of health workers each year and the overall QLFS sample from which these were drawn. Table 2.4: Sample of health workers and the overall QLFS sample Year Overall QLFS sample N Sample of health workers N Frequency Frequency 2008 93,062 627 2009 87,653 570 2010 83,357 575 2011 84,953 578 2012 85,112 614 2013 87,523 581 2014 84,274 529 2015 70,345 490 2016 69,231 476 2017 68,351 462 Total 813,861 5,502 31 2.4 Variable description and data management Table 2.5 shows the original variables in the QLFS, their definitions, and the data management process. Table 2.5: Data management process QLFS variable Definition Data Management Process Sociodemographic variables UQNO: This is an 18-digit ID allocated to each household, with 1st eight digits being the PSU number. Extracted PSU as a new variable and used that to adjust for the survey design all analyses. Weight Original survey weights: This is the probability or sampling weight proportional to the inverse probability of being sampled because of the survey design. This variable was used to adjust for the survey design Weight was added as svyset psu [pweight=weight]. Person No, Person No is the respondent number of a participant in a single household. We generated a unique ID number for each respondent, which was used to check for duplicates. Q14age Respondents' current age, the original QLFS defines its working age as individuals between 15-64 years old. Age was restricted to 18-64 years and further changed to 5 categories as indicated below:1(18-25); 2(26-35); 3(36-45); 4(46-55); 5(56-64). q16maritalstatus Respondents’ marital status: 1. Married 2. Living together like husband and wife 3. Single 4. never married, 5. divorced/separated 6. widow/widower This variable had low numbers in some of the groups and was changed to two categories for analysis purposes and recoded as: 1-Married (living together like husband and wife, married) 2-Single (never married, divorced/separated, widow/widower) q17education Respondents’ highest education This variable was excluded from the analysis due to data entry errors; over 45 % of this variable had incorrect information where Doctors and Nurses, for example, had no schooling or had primary schooling as their highest education. Province Current province of residence 1- Western Cape, 2- Eastern Cape, 3- Northern Cape, 4- Free State, 5- Kwazulu Natal, 6- North-West, 7- Gauteng, 8- Mpumalanga, 9- Limpopo This variable was rearranged into alphabetical order and coded as: 1-Eastern Cape, 2- Free State, 3-Gauteng,4- Kwazulu Natal, 5-Limpopo, 6-Mpumalanga, 7- Northern Cape, 8- North-West, 9-Western Cape 32 QLFS variable Definition Data Management Process Geo_type Respondents' geography type/ type of areas they lived. 1. Urban 2 Rural 3. Farms, traditional mining This variable had low numbers in some groups and was changed to two categories for analysis. recorded as: 1 -urban (urban) 2 -rural (farms, traditional, mining) Occupational and economic variables Q42occupation The actual occupation of respondents who were employed when the survey was conducted. There were approximately 380 listed occupations Used the codes derived from the South African Standard Classification of Occupation to extract all employed health workers and classified them into four groups: Group 1- Dentists and doctors Group 2- All categories of nurses Group 3- Mid-level health workers (these include pharmacy, medical and dental assistants) Group 4- Allied health workers (these include physiotherapists and health workers involved with the rehabilitation of patients) q13prevoccupation The last occupation unemployed respondents had when they were last employed. There were about 400 listed occupations. Used the codes derived from the South African Standard Classification of Occupation to extract all unemployed health workers q43industry and q316previndustry q43industry- current industry of employed respondents. q316previndustry- previous industry of unemployed respondents. 1. social and personal services;2. Agriculture hunting. 3. forestry and fish;4. Mining and quarrying, manufacturing, 5. electricity: gas and water supply; 6. Construction; 7 wholesale and retail trade; 8 transport ;9 financial intermediation, private households A new variable was created (Industry) and recorded as 1. 1.Health industry (social and personal services) 2. 