Working life expectancy of physicians: the case of primary care physicians in Czechia

Abstract

Background

The decrease in the number of healthcare workers and the resulting deterioration in healthcare quality and availability have been subjected to intensive discussion in Czechia in recent years. Estimating future healthcare worker capacities requires a detailed analysis of their “movement” within the healthcare system. This study focuses on exits of the primary care physicians from the healthcare system in Czechia.

Methods

Using anonymised data obtained from the largest Czech health insurance company (2012–2022), we constructed working life tables and calculated working life expectancy, which indicates the expected average number of remaining years of work at the exact age of the physician. The study focuses on primary care physicians, who are crucial for the effective functioning of the healthcare system.

Results

At age 50, working life expectancy was 20 years for female physicians and approximately 21 years for male physicians. Over the monitored period, working life expectancy decreased by 1 year for both genders. Gynaecologists had the longest working life expectancy, while dentists had the shortest.

Conclusions

The decrease in the working life expectancy and the length of tenure indicates the need to create favourable conditions for the extension of the working lives of physicians to avoid early exits from the system.

Background

One of the manifestations of population ageing that is currently faced by most developed countries concerns an increase in demand for health services [1], which is often accompanied by a decrease in the supply of such services due to the ageing of the healthcare workforce (HWF) [2]. This trend is also evident in Czechia and forms the subject of this article.

While the availability of a sufficient number of healthcare workers is essential for the functioning of health services, this issue has not, to date, received sufficient attention in Czechia. Previous analyses have focused primarily on the availability of health services, concerning which the workload of healthcare workers comprised only one of several factors. The performance of a detailed analysis of healthcare worker movements within the system (entries and exits) was not the main objective of these analyses. Estimates of the future development of the number and age structure of doctors in Czechia have been published only for selected medical specialisations due to data limitations [3]. The improvement in the structure of data in recent years, however, allows for the enhancement and refinement of such analysis (details are provided in the Methods chapter).

Despite improved data options, there remains a paucity of studies on the estimation of the HWF for Czech healthcare service staffing planning purposes. One of the exceptions in this respect concerns estimates of outgoing general practitioners (GP) up to 2035 from the regional perspective [4]. Moreover, whereas many studies on HWF planning have been published in other countries [e.g., 5–9], they are mainly of a theoretical nature and do not provide comprehensive projections supported by detailed data. Most countries have very limited access to data on healthcare workers in terms of their specialisation, age and geographical distribution [5, 10]. Only a small number of developed countries, e.g., the Netherlands, Norway and Finland, have created comprehensive models for the projection of the HWF [11].

Certain initiatives are being adopted in Czechia aimed at addressing these gaps, including participation in the European JA HEROES project, which focuses on improving health workforce capacity planning [12]. One of the aims is to develop effective tools and methods for HWF planning purposes. This study aims to contribute to this aspect, specifically with concern to refining planning methods, which depends on forming a detailed understanding of the age and gender structure of the workforce and mapping the movement of workers within the system. Demographic approaches, such as population projection and demographic analysis, are particularly useful for estimating entries and exits to/from the healthcare system [11].

This article specifically addresses the exit of primary care physicians from the healthcare system by applying the demographic method of life tables so as to construct working life expectancy tables for physicians. The output of the tables is the working life expectancy ($WLE$) indicator representing the expected average number of remaining years of work at exact age (details of calculation are provided in the Methods chapter). Due to the availability of unique detailed and accurate data, the paper applied the cohort (generational) perspective, which comprises both an innovative and useful approach to estimating health worker outcomes, to track workers over time. The analysis considered primary care physicians, who are crucial in terms of the efficiency of health systems and the quality of health services globally, since they resolve 90% of cases or refer patients to the appropriate specialised level of care. [13, 14]. Retaining these highly demanded professionals is becoming increasingly challenging in view of their increasing workloads and deteriorating working conditions following years of underinvestment [14].

Primary care in Czechia does not hold the status it deserves within the healthcare sector, which is reflected in the overall performance of the system. Unlike some countries, Czechia lacks a gatekeeping system that prevents the access of patients to specialists without referral by a GP, a system that acts to reduce the overall use of, and expenditure on, health services. In general, a robust primary healthcare system serves to improve the health and satisfaction of patients and to reduce healthcare costs, thus its availability should be a priority for developed nations [15, 16].

