Workforce crisis in primary healthcare worldwide: Hungarian example in a longitudinal follow-up study.

OBJECTIVE: The study was designed to explore the development of the general practitioner (GP) shortage in primary care and its characteristics in Hungary.
DESIGN: Longitudinal follow-up study over the decade 2007-2016.
METHODS: Analyses were performed on changes in number, age and sex of GPs by practice type (adult, paediatric and mixed), as well as on their geographical distribution and migration between areas characterised by deprivation index (DI) at municipality level. The association between deprivation and vacancy for GPs was studied by risk analysis. The number of population underserved was defined by DI quintile. SETTING AND
SUBJECTS: The study involved all general practices and GPs in the period examined. MAIN OUTCOME MEASURE: It is showed that the number of general practices with unfilled GP posts was increasing exponentially, mainly in the most deprived areas of the country.
RESULTS: A decrease in the number of GPs in all types of practices, especially in mixed (by 7.7%; p<0.001) and paediatric (by 6.5%; p<0.001) ones, was shown; the number of adult practices with unfilled GP posts doubled, while the number of paediatric practices with a vacancy for a paediatrician more than tripled. The average age of GPs was increased by 3.7 years (p<0.001) in adult, by 5.4 years (p<0.001) in paediatric and by 4.2 years (p<0.001) in mixed practices. In 2007, 52.27% (95% CI 51.03 to 53.5) of the GPs were women, and this rate increased to 56.19% (95% CI 54.93 to 57.44) by the end of the decade. An exponential association between relative vacancy rate and deprivation was confirmed. As a result of the migration of GPs, in the most deprived areas, the number of GPs decreased by 8.43% (95% CI 5.86 to 10.99).
CONCLUSIONS: The workforce crisis in Hungarian primary care is progressively deepening and resulting in more severe inequity in access to healthcare. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

The study is based on a comprehensive longitudinal follow-up survey at a national level to evaluate major changes in the number and distribution of general practitioners (GPs).

The survey used an area-based composite indicator to study the relationship between deprivation and shortage in GPs in the country.

The study analysed the migration of GPs between practices, but the reasons for leaving primary care were not investigated in detail.

Data were available for the first day of each year; thus, the study could not detect transient changes in a year.

Introduction

The workforce crisis in primary care (PC) is a worldwide phenomenon. The shortage in general practitioners (GPs) is particularly severe in low-income and middle-income countries in Africa, Asia and the Pacific,1–3 but it is becoming more striking in high-income countries as in the USA,4 5 UK6–8 and Switzerland9 10 as well. In 2016, England was already approximately 6500 GPs below the ideal number, and this gap would increase to 12 100 by 2020.6 Switzerland is also facing an impending PC workforce crisis since almost half of all PC physicians are expected to retire in the next decade.9 10 As the ‘Health at a glance, 2016’ report11 shows in the Organisation for Economic Co-operation and Development (OECD) countries―among them in the European Union member states―the workforce crisis at the level of PC is so deep that in the period of 2011–2013 a significant portion of the emergency department (ED) visits occurred because PC was not available. In certain Central Eastern European (CEE) countries (Czech Republic, Slovakia), this proportion of ED visits was as high as 52%–74%.

The shortage of GPs seems to be the key issue of the workforce crisis in PC throughout the world which may strongly contribute to the limited access of people living in socioeconomically deprived conditions to PC and other health services.12 The Position Paper of the Irish Medical Organisation on health inequalities denotes that vulnerable rural and deprived urban communities have no adequate GP cover in Ireland.13 Demographic studies identified a shortage of PC service providers located in the countryside and a relative oversupply in metropolitan areas in Germany, and in German cities with a population size of more than 500 000 the distribution patterns of GPs were significantly correlated with the geodemographic index, an area measure of deprivation.14 Although the Royal College of General Practitioners reported that the so-called ‘underdoctored’ areas are the most-deprived regions of the UK, that is, ‘the general practice workforce is unevenly spread across the country, with the fewest doctors in the most-deprived areas, exacerbating health inequalities’,15 it is worth mentioning that there was a substantial reduction in socioeconomic inequality in family physician supply associated with national policy for establishing new practices in the most deprived areas under the ‘Equitable Access to Primary Medical Care (EAPMC)’ programme.16

Substantial geographical inequalities in family physician supply was demonstrated between large subnational areas even in high-income countries with universal health coverage. However, because these studies have focused on large areas, they have not been able to accurately describe socioeconomic inequality in PC supply by pinpointing PC shortages in specific disadvantaged neighbourhoods.17–26

WHO Department of Human Resources for Health provided an overview on the inequitable access of populations to PC. The underserved populations were identified in remote and rural areas where the socioeconomic environment, such as working and living conditions, as well as access to education for children were unfavourable.27 28 Data are available almost exclusively from cross-sectional rather than longitudinal surveys6 29–31; therefore, these studies are unable to report on previous and actual translations in the PC system. In addition, these surveys were conducted among GPs typically in a single region of a country,30 32–35 which may not be entirely typical of a country as a whole.