2. Non-health industry (Agriculture hunting; forestry and fish, mining and quarrying, manufacturing, electricity: gas and water supply, construction, wholesale and retail trade, transport, financial intermediation, private households) Status_expanded This variable was used to classify the employment status of all respondents: 1. Employed 2. Unemployed 3. Not economically active A new variable was generated (employment status) using the status_exp variable and was recorded as: 1-employed (employed) 0-unemployed (unemployed, not economically active) 33 QLFS variable Definition Data Management Process q415typebusns Respondent's employment sector 1-National/Provincial/Local government, 2-Government controlled business (e.g.Eskom/Telkom), 3-private enterprise, 4-Non-profit organisation (NGO/CBO), 5-Private household) This variable had low numbers in some categories and was changed to two categories for analysis purposes and re coded as 1-public sector (National/Provincial/Local government, government-controlled business (e.g., Eskom/Telkom) 2-private sector (private enterprise, Non-profit organisation (NGO/CBO), private household) q41multiplejobs Respondents who had more than one job 1-Yes 2-No This variable was kept as coded initially 1-Yes 2-No Other Year Year of survey 2.5 Data analysis 2.5.1 Trends in sociodemographic characteristics trends of health workers. In addressing objective 1, the median age with the interquartile range was calculated from 2008 to 2017 and presented in a box plot. A table presenting both the mean with standard deviation (SD) and median with interquartile range (IQR) by each health occupation was presented. Proportions were calculated for the categorical variables (health occupation, gender, population group, and geographic distribution), presented in line graphs, tables, and bar graphs. All variables were stratified by health worker categories and a Chi-squared test was conducted to identify significant differences between them. A non-parametric test for trends in ordered groups was used to assess whether the trends in the different years were statistically significant. This was done using the nptrend (Cochrane-Armitage) command in Stata. The Stata svyset and svy: commands were used to account for the stratified two-stage sampling design in the primary study. 2.5.2 Trends in occupational characteristics and economic activities of health workers. In addressing objective 2, proportions for economic and occupational variables (employment status, employment sector, industry, multiple job holding, and geographical distribution) were calculated. They were presented in tables and bar graphs. All variables were stratified by health worker categories and evaluated using a chi-squared test. A non-parametric test for trends in ordered groups assessed whether the trends in the different years were statistically significant. This was done using the nptrend (Cochrane-Armitage) command in Stata. 34 2.5.3 Demographics and occupational factors associated with employment status In addressing objective 3, the outcome variable was employment status (1-employed/ 0- unemployed). A bivariate analysis (Chi-square test) assessed the association between employment status and each explanatory variable. These included demographic variables (age group, gender, population group) and economic and occupational variables (health worker categories, industry, geography type). A cut-off point for association with a p-value of less than 0.2 was used to select variables for the final regression model. Multiple logistic regression analysis determined the demographic, economic and occupational factors associated with employment status in health workers surveyed in the QLFS. Adjusted odds ratios are presented. The Archer-Lemeshow post-estimation goodness of fit test suitable for survey data was done to check whether the final model fit. The level of significance (α) was set at 0.05. The Stata svyset and svy: commands were used to account for the stratified two-stage sampling design in the primary study. 2.6 Ethical considerations The University of the Witwatersrand Human Research Ethics (Medical) Committee (HREC) granted a waiver (see Appendix 3) as a clearance certificate was not required: number W-CBP- 190222-2. This is because the study analyses the information in the public domain, which is de-identified with no personal identifiers (names, home addresses and identification numbers). The QLFS is well-established, and the outcomes of interest (occupational characteristics and economic activities) did not present any ethical dilemmas. Importantly, the downloaded datasets excluded any information on earnings due to their sensitive nature, while the de- identified information removed the potential for discrimination against particular groups or individuals. 35 CHAPTER 3: RESULTS 3.1 Introduction This chapter presents the results of the secondary data analysis of the QLFS from 2008 to 2017, in line with the study objectives. Section 3.2 presents descriptive statistics of the sample used in the final analysis. Section 3.3 describes the trends in the demographic characteristics of health workers surveyed in the QLFS from 2008-2017. Section 3.4 shows the economic and occupational characteristics of health workers surveyed in the QLFS from 2008-2017. Section 3.5 presents the logistic regression results used to examine the demographic and occupational factors associated with employment status amongst health workers. 