This is linked to the fact that the capacity of outpatient specialists in Czechia is significantly higher than that of GPs, especially in larger cities [17, 18]. This specialist-heavy model contributes to increasing the workload of specialists and a shortage of the comprehensive care typically provided by GPs, which is unsustainable under the conditions of the current system. However, efforts are being made to change the situation and to make better use of the potential. One of the main objectives of the “Health 2030 strategy” is to reform the primary care system. In addition to introducing a gatekeeping system and thereby emphasising the importance of GPs, the strategy will aim to reduce administration [19].

In terms of demographic ageing, the age structure of primary care physicians is the most negatively loaded structure in the healthcare system due to the comparatively lower prestige and attractiveness associated with the profession [14, 20]. The average age of primary care physicians in the EU is around 50–53 years [21, 22], whereas the average age of GPs in Czechia is 55 years, paediatricians over 57 years, dentists 48 years and outpatient gynaecologists 56 years [18]. The main challenge for EU states, particularly Czechia, is, therefore, to ensure the smooth generational change of primary care physicians [23, 24].

The analysis of the exit of physicians is crucial in terms of constructing reliable projections of future healthcare capacities. It is necessary to determine the $WLE$ of physicians and how it changes over time. For example, in the Netherlands and France the projection model considers that some doctors will continue to work even after reaching retirement age [25, 26]. However, in many cases, exits occur before this age due to factors including working conditions and legislative changes [11, 27].

In Czechia, retirement entitlement varies by the year of birth, and for those born before 1965, by gender and the number of children raised. For those born after 1965, the retirement age is set at 65 years regardless of gender or family size. Men born before 1965 retire at an age that is reduced by 2 months per year from the set age of 65 years. The retirement age of women born before 1965 is lower if they had children; for example, women born in 1950 who had two children retired 4 years earlier than men, while women born in 1960 who had two children retired 2 years earlier than men [28].

An increase in the retirement age to a maximum of 67 years is currently being considered. Early retirement is permitted up to 3 years before the retirement age with a 1.5% pension reduction per each 90 days. Most countries worldwide are concerned about the early exit of workers from their healthcare systems and are, therefore, endeavouring to improve working conditions, especially for nurses and auxiliaries who perceive a deterioration in their working conditions [29] accompanied by increased stress levels, which have been largely attributed to the impact of the COVID-19 pandemic, which in Canada, for example, led to a large number of GPs leaving the system [30]. Adverse working conditions are strongly linked to early retirement [31]; moreover, challenges such as increased multimorbidity, inadequate financial compensation and the introduction of eHealth systems (e.g., mandatory e-prescriptions in Czechia since 2018 [28]) are most likely contributing to the early exit of primary care physicians [27].

This study applies demographic analysis methods (working life expectancy tables) to identify trends in the intensity and timing of primary care physicians exits from the healthcare system in Czechia. The study focuses on whether the $WLE$ of physicians has increased in response to mounting pressure to continue working due to the declining availability of primary care and whether $WLE$ has been negatively impacted by external factors, such as the COVID-19 pandemic and legislative changes (e.g., concerning eHealth). The analysis further examines the extent to which $WLE$ varies by the gender and specialisation of the physician.

Data and methods

Data source

Czechia has relatively detailed data sources on healthcare workers. The Institute of Health Information and Statistics of the Czech Republic (IHIS CR) maintains the main general information register. Although the National Register of Healthcare Workers (NRZP) and the National Register of Reimbursed Health Services (NRHZS) include information on the HWF, neither register is currently suitable for planning purposes, due to incomplete or out-of-date data (NRZP) or a lack of historical data (NRHZS—no suitable data are available for HWF planning prior to 2023). However, these resources will become increasingly relevant for this purpose going forward.

Therefore, an alternative data source was used, i.e., the basic data set on primary care physicians as maintained by the General Health Insurance Company of the Czech Republic (GHIC), based on a cooperation agreement with the authors’ research organisation. The GHIC is the largest Czech health insurance company and has contractual agreements with almost all the country’s providers (almost 100% in the GP segment); hence, we had an almost complete source of data at our disposal [32]. The fact that around 60% of the population is covered by this insurance company did not affect the results in this article, which focuses on providers. Currently, no more comprehensive data is available for the monitored period in Czechia than that used in this analysis. The data set consisted of contractual individual anonymised data on primary care physicians by gender and age for the period 2012–2022 (status as of 31 December). Primary care physicians comprise general practitioners (GP), general paediatricians (PED), general dentists (DENT) and outpatient gynaecologists (GYN) (see Appendix 1 for data description).