Concerning the population size of Hungary, as well as the fact that the workforce crisis in PC was evolving in the last decade the country may serve as an ideal field to a study targeting the characterisation of the workforce crisis in its development. In Hungary, responsibilities for PC have been centralised at national level and the only source of financing is the National Health Insurance Fund. The local governments are responsible for ensuring that PC services are available to their population, so practices are organised by them at municipality level. The size of the practices, reflecting the number of clients served, varies widely (800–3000 persons/practice), and the average size is approximately 1550 persons/practice. Generally, more family practices operate in more highly populated municipalities, whereas one family practitioner serves more than one municipality in less populated areas. PC physicians are mostly (95%) self-employed service providers contracted to the National Health Insurance Fund, and they work almost exclusively in solo practices, that is, one GP—with the assistance of one practice nurse—serves the clients. GPs provide a complexity of care (screening, health status assessment, diagnosis, treatment, follow-up and referral to the secondary level of care) through capitation-based insurance coverage for this basic healthcare service. Concerning the type of practices, about half of them serves only adults, one quarter of them serves exclusively children, while the remaining are mixed practices. If a GP post becomes vacant, it is very difficult to recruit replacement GP, in general GPs from the nearby practices as locum tenents provide services in very limited working hours, so the population belonging to the practice with vacant GP post will be definitively underserved. Theoretically, for GPs, there are some competitive work opportunities, because about 40% of them in addition to specialisation in family care have medical specialisation in paediatrics or internal medicine, but the healthcare institutions at secondary or tertiary levels rarely provide GPs with positions, so they can work almost exclusively in the emergency service at PC level. On the basis of the results obtained in the Hungarian Public Health Focused Primary Care Model Programme,36–39 the government started to disseminate the GPs’ cluster work in a new pilot programme in 2018. GPs’ clusters are built on the cooperation of six GPs employing health professionals (public health professionals, community nurses, physiotherapists, dieticians, health psychologists) who—in addition to the traditional patient care—are competent to plan and implement various public health services at cluster level. In 2019, approximately 6% of the PC practices is involved in this pilot, that is, solo practice is still the norm in Hungary. On the basis of the three dimensions of PC structure—governance, economic conditions and workforce development—the Hungarian PC system is qualified into the medium category,40 but the coordination and comprehensiveness are found to be weak.18

Our study is designed to follow the development of the GP shortage and to describe the characteristics of changes in Hungary, a country which illustrates a prime example of CEE countries with a workforce crisis in PC. The secular trend by changes in the number, age and sex of GPs by type of practices, as well as the distribution of GPs and vacant GP positions by deprivation is characterised in a longitudinal follow-up design in the 10-year period that began on 1 January 2007 and ended on 1 January 2017.

Methods

Data on general practices analysed

The analysis is based on data provided by the National Institute of Health Insurance Fund Management (NIHIFM) for the period of 1 January 2007 to January 2017 in Hungary. The NIHIFM provided detailed information for the first day of each year on the number, type (adult, paediatric or mixed) and geographical (municipal) location of general practices, identified by numeric practice codes. It was reported whether the GP post of the practice was filled or unfilled on the first day of each year. In the case of practices with filled GP posts, the age and sex of the GPs were registered. If the GP post was unfilled, the period during which there was an inability to recruit GP, as well as the number of population underserved (ie, the number of people belonging to practice with vacant GP post) was also indicated. If the reason for vacancy involved moving a GP from one PC practice to another, the deprivation status of the municipalities where the former and the newly engaged practices were found was also considered in the analysis; the number of practices affected was defined by deprivation index (DI) quintiles.

In the analysis performed, the change in the number of practices with filled and unfilled GP posts and the change in the average age of GPs were defined by the type of practices for the period investigated. In addition, the change in distribution by sex was also determined.