3.2 Study participants In this study's final eligible sample of health workers, 461 (8.37%) were unemployed, and 5 041 (91.62%) were employed, health workers. Figure 3.1 shows the total number of health worker cadres in each category. The highest proportion of health worker cadres in the QLFS were nurses (60.01%), and the lowest proportion were allied health workers (8.10%). Figure 3.1: Total number of health workers by occupation category, 2008-2017 Doctors/Dentists, 549, 10% Nurses, 3307, 60% Mid-level health workers, 1201, 22% Allied health workers, 445, 8% Doctors/Dentists Nurses Mid-level health workers Allied health workers 36 3.3 Trends in socio-demographic characteristics of health workers 3.3.1 Age of health workers Figure 3.2 presents the trends in the median age of health workers from 2008 to 2017. The median age remained relatively stable over the study period, with minor fluctuations observed. In 2008 the median age of health workers was 41.0 years, which slightly increased to 42.0 years in 2012 returning to 41.0 years in 2017. The overall test for trends in the age of health workers from 2008-2017 was not statistically significant. Figure 3.2: Median age distribution among health workers, 2008-2017 Table 3.1 shows the mean with standard deviation (SD) and median with interquartile range (IQR) for age across different health worker categories. The mean age of doctors/dentists, nurses, mid-level health workers and allied health workers was 41.8 (SD±10.8), 43.6 (SD±10.3), 38.6 (SD±10.4), 37.8 (±10.8) respectively. 37 Table 3.1: Mean (SD) and median (IQR) of ages by health worker category Health occupations Means SD Median IQR Doctors/Dentist 41.8 10.8 41 33-50 Nurses 43.6 10.3 44 36-52 Mid-level health worker 38.6 10.4 37 30-46 Allied health workers 37.8 10.8 36 29-45 Table 3.2 below presents age groups of different health worker categories. Notably, the majority of nurses (72.2%) were considerably older (between 56-64 years) than other age health worker’s categories, as also reflected by the mean and median provided in Table 3.1. Although doctors/dentists constituted 10.0% of the total health workers, only a small proportion (4.9%) were found in the 18-25 age-group. Table 3.2: Age groups by health worker categories, 2008-2017 Age group Doctors/Dentists n(%) Nurses n(%) Mid-level health workers n(%) Allied health workers n(%) Total n(%) 18-25 27 (4.9%) 121 (3.7%) 114- (9.5%) 55 (12.4%) 317(5.8%) 26-35 161(29.3%) 688(20.8) 406(33.8%) 153(34.4%) 1408(25.6%) 36-45 150(27.3%) 1023(30.9%) 363(30.2%) 126(28.3%) 1662(30.2%) 46-55 146(26.7%) 988(29.9%) 232(19.3%) 74(16.6%) 1440(26.2%) 56-64 65(11.8%) 487(14.7%) 86(7.2%) 37(8.3%) 675(12.3%) Total 549(100%) 3307(100%) 1201(100%) 445(100%) 5502(100%) 3.3.2 Gender Figure 3.3 presents trends in the gender distribution of health workers from 2008-2017. Throughout the study period, females consistently comprised a higher proportion of health workers surveyed compared to men. However, there was a slight decrease in the proportion of female health workers over the years (82% in 2008 to 78% in 2017). The proportion of males gradually increased accordingly from 18.0% in 2008 to 22.0% in 2017. The overall test for the trend in the gender of health workers from 2008-2017 was statistically significant (p=0.003), suggesting an increasing proportion of male health workers over time. 38 Figure 3.3: Gender distribution of health workers, 2008-2017 Figure 3.4 below provides a breakdown of health worker categories by gender. Nursing had the highest proportion of females (91.0%) compared to other health worker categories. In contrast, there was a higher proportion of males (53.0%) than females in the doctors/dentists’ category. Figure 3.4: Break down of health worker categories by gender 18% 16% 15% 19% 20% 19% 19% 21% 21% 22% 82% 84% 85% 81% 80% 81% 82% 79% 79% 78% 0% 20% 40% 60% 80% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 Year Male Female 53% 9% 27% 25% 47% 91% 73% 75% 0% 20% 40% 60% 80% 100% Doctors/Dentists Nurses Midlevel health workers Allied health workers Health workers category Male Female 39 3.3.3 Population groups Figure 3.5 presents trends in the population group of health workers from 2008-2017. Black Africans accounted for the highest proportion of health workers surveyed, comprising 64.5% of the total. Specifically, the percentage of Black African health workers decreased from 64.0% in 2008 to 57.0% in 2012 and then increased to 69.0% in 2017. Indian health workers accounted for the smallest proportion of