Calculation and smoothing of exit probabilities

This article aimed to analyse the exit of primary care physicians from the healthcare system. At the same time, the objective was to observe both differences in physician attrition over the years as well as between primary care physicians of different specialties. To accomplish the objectives, methods based on constructing so-called life tables (a basic analytical tool used widely in demographic analysis [33, p. 32, 34]) was used. Specifically, we applied the method of single-decrement life tables. Due to the lack of detailed data, we did not consider multi-state tables, which are based on more detailed estimates, because they decompose the overall life expectancy into segments of life span by different states [35].

The main output of the life tables concerns the potential for the calculation of the $WLE$. This indicator has previously been calculated abroad, but only for the entire working population, i.e., not specifically for healthcare workers [36, 37].

The main input for the tables of doctors who leave the system consisted of the calculation of the probability of exit from the healthcare system due to retirement, death, emigration, moving to a different segment of the labour market during a given year or other reasons. Using the detailed data on individual healthcare workers, it was possible to apply the direct probability calculation method [33, p. 32], which can be simplified as follows:

$${}_{t}^{c} q_{x}^{x + 1,g,s} = \frac{{{}_{t}^{c} O_{x}^{x + 1,g,s} }}{{ {}_{31.12.t-1}{}^{c}{P}_{x}^{g,s}}}$$

where ${{}_{t}^{c}{q}_{x}^{x+1,g,s}}$ is the probability of exits from the system during year $t$ between ages $x$ and $x+1,$ ${{}_{t}^{c}{O}_{x}^{x+1,g,s}}$ represents the number of doctors who exit the system in year $t$ (based on the tracking of the presence of individual workers on the list of doctors at both the end of year $t-1$ and the end of year $t$; if the worker is not listed in year $t$, they are considered to have exited the system) and ${}_{31.12.t-1}{}^{c}{P}_{x}^{g,s}$ is the number of doctors at the end of year $t-1$. $C$ is the generation (the year of birth) of the doctors, $g$ is the gender and $s$ is the specialisation.

Prior to constructing the tables, it was crucial to establish the age at which the exit of doctors from the system becomes significant. The length of the period of education is an important factor. General medicine studies last at least 6 years in Czechia, followed by specialised training for approx. 4.5 years. Practical dentists are permitted to open their own practices after completing the basic 5-year dentistry study period. After including basic practice and the combination of family and working life (especially for women), physicians typically fully enter the system at around the age of 35, with inputs significantly exceeding outputs up to around the age of 40. Between the ages of 40 to 50, the turnover of doctors is low, as confirmed by the average net workforce change by age in the period 2013–2022 (Fig. 1). We set the age at which exits begin to outnumber entries at 50. The balance values are close to zero for both men and women from age 50 to 57, whereupon the number of exits begins to predominate, especially for women.

Fig. 1

(Source: [32]; author’s own calculations)

Average net workforce change by age of primary care physicians, Czechia, 2013–2022. Net workforce change = the difference (balance) between the number of physicians entering and exiting the system during the year per 1000 physicians. WMA 9 = the 9-period weighted moving average (see the calculation below)

To ensure the robustness of the data, the exit probabilities were calculated for each year from 2013 to 2022 and then averaged. Since the probabilities by age were observed to fluctuate considerably due to the low number of events, especially at the highest ages, the values were subsequently smoothed. Due to the development of the exit probability curve, the use of multiple smoothing methods by age was considered the most suitable approach. Up to the age with enough relatively stable observations, smoothing was applied applying the 9-period weighted moving average (WMA 9). Period 9 was chosen so as to avoid considering all the minor fluctuations. The smoothed probability using the weighted moving average followed the formula:

$$q_{x}^{{{\text{smoothed}}}} = \frac{{w_{0} *q_{x} + w_{1} *\left( {q_{x - 1} + q_{x + 1} } \right) + \cdots + w_{4} *\left( {q_{x - 4} + q_{x + 4} } \right)}}{{w_{0} + w_{1} + \cdots + w_{4} }}$$

where ${q}_{x}$ is the exit probability and $w$ are the weights.