health workers surveyed. The overall test for trend in the population group of health workers from 2008-2017 was statistically significant (p=0.011), indicating an overall increase of Black African health workers. Figure 3.5: Trends in the population group of health workers from 2008-2017 Figure 3.6 below describes the categories of health workers by population group for the period 2008-2018. Among doctors/dentists, the highest proportion (48.0%) was White, while the lowest proportion was Coloured (8.0%). For nurses, the majority (72.0%) were Black Africans, with the lowest proportion being Indian (3.0%). Coloureds accounted for a higher proportion of mid-level health workers (15.0%), compared to Whites (8.0%) and Indians (4.0%). Among allied health workers, the highest proportion was White (49.9%) with Indians (9.0%) and Coloureds (11.0%) having the lowest proportion. 64% 65% 64% 62% 57% 63% 63% 71% 70% 69% 14% 16% 16% 14% 16% 14% 15% 9% 11% 11% 6% 4% 5% 5% 6% 5% 5% 5% 4% 5% 17% 15% 14% 19% 22% 18% 17% 15% 15% 15% 0% 20% 40% 60% 80% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 year Black Coloured Indian/Asian White 40 Figure 3.6: Breakdown of health workers’ categories by population group 3.3.4 Geographical distribution Figure 3.7 presents the trends in the geographical distribution of health workers from 2008- 2017. Overall, most of the health workers surveyed were in urban areas rather than rural areas. In 2008, the proportion of health workers living in urban areas increased from 85.0% to 88.0% in 2012, before slightly declining to 84.0% in 2017. The overall test for the trend in the rural- urban distribution of health workers from 2008-2017 was not statistically significant (p=0.115), indicating little change in this distribution over time. 28% 72% 73% 32% 8% 15% 15% 11% 16% 3% 4% 9% 48% 10% 8% 49% 0% 20% 40% 60% 80% 100% Doctors/Dentists Nurses Mid-level health workers Allied health workers health worker categories Black Coloured Indian White 41 Figure 3.7: Trends in the geographical distribution of health workers, 2008-2017 Figure 3.8 provides a breakdown of the geographical distribution of health workers. The proportions of nurses and mid-level workers in rural areas were significantly higher at 18.0% and 21.0%, respectively, compared to 5.0% of doctors/dentists and 7.0% of allied health workers in rural areas. Figure 3.8: Geographic distribution of health worker categories 85% 84% 82% 85% 88% 81% 84% 81% 81% 84% 15% 16% 18% 15% 12% 19% 16% 19% 19% 16% 0% 20% 40% 60% 80% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 Year Urban Rural 95% 82% 79% 93% 5% 18% 21% 7% 0% 20% 40% 60% 80% 100% Doctors/Dentists Nurses Mid-level health workers Allied health workers Health worker categories Urban Rural 42 3.4 Trends in occupational characteristics and economic activities of health workers 3.4.1 Health worker occupations Table 3.3 presents the health worker categories surveyed in the QLFS from 2008 to 2017 shown as numbers (n) and proportion (%). The highest proportion of health workers surveyed were nurses (60.1%), although there was a decreasing trend over the study period. Doctors/dentists category increased from 9.9% in 2008 to 11.4% in 2012 but decreased to 10.6% in 2017. Mid-level health workers accounted for 19.0% in 2008, which decreased to 15.6% in 2012 and then increased to 27.1% in 2017. Allied health workers accounted for the lowest proportion of health workers but showed an increasing trend. The variations in the trends of health occupations from 2008-2017 were statistically significant (p= 0.001). Table 3.3: Trends in occupational characteristics of health workers, 2008-2017 Years Doctors /Dentists n (%) Nurses n (%) Mid-level health workers n (%) Allied health workers n (%) 2008 62 (9.9%) 412 (65.7%) 119 (19.0%) 34 (5.4%) 2009 46 (8.1%) 359 (63.0%) 137 (24.0%) 28 (4.9%) 2010 52 (9.0%) 365 (63.5%) 120 (20.9%) 38 (6.6%) 2011 58 (10.0%) 378 (65.4%) 87 (15.1%) 55 (9.5%) 2012 70 (11.4%) 387 (63.0%) 96 (15.6%) 61 (9.9%) 2013 68 (11.7%) 357 (61.4%) 101 (17.4%) 55 (9.5%) 2014 45 (8.5%) 288 (54.4%) 152 (28.7%) 44 (8.3%) 2015 46 (9.4%) 270 (55.1%) 140 (28.6%) 34 (6.9%) 2016 53 (11.1%) 251 (52.7%) 124 (26.1%) 48 (10.1%) 2017 49 (10.6%) 240 (51.9%) 125 (27.1%) 48 (10.4%) Total 549 (10.0%) 3307 (60.1%) 1201(21.8%) 445 (8.1%) n: numbers; %: proportions 43 3.4.2 Employment status Figure 3.9 presents the trends in the employment status of health workers from 2008 to 2017. The proportion of employed health workers increased from 93.0% in 2008 to 94.0% in 2012 before gradually declining to 89.0% in 2017. The overall test for the trend in the employment status of health workers from 2008-2017 was statistically significant (p=0.001), indicating an overall increasing trend of unemployed health workers compared to employed health workers. Figure 3.9: Trends in employment status of health workers 2008-2017 Figure 3.10 shows health workers categorised by employment status. Among the various health worker categories, doctors had the highest proportion (97.0%) of employed individuals compared to other health categories. Mid-level health workers (13.0%) had the highest proportion of health workers who were unemployed as compared to other cadres. 