The weights were symmetrical around the middle value; moreover, the middle-balanced value was assigned the highest weight, with decreasing weights on either side. The basic condition was that the sum of the weights was equal to one. The weighting values were calculated using the least squares method based on the formula [38, p. 34]:

$$w_{j} = \frac{3}{{4p\left( {p^{2} - 4} \right)}}*\left( {3p^{2} - 7 - 20j^{2} } \right),\quad {\text{where}}\;j = - m, - m + 1, \ldots - 1,0,1, \ldots m - 1,m;m = \frac{p - 1}{2}$$

where $p$ is the period (the length of the moving average).

Concerning the higher ages, instead of using moving averages, the exit probabilities were smoothed using the Gompertz–Makeham function, often used in the calculation of life tables to smooth the age-specific probabilities of death at higher ages [39]. This function was calculated using the King–Hardy method, the key issues of which concern the appropriate selection of the input parameters—the initial age (${x}_{0}$) and the length of three consecutive intervals ($k$) [40]. The appropriateness of the parameters was first visually evaluated and then verified statistically by calculating the Akaike Information Criterion (AIC).

The interval length ($k$) was most often chosen at between 5 and 6 years. The starting age was set at 66 years for female primary care physicians and 68 years for males (Table 1). The maximum age for women was set at 88 years, with the exit probability set at 1 at the age of 89, i.e., we assumed that doctors would not work beyond age 90 (although rare cases were observed). The highest age for men was determined at 1 year higher than for women. Moreover, the highest ages featured low numbers of doctors; hence, the smoothing method selected for exit probabilities was considered sufficient. The value of the smoothed probability of exits from the system concerning the highest ages for the period 2013–2022 for the considered specialisations was determined as being in the range 0.20–0.25.

Table 1 Input parameters ${x}_{0}$ and $k$ for the Gompertz–Makeham function.

Construction of working life tables

Working life tables were compiled based on the exit probabilities. The output of the tables comprised the following working life table functions:

• ${l}_{x}$: the table number of physicians who continue to work at exact age $x$ [of a hypothetical initial cohort of 100 000 at the age of 50 (${l}_{50}$)]

• ${d}_{x}$: the table number of physicians who exit from the system between exact ages $x$ and $x+1$

• ${L}_{x}$: the table number of person-years spent working between the exact ages $x$ and $x+1$

• $WLE$: the working life expectancy—the expected average number of remaining years of work at exact age $x$ [36, 37]; to represent the degree of uncertainty in the results, confidence intervals were calculated [41].

These functions described the age-specific exits of doctors over the 10-year period analysed. Aimed at identifying the changes in the timing of exit over the years, the study period was divided into three intervals, i.e., 2014–2016, 2017–2019 and 2020–2022. The intention was to determine whether COVID-19 exerted an impact on the exit probability in the period 2020–2022 or whether the change occurred between previous periods in which a change in legislation occurred (for example, concerning the e-health system and the associated increased administrative burden, see above).

Previously, the average retirement age comprised the main indicator of the end of productive life. Today, however, as many seniors continue to work part-time after retirement age, the working life expectancy ($WLE$) was considered a more appropriate indicator.

Calculation of other indicators based on working life tables

Other indicators considered in the working life tables served for the analysis of the exit compression process (the reduction in the variability of the exit age) and the rectangularisation process (whether exits are moving to older ages).

The key indicator in terms of evaluating changes in the exit age distribution comprised the mode ($M$) of the table number of physicians exiting from the system, also referred to as the modal exit age. It is known that most exits occur at around the age at which the ${d}_{x}$ is at the maximum.

The distribution of exits by age is primarily described using quartiles. For the purposes of this study, the upper quartile (75th percentile) indicates the age at which 75% of the hypothetical cohort of 100 000 are likely to exit, while the lower quartile (25th percentile) represents the age at which 25% of this cohort exit from the system. The interquartile range ($IR$), calculated from the upper and lower quartiles, indicates the number of years in which half of the hypothetical cohort will exit after one quarter of the generation has already exited [42].

The main indicator of the rectangularisation of the ${l}_{x}$ curve comprises the $FR$ (fixed rectangle), which specifies the proportion of the possible 40 years for women and 41 years for men spends at work a working person at the age of 50, maintaining the exit pattern. An increasing $FR$ value indicates the rectangularisation of the curve [42, p.17]:

$${FR} = \frac{{\mathop \sum \nolimits_{50}^{89} L_{x} }}{{l_{50} }}*\frac{1}{100}$$

With concern to the overall monitoring of the available work capacity, it is important to track not only the physical number of staff but also their full-time equivalent (FTE). The Czech primary care system sets 1 FTE according to the number of office hours; 1 FTE = 25 contracted consultation hours for general practitioners, 30 h for outpatient gynaecologists and 35 h for dentists. The data set included information on the FTE of physicians, which allowed for the calculation of average FTE values for each specialty by age and gender from 2013 to 2022. These values were subsequently smoothed applying a moving average.

Results

The probabilities of exits from the healthcare system, which subsequently formed the basis for the working life tables, were calculated by age and gender. First, the overall exit probabilities for primary care physicians were calculated and, subsequently, the differences between the considered primary care specialisations were evaluated. This was followed by the analysis of the development of the exit of doctors from the system over time.

On average, the exit probability begins to increase significantly from around the age of 60 (Fig. 2); women exit from the system on average slightly earlier than men, which is strongly influenced by the differing retirement age of men and women in Czechia. The increase in the probability of men exiting the system is slower but more constant than that of women. Around the age of 75, the exit probability is equal for men and women. Women exhibit a more moderate exit probability increase rate than men at the highest ages.

Fig. 2

(Source: [32]; author’s own calculations)

Probability of exits from the system by age and gender, Czechia, 2014–2022

Changes in the exit probability over time represents a further important aspect. For comparison purposes, the probabilities were calculated for 3-year intervals (2014–2016, 2017–2019, and 2020–2022). A more significant change in the exit probability by age was recorded for women over the given periods than for men. The 62–66 age group saw an increase in the exit probability over time, whereas the younger age group evinced a decrease in the period 2020–2022. Concerning the highest ages, the exit probability increased the most in the periods 2014–2016 and 2017–2019, with a higher increase among women than men. This may have been linked to the introduction of compulsory electronic prescriptions in 2018, which may have prompted older doctors to close their practices. In addition, the overall increase in the administrative burden may also have contributed to this trend. These findings are in line with a 2018 survey of general paediatricians [43], which identified legislative requirements, economic factors and the administrative load to be among the primary reasons for the closure of practices, in addition to age.

We concluded that the COVID-19 pandemic contributed to an increase in the exit frequency of women doctors at around the retirement age (from approx. 60), whereas no significant differences were noted in this respect in the given period for male doctors. The exit probability differed with respect only to the highest ages, but as mentioned previously, this indication may have been due to the smoothing bias.

The $WLE$ at the age of 50 was on average 19.8 (95% CI 19.5–20.1) years for female primary care physicians and 21.0 (95% CI 20.6–21.5) years for their male counterparts in the period 2014–2022 (Table 2). The expected average number of remaining years of work at age 50 decreased over this period by more than 1 year for both genders. The $WLE$ at age 50 for women in the period 2014–2016 was 20.6 (95% CI 20.0–21.3) years; however, it stood at 1 year less in the following period and stabilised at this level moving into the third period. The $WLE$ decreased gradually for men by approximately 0.5 years between the defined periods. The assumption that doctors would stay in practice longer as a result of the shrinking availability of healthcare services was not proved.

Table 2 Indicators calculated based on the working life tables by gender, Czechia, 2014–2022.

The age at which most doctors left the system, i.e., the modal exit age, was 67.3 years for women and approx. 2 years higher for men. This age decreased over the given periods for women; the most significant decrease occurred in the third period. Whereas in the period 2014–2019, the modal exit age was around 68.0 years, it fell by 3 years over the period 2020–2022, which may have been due to the COVID-19 pandemic. The opposite trend was evident for men, i.e., the modal exit age increased over this period. However, it can be observed from the development of the table number of physicians who exit from the system between exact ages, that concerning 2020–2022, increases in exits are evident for the lowest ages and for the 65th and 72nd years compared to the previous years. A general decline in exits is evident for all the other ages over the given period (Fig. 3).