7% 10% 7% 5% 6% 8% 8% 10% 13% 11% 93% 90% 93% 95% 94% 92% 92% 90% 87% 89% 0% 20% 40% 60% 80% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 Year unemployed employed 44 Figure 3.10: Breakdown of health worker categories by employment status 3.4.3 Employment industry Figure 3.11 presents trends in the employment industry of health workers from 2008 to 2017. The proportion of health workers working within the health industry increased from 85.0% in 2008 to 88.0% in 2012 and remained steady at 88.0% in 2017. The overall test for trends in the labour force industry of health workers from 2008-2017 was statistically significant (p=0.003). 4% 8% 13% 7% 97% 92% 88% 93% 0% 20% 40% 60% 80% 100% Doctors/Dentists Nurses Mid-level health workers Allied health workers unemployed employed 45 Figure 3.11: Industry distribution of health workers, 2008-2017 Figure 3.12 presents a description of health workers by their employment industry. Mid-level health workers (27.0%) and allied health workers (22.0%) had the highest proportion of health workers employed outside the health industry. Nurses (94.0%) had the highest proportion of health workers employed in the health industry. 15% 13% 14% 12% 12% 11% 10% 9% 11% 12% 85% 87% 86% 88% 88% 89% 90% 91% 89% 88% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 non health industry Health Industry 12% 6% 22% 27% 88% 94% 78% 73% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Doctors/Dentists Nurses Mid-level health workers Allied health workers non health industry health industry Figure 3.12: Breakdown of health worker’s categories by industry 46 3.4.4 Employment sector Figure 3.13 presents the trends of health workers employed in the public sector from 2008 to 2017. In 2008, 2012 and 2017, 59.0% of health workers were employed in the public sector while 41.0% were in the private sector. The overall test for trends in the employment sector of health workers from 2008-2017 was not statistically significant (p=0.243), indicating no change in the distribution over time. Figure 3.13: Trends in the distribution of health workers by employment sector, 2008-2017 Figure 3.14 shows the distribution of the health worker cadre by employment sector. The category with the highest proportion of health workers employed in the public sector was nurses, at 70.0%. There were more doctors/dentists (57.5%) and allied health workers (65.2%) in the private sector than in the public sector. 59% 59% 61% 62% 59% 62% 61% 64% 57% 59% 41% 41% 39% 38% 41% 38% 39% 36% 43% 41% 0% 20% 40% 60% 80% 100% 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 year Public sector Private Sector 47 Figure 3.14: Breakdown of health workers by employment sector, 2008-2017 3.4.5 Multiple job holding Table 3.4 presents trends in health workers who reported dual work or multiple job holding from 2008 to 2017. The majority (99.8%) of health workers reported that they were not engaged in multiple job holding. However, a small proportion (0.2%) reported having a second job; this slightly increased to 0.3% in 2012 before decreasing in 2017. The overall test for trends on multiple jobs held amongst health workers from 2008-2017 was statistically significant (p=0.001). 43% 70% 52% 35% 58% 30% 48% 65% 0% 20% 40% 60% 80% 100% Doctors/Dentists Nurses Midlevel health workers Allied health workers Public sector Private Sector 48 Table 3.4 :Trends in reported MJH among health workers, 2008-2017 Multiple job holding 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Total Yes 1 (0.2%) 1 (0.2%) 1 (0.2%) 1 (0.2%) 2 (0.3%) 2 (0.4%) 3 (0.6%) 1 (0.2%) 0 (0.0%) 0 (0.0%) 12 (0.2%) No 582 (99.8%) 510 (99.8%) 532 (99.8%) 547 (99.8%) 576 (99.7%) 532 (99.6%) 483 (99.4%) 440 (99.8%) 416 (100.0%) 411 (100.0%) 5029 (99.8%) Total 583 (100%) 511 (100%) 533 (100%) 548 (100%) 578 (100%) 534 (100%) 486 (100%) 441 (100%) 416 (100%) 411 (100%) 5041 (100%) 3.5 Demographic and occupational characteristics associated with employment status Table 3.5 and Table 3.6 present the results of the bivariate analysis associations of the demographic variables and the occupational variables respectively with employment status. All demographics and occupational variables were found to be significant at the 0.2 level and therefore included in the multiple regression. 