Fig. 3

(Source: [32]; author’s own calculations)

Table number of exiting physicians by age and gender, Czechia, 2014–2022

The calculation of the interquartile range revealed that in the periods 2014–2016 and 2017–2019, the length of the interval in which 50% of the hypothetical cohort of physicians would exit the system after the first quarter of a generation shortened, thus suggesting the compression of the exit of physicians from the system. However, a slight decompression was observed in the third period.

The final indicator used to evaluate changes in the exit distribution by age comprised the fixed rectangle, which served for the assessment of changes during the development of the function of the table number of physicians who continued to work at exact age $x$. This indicator examined whether exits were shifting to older ages, i.e., whether the ${l}_{x}$ curve was becoming rectangular. An increasing value would indicate the rectangularisation of the ${l}_{x}$ curve; however, in this case the opposite process is evident. The number of physicians who continued to work after the age of 60 slightly decreased over the period for both genders, as can be seen in the graph (Fig. 4).

Fig. 4

(Source: [32]; author’s own calculations)

Table number of physicians who continue to work by age and gender, Czechia, 2014–2022

The assessment of the exit probability by age for the various primary care specialisations revealed several differences (Fig. 5). For example, the exit probability of gynaecologists increased relatively slowly at younger ages then increased significantly from the age of 65. In contrast, the age of the commencement of the exit probability of dentists was the lowest of all the specialisations, i.e., 57 for women and 60 for men.

Fig. 5

(Source: [32]; author’s own calculations)

Probability of exits from the system by age, gender and specialisation, Czechia, 2013–2022

The expected average number of remaining years of work at age 50 also varied between the primary care specialisations (Table 3). Gynaecologists evinced the highest number of remaining years—23.1 (95% CI 21.9–24.5) years, while dentists evinced the least—18.4 (95% CI 18.0–18.9) years; the other specialisations ranged from 19.8 to 20.6 years. Paediatricians were found to have the lowest $WLE$ for men; however, this indicator was influenced by the low number of male paediatricians and the consequent significant related fluctuations. The level of uncertainty in the result was represented by the confidence interval. Male general paediatricians numbered 330 on average over the studied period, which represented just 15% of all general paediatricians.

Table 3 Average working life expectancy of primary care physicians at age 50 by gender and specialisation, Czechia, 2013–2022.

A further important aspect that must be considered when evaluating the $WLE$ of doctors concerns their working hours in terms of the full-time equivalent (FTE). The average FTE for female doctors decreases at the beginning of their careers and increases so as to reach the maximum, i.e., more than 0.9, between the ages of 45 and 65 (Fig. 6). The average FTE of male doctors, however, is relatively constant (above 0.8) from the beginning of their careers up to approximately the age of 50, and only then does it begin to increase to a maximum of above 0.9 between the ages of 57 and 61. After reaching the maximum level, sharp declines are evident for both genders—at the age of 80, the average FTE is around 0.7, and 0.4 at the age of 90. Thus, many doctors continue to work beyond the retirement age in combination with a gradual reduction in their working hours. This information is crucial in terms of the planning of future workforce requirements and the discussion on the continued participation of physicians who are nearing retirement age so as to ensure sufficient levels of healthcare provision capacity.

Fig. 6

(Source: [32]; author’s own calculations)

Average FTE by age, gender and specialisation, Czechia, 2013–2022. FTE = Full-time equivalent. The lower average FTE recorded for gynaecologists was due to the frequent overlapping of the performance of work in the obstetric–gynaecological departments of inpatient service providers

Considerable differences were evident in terms of the full-time equivalent between the various specialisations due to the nature of their work. It is, therefore, important to consider that although GYN physicians have the highest ${WLE}$, they also have the lowest FTE. Conversely, although dentists have a lower $WLE$, they work a high FTE from the beginning of their careers. The average FTE changed only slightly over the studied period.

Discussion

It was considered important in this context to compare the working life expectancy of primary care physicians with that of the general working population. According to Eurostat [44], Czechs generally work for fewer years than their counterparts in Western European countries, which is most likely due to length of the educational process and the age of retirement. Primary care physicians, therefore, generally work for more years than the average working population, often beyond the retirement age.

Unfortunately, no studies have been conducted that track changes in the $WLE$ of physicians. However, a systematic review of the retirement planning of physicians [45] determined that they typically retire at around age 69, which is on average 3 years later than the general population. The delaying of retirement is influenced by a number of factors including flexibility surrounding, and the intensity of, working hours, the intrinsic value, relationships with co-workers etc. Indeed, working hours flexibility and the option to reduce working hours with increasing age is an advantage that is not enjoyed by most of the working population, particularly those with lower educational qualifications. The increased take-up of part-time work at older ages is consistent with our finding that, following peaks in the FTE at around age 60, the average FTE for all primary care specialisations gradually declines. Other studies have also highlighted that GPs have fewer patients in their final years of practice [46, 47].

Studies that have focused on the $WLE$ according to educational level attained have revealed that those with higher educations have the longest working careers [48, 49]. The $WLE$ of those with low education levels is more than one quarter shorter than those with high educational attainment [50]. Moreover, those with lower socio-economic status evince lower $WLEs$ and more years of working life lost than those with higher socio-economic status [51]. Moreover, the findings of the Czech Labour Force Survey revealed that the higher the education level, the higher the proportion of people who work even after reaching retirement age [52]. Therefore, our results are in line with studies that indicated that physicians work on average more years than the rest of the working population.

Regarding the explanation of the gender gap in $WLE$, one of the factors why women in their 50 s have fewer years left at work than men may be that they are more likely to be caring for their relatives [53, 54]. In Czechia, this trend is also observed in relation to the lack of availability of daycare for children and homecare for the elderly [55]. Another factor may be, that women felt less inclined to work after age 65 if they felt superfluous at work, and if they had a positive attitude to the private sphere [53]. The question is whether, as in Sweden, the gender gap will narrow in the coming years [53].

The differences in $WLE$ between medical specialisation have no simple explanation. Even in studies that have looked at the retirement age of physicians, there is no consensus. In the UK, GPs were found to retire earlier than doctors of another specialisation [56]. They also found that GPs had higher burnout rates than other specialists [57]. This may also explain our findings that GPs have a lower $WLE$ than gynaecologists. However, there are studies that observe the opposite trend or report no differences in retirement age between specialties [46]. Future research should investigate this issue further, with a focus on other specialties.

In terms of the change in working life expectancy over the years, the $WLE$ for the whole population in Europe is gradually increasing, except for slight decreases in 2020 and 2021, i.e., 0.5 years for women and 0.3 years for men. Concerning primary care physicians, we also observed a decrease in $WLE$ between 2020 and 2022, which was attributed to the COVID-19 pandemic, which led to early departures due to overwork, increased stress and the worsening of working conditions [22, 58]. However, compared to the general working population, a decrease was also observed for primary care physicians in previous years.

While the age at which most people in the average workforce leave the labour market has increased in parallel with the increasing retirement age [52], the physician populations considered in this study evinced opposite trends, which suggests that their exit from the system is related less to reaching the retirement age than for the average working population. However, many factors lead to some doctors leaving the system earlier than others, including excessive workload, burnout syndrome, unhealthy work–life balance, technological advances and the emergence of new treatment and diagnosis methods [43, 45, 58]. In recent years, the impact of the COVID-19 pandemic has also been widely cited in relation to worsening working conditions. However, other factors lead to poorer working conditions, e.g., the increasing complexity of healthcare (increases in multimorbidity), inadequate financial remuneration and the ever-increasing administrative burden [27]. The introduction of eHealth innovations (electronic medical records) may also have influenced the exit of physicians [59, 60]. In Czechia, for example, the obligation to prescribe prescriptions only electronically has been in force since 1 January 2018 [28], a development that is thought to have impacted the increase in the departure of older physicians.

In response to concerns about the increase in early retirement, attention should be devoted to improving the working conditions of health workers so as to ensure their retention in the system. While the pandemic led to a number of improvements, e.g., the increased use of teleconsultations, there is still a long way to go [27].

In addition, the modification of the role of older GPs should be considered, e.g., by reducing their workload, the mentoring of younger colleagues in group practices, etc., especially since older ages and long lengths of practice have been associated with lower quality care (especially concerning demanding specialisations, e.g., dentists) [61].

Despite the finding that in recent years the $WLE$ of doctors has decreased, unlike the general population, doctors generally work beyond the retirement age, which should be considered in the resulting projection models as it is in other countries, such as France and the Netherlands [25, 62].