49 Table 3.5: Association between demographic variables and employment status Sociodemographic Variables Unemployed health workers (n-461) Employed health workers (n=5041) p-value Freq. % Freq. % Age 18-25 44 13.88 273 86.12 <0.001 26-35 121 8.59 1287 91.41 36-45 86 5.17 1576 94.83 46-55 73 5.07 1367 94.93 56-64 137 20.30 528 79.70 Gender Female 403 9.02 4064 90.98 <0.001 Male 58 5.60 977 94.40 Population group African 327 9.22 3,220 90.78 <0.001 Coloured 68 8.92 694 91.08 Indian 11 4.09 258 95.91 White 55 5.95 869 94.05 Geography type Urban 415 7.99 4777 92.01 <0.001 Rural 46 14.84 264 85.16 Table 3.6: Association between occupational characteristics and employment status Occupational variables Unemployed health workers Employed health workers p- value Freq. % Freq. % Health Occupations Doctors/Dentist 19 3.46 530 96.54 <0.001 Nurse 261 7.89 3.046 92.11 Mid-level health workers 150 12.49 1051 87.51 Allied health workers 31 6.97 414 93.03 Industry Health industry 365 7.52 4486 92.48 <0.001 Non- health industry 96 14.75 555 85.25 Table 3.7 shows the results of the multiple logistic regression. The overall model of the logistic regression was statistically significant (p=0.000). The Archer-Lemeshow post-estimation goodness of fit test confirmed the final model to be a good fit (p=0.135). 50 The results showed that health workers aged 36-45 and 46-55 had significantly (p<0.05) higher odds of employment as compared to health workers aged 18-25 years, with odds ratios of [2.8 (CI 1.8-4.4)] and [2.7 (CI 1.6-4.1)], respectively. Additionally, health workers aged 56-64 had 0.5 times higher odds of employment compared to those aged 18-25, with a statistically significant p-value of 0.011 and a confidence interval ranging from 0.4 to 0.9. Furthermore, female health workers had significantly (p=0.003) lower odds of employment compared to male health workers, with an odds ratio of 0.5 (CI 0.3-0.9). The results suggest that age and gender are significant predictors of employment among health workers. Indian health workers were more likely to be employed than Black African health workers, but the association was not statistically significant (p > 0.05). However, White health workers had 1.5 times higher odds of employment compared to Black African health workers, with a statistically significant p-value of 0.040. In addition, Coloured health workers had lower odds of employment compared to Black African health workers, although the association was not statistically significant (p > 0.05). This suggests that there are differences in employment status among health workers of different population groups. Health workers from rural areas were 0.47 less likely to be employed compared to health workers from urban areas, this association was statistically significant with a p-value of 0.001. Furthermore, mid-level health workers had significantly lower odds of employment compared to doctors/dentists (OR=0.38, p=0.004). In contrast, the odds of employment for nurses and allied health workers did not significantly differ from those of doctors/dentists. Employment of Health workers outside the health industry was 0.53 less likely than those employed in the health industry which was statistically significant (p<0.001). Additionally, a significant (p<0.001) difference in the odds of employment between health workers employed in the health industry and those employed outside the health industry. Health workers employed outside the health industry had 0.53 times lower odds of employment compared to those employed within the health industry, and this association was found to be statistically significant. It can be inferred that there are notable disparities in the likelihood of employment among health workers, which can be attributed to multiple factors including the geographical setting, health occupation, and industry of employment. 51 Table 3.7: Multiple logistic regression of predictors of employment in health workers Predictors Adjusted odds ratio Confidence intervals [ 95%] p-value Age group 18-25 1 (ref) 26-35 1.81 1.18 - 2.76 0.006 ** 36-45 2.83 1.82 - 4.41 <0.001 *** 46-55 2.65 1.61- 4.14 <0.001 *** 56-64 0.56 0.36 - 0.87 0.011 * Gender Male 1 (ref) Female 0.56 0.39 -0.82 0.003 ** Population group African/Black 1 (ref) Coloured 0.91 0.63 - 1.30 0.617 Indian/Asian 1.98 0.88 - 4.46 0.097 White 1.52 1.01 - 2.27 0.040 * Geography type Urban 1 (ref) Rural 0.47 0.31- 0.71 <0.001 *** Health worker category Doctors/Dentists 1 (ref) Nurses 0.63 0.32 -1.22 0.175 Mid-level health workers 0.38 0.20-0.73 0.004 ** Allied health workers 0.72 0.35-1.47 0.374 Industry Health industry 1 (ref) Non-health industry 0.53 0.310- 0.71 <0.001 *** ***: p<0.001, **: p<0.01, *: p<0.05 52 CHAPTER 4: DISCUSSION 4.1 Introduction This was one the first studies to analyse the QLFS data on health workers in South Africa. This chapter begins with a summary of key study findings (Section 4.2). Section 4.3 discusses the socio-demographic findings followed by section 4.4 which discusses the findings on occupational and economic activities. The demographic, occupational and economic factors associated with employment are discussed in section 4.5. Section 4.6 concludes with the limitations and strengths of this study. 4.2 Summary of key results Trends in socio-demographic characteristics • Nurses constituted the highest proportion of health workers in the QLFS albeit with a decreasing trend. • The overall median age of health workers remained steady at around 41 years, with nurses being older at 44 years. • There were more female health workers than males, with the majority of nurses being female. • The results also showed significant imbalances in the distribution of health workers, with a concentration of health workers in urban areas as compared to rural areas. Trends in occupational characteristics • The majority of health workers reported that they did not have more than one job. • Most healthcare workers were employed in the health industry, and in the public health sector, but with variations by category. • The private sector had a higher concentration of doctors/dentists and allied health workers compared to the public sector, whereas nurses and mid-level health workers formed the largest proportion in the public sector. Factors associated with the employment of health workers: • Age and gender are significant predictors of employment among health workers. Health workers aged 36-45 and 46-55 had higher odds of employment compared to those aged 18-25 years. • Female health workers were less likely to be employed than male health workers. • White health workers had higher odds of employment compared to Black African health workers, whereas Coloured health workers had lower odds of employment compared to Black African health workers. • Health workers from urban areas were more likely to be employed than those from rural areas. • Mid-level health workers had lower odds of employment compared to doctors/dentists. • Employment of health workers outside the health industry was less likely than those employed in the health industry. • The variations in the likelihood of employment among health workers can be attributed to multiple factors, including geographical setting, health occupation, and industry of employment. 53 4.3 Socio-demographic characteristics This study highlighted an ageing health workforce, especially of nurses who were older compared to the other professions. The concerns about an ageing workforce of nurses in South Africa are well documented, with almost one in two professional nurses older than 50 years (35). There were very few younger doctors/dentists (18-25 years) in the surveyed health workforce as compared to those approaching retirement age (56-64 years). The 2007 study by Dal Poz et al. (59) found that there was a reduced number of younger doctors entering the health workforce. However, the study is more than a decade old. In South Africa, the lower proportion of younger doctors/ dentists in the OLFS could relate to funding constraints and inability of the public sector to absorb new medical and/or dental graduates. In this MSc study there were more female health workers as compared to males, thus supporting the findings of other studies on the feminisation of the health workforce (9;60). Furthermore, nurses constituted the majority of the health workers surveyed, which is dominated by females. This study also found that 53.0% of doctors/dentists were male. The findings of this study are similar to those of a study that analysed gender equity in 104 countries (38). The multi-country analysis found a predominance of females in the nursing and midwifery professions and a predominance of males in the medical, dental or pharmacy professions (38). The study findings suggest the need to enhance the gender balance across all health occupations to ensure the delivery of high-quality and equitable healthcare services to all people in South Africa (1, 6, 36). South Africa’s 2030 HRH Strategy highlights the increasing feminisation of the health professions and emphasises the need for gender-transformative policies and practice environments (1). Overall, this study showed that the majority of health workers surveyed were in urban areas compared to rural areas, increasing from 85.0% in 2008 to 88.0% in 2012, although it slightly decreased to 84.0% in 2017. Amongst the health worker categories, the findings showed that there were very few doctors/dentists (5.0%) and allied health workers (7.0%) in rural areas, whilst the proportions of nurses (18.0%) and mid-level workers (21.0%) were slightly higher. These findings are similar to those of prior studies on HRH in Africa and other low-middle- income nations, demonstrating that HRH are concentrated in urban areas (6, 23, 45). In India, a study found that 74.0% of health workers were in their urban areas (50). In South Africa, Zihindulai et al. (41) reported that 43.6% of the population living in rural areas have 12.0% of doctors and only 19.0% of nurses servicing them. The implications of the MSc study findings 54 are that policy intervention is needed to address the maldistribution of health workers between urban and rural areas, through a combination of incentives, and ongoing monitoring of the success of these interventions. 