Since women end their working lives earlier than men, it is also necessary to consider the increasing feminisation of the healthcare sector when making projections. The share of women doctors increased in OECD countries between 2000 and 2019 from 40 to 49%. Moreover, Czechia has traditionally had one of the highest representations of female doctors in Europe; in 2019, 56% of all Czech doctors were women and up to 85% were general paediatricians [17].

Limitations

It is important to consider that the results, especially those relating to specific time periods and specialisations, may have been affected by the fluctuating and generally low numbers of observations at the highest ages, concerning which it is very important to choose the most appropriate model in terms of smoothing the probability of the leaving curve. We attempted to apply the most relevant statistical approaches and demographic analysis methods in this study so to avoid potential inaccuracies, including the calculation of confidence intervals to represent the degree of uncertainty in the results.

In addition, the results were compiled at the national level (for Czechia as a whole). We are aware that the resulting exit probabilities vary at the regional level due to the level of urbanisation, connectivity to other health services, the economic conditions in the region, etc. While regional differences and the factors that influence the intensity of exiting the system were not the focus of this paper, they nevertheless potentially provide ground for the conducting of further follow-up studies.

Conclusion

The analysis of the exit probability of doctors who work in the primary healthcare sector revealed several key findings crucial for estimating the future capacities of health workers. The exit probability for women begins to increase at the age of 57, i.e., approx. 3 years earlier than for men, partly because the working life expectancy ($WLE$) at age 50 is around 1 year lower for women. On average, primary care physicians have a $WLE$ at age 50 of around 20 years, which is higher than the average working population. The $WLE$ decreased by 1 year for both genders in the period 2014–2022, which may have been related to the COVID-19 pandemic and the introduction of compulsory electronic prescriptions. The specialisation also impacts $WLE$; general gynaecologists were observed to have the longest $WLE$ (23.5 years) and dentists the shortest $WLE$ (18.5 years). However, it is important to consider the shorter working hours of the former and, in general, the slightly decreasing trend in the FTE in recent years.

In conclusion, it can be stated that the slight decrease in the $WLE$ of primary care physicians is unfavourable due to the ageing of the healthcare workforce and the level of expected exits from the sector going forward. The objective should, therefore, be to support those who wish to continue working and to create the conditions that prevent premature exits, such as reducing the administrative burden or introducing physician group practices. The continued analysis of the exit of doctors and the monitoring $WLE$ trends is essential. Moreover, attention should also be devoted to other healthcare sector fields, especially nursing staff amid concerns about their high turnover due to the ageing of the workforce.

Appendix 1: Number of primary care physicians by age, gender and specialisation, Czechia, 2012–2022 (status as of 31 December)

	Age	25–34	35–44	45–54	55–64	65–74	85 +	Total
Females
GP	2012	219	647	863	1276	480	67	3552
	2015	289	630	718	1401	545	76	3659
	2018	328	693	793	1221	620	129	3784
	2022	379	902	868	834	790	196	3969
PED	2012	31	202	557	872	256	21	1939
	2015	52	191	445	847	340	24	1899
	2018	72	203	380	749	410	56	1870
	2022	83	312	338	550	492	67	1842
DENT	2012	797	628	696	2020	429	23	4593
	2015	1019	639	647	1826	553	33	4717
	2018	1226	680	726	1213	833	63	4741
	2022	1157	1037	687	624	1008	103	4616
GYN	2012	142	252	206	185	127	14	926
	2015	129	262	211	207	129	24	962
	2018	102	318	240	204	133	37	1034
	2022	88	331	314	205	140	48	1126
Males
GP	2012	109	282	533	847	305	77	2153
	2015	129	293	422	877	355	79	2155
	2018	138	300	431	725	468	80	2142
	2022	180	319	413	550	584	109	2155
PED	2012	6	45	92	143	33	9	328
	2015	8	30	78	146	56	9	327
	2018	8	30	67	133	80	9	327
	2022	16	27	58	106	100	9	316
DENT	2012	480	355	325	882	268	35	2345
	2015	613	437	317	749	347	58	2521
	2018	694	542	337	502	500	77	2652
	2022	637	676	380	313	545	78	2629
GYN	2012	58	231	330	405	153	31	1208
	2015	53	175	312	437	195	30	1202
	2018	39	157	293	382	269	41	1181
	2022	36	144	246	334	312	75	1147

1. Source: [32]; author’s own calculations
2. GP general practitioners, PED general paediatricians, DENT general dentists, GYN outpatient gynaecologists