4.4 Occupational characteristics and economic activities The study found that the majority of health workers surveyed during the study period were nurses (60.0%), albeit with decreasing trends as the proportion of other health worker categories increased. This predominance of nurses is similar to the findings of studies in other African or low- and middle-income countries (4, 26, 34, 61, 62). The South African health system is primarily nurse-based, especially at the primary health care level. Doctors or dentists represent individuals with scare skills, and shortages remain a persistent challenge in many African countries (23, 38, 63, 64). In South Africa, the skills-mix vary across the nine provinces, with a higher concentration of doctors and dentists in the urban provinces of Gauteng, Kwa-Zulu Natal, and the Western Cape (1). In this study, the public sector was found to employ a higher proportion (59.0%) of healthcare workers compared to the private sector (41.0%). This trend is consistent with a study conducted in Cameroon (32), where the majority (65.9%) of healthcare workers were employed in the public sector. However, a greater proportion of doctors/ dentists (58.0%) and allied health workers (65.0%) were employed in the private sector, while nurses (70.0%) and mid-level health workers (52.0%) were dominant in the public sector. The inequitable distribution of healthcare workers between the public and private sectors in South Africa is well documented (7, 65), and it of concern, as the private health sector provides care to less than 20% of the South African population. The analysis of the OLFS provides evidence of the maldistribution of health professionals in South Africa, as Barron et al. (65) highlighted the limitation of health professions registers for obtaining reliable data on the sectoral distribution of health workers. The study findings point to the need for concerted policy Intervention to address the maldistribution of health workers between the public and private health sectors. This was also highlighted in the recommendations contained in the 2030 HRH Strategy (1). 55 4.5 Factors associated with the employment of health workers The multiple logistic regression demonstrated a significant association between age and employment, consistent with the findings of Dunga and Sekatane (66). Specifically, individuals in the age groups of 36-45 years and 46-55 years were more likely to be employed, possibly due to the availability of employment opportunities in the health sector for adults with more work experience. However, there is a need for concerted efforts to recruit and retain early- career health professionals as they are the future health workforce (2, 67). Additionally, a younger health workforce is meet the health needs of an ageing population (28), and overcoming the potential crisis created by the combination of an ageing population and an ageing health workforce (29). Notwithstanding a predominantly female workforce, women were 0.56 times less likely to be employed than men, suggesting a combination of structural and relational barriers to women's engagement in the health labour market, as has been found in other studies (3, 4, 5, 30, 37). South Africa has robust gender equity legislation and policies (73). The study findings highlight again the need for a gender-transformative health system and the implementation of the recommendations of the 2030 HRH Strategy (1). This study found that health workers were less likely to be employed in rural areas than urban areas. This finding is similar to a study in Peru (26) that found that physicians and nurses were 5 times and 14 times more likely to be employed in urban areas. A combination of factors influences rural employment, including salary differences, resource availability that enable health workers to provide quality care, safety of health workers, and lack of basic amenities such as accommodation and schools for children (68). The unequal distribution of health workers between urban and rural areas presents a significant challenge to healthcare access and delivery in South Africa, highlighting