Deprivation

To see the relationship between the distribution of unfilled practices and the socioeconomical status (SES) of the settlements where the practices are located, association analyses were carried out. The SES was characterised by the DI, an area-based composite indicator at the municipality level. The DI was built from seven different indicators (income, the level of education, the rate of unemployment, the rate of one-parent families, the rate of large families, the density of housing and car ownership) derived from the Regional Informational System of the Ministry of Local Government and Regional Development. The data were originally obtained from the Hungarian Central Statistical Office (from the Census 2011) and the Hungarian Tax and Financial Control Administration (for the year of the census). The development of the DI was carried out by using principal component analysis. All variables were first transformed by using the natural log-transformation and then standardised. Areas with positive (higher) index values were municipalities with a lower socioeconomic status than the national average, and the converse was shown in areas with negative (lower) index values. This method to calculate DI values has been published previously by us41 and successfully used in several studies.42–45 The same or similar approaches measuring SES were also applied in studies carried out in other countries all over Europe46 such as the UK,47–50 France,51 Spain,52 53 Germany,54 Denmark,55 Switzerland,56 Italy,57 Czech Republic58 and Slovenia.59

Mapping and risk analysis of vacancy

A hierarchical Bayesian binomial model was used to predict the relative vacancy ratio at the municipality level. The data consisted of the observed number of practices with unfilled GP posts and total number of practices for each municipality. The convolution model, originally proposed by Besag et al,60 was used and incorporates both a heterogeneous and spatially structured random effect. The parameters for the spatial model were estimated via Integrated Nested Laplace Approximations (INLA) directly in R using the package INLA.61 62 The map for vacancy shows the relative vacancy rate and exceedance posterior probabilities.

Using the risk analysis capabilities of the Rapid Inquiry Facility,63 the association between deprivation and vacancy for GPs was calculated. Relative frequencies of vacancy in relation to the national average were calculated by DI quintiles as a municipality-based categorical covariate. χ2 tests for homogeneity and for linear trend were also carried out to test the global association of the DI and the vacancy.

Patient and public involvement

Patients and or public were not involved.

Results

The number of local administrative units 2, called municipalities or settlements in Hungary, was 3176 with a total population number decreasing from 10 066 158 (2007) to 9 797 561 (2017) in the period examined.

At the beginning of the decade covered by the study, there were 6427 PC practices serving the country’s population, and at the end of the period, 6350 practices were operating. During the survey period, the number of practices decreased by 77. Behind the change in the number of practices were decisions made by local governments having the right to merge practices that typically could not be fulfilled for years in small-sized settlements with other practices belonging to the same administrative municipality.

Change in number of practices with unfilled GP posts

In Hungary, the number of GPs in the beginning of the decade examined was 6247. Of these, 53.2% served adults, 24.3% served children exclusively and the remaining 22.5% served mixed practices. For the end of the period, 6003 GPs were registered with the following distribution: 54.6% in adult practices, 23.8% in paediatric practices and 21.6% in mixed practices (table 1).

Table 1

Number and average age of general practitioners (GPs) and the number of practices with a vacant GP post by the type of practices, as well as, the sex distribution of GPs in Hungary, 2007–2016

	1 January 2007	1 January 2008	1 January 2009	1 January 2010	1 January 2011	1 January 2012	1 January 2013	1 January 2014	1 January 2015	1 January 2016	1 January 2017	2007	2017
Number of GPs by type of practice												Change (p for trend)
Adult practices	3315	3316	3339	3331	3319	3317	3307	3303	3301	3301	3279	−1.1% (p=0.004)
Paediatric practices	1524	1526	1520	1522	1504	1494	1488	1475	1459	1440	1425	−6.5% (p<0.001)
Mixed practices	1408	1409	1400	1399	1393	1378	1370	1357	1341	1314	1299	−7.7% (p<0.001)
Total number of GPs	6247	6251	6259	6252	6216	6189	6165	6135	6101	6055	6003	−3.9% (p<0.001)
Total number of general practices	6427	6415	6417	6408	6395	6382	6383	6372	6370	6356	6350	−1.2% (p<0.001)
Practices with a vacant GP post and their distribution by type of practice (%)												Proportion of practices with a vacant GP post (95% CI)
Adult practices	45 (25.0%)	44	38	38	46	41	53	58	63	64	86 (24.8%)	1.36% (0.96 to 1.75)	2.62% (2.08 to 3.17)
Paediatric practices	22 (12.2%)	16	20	19	30	33	38	46	60	65	71 (20.5%)	1.44% (0.84 to 2.04)	4.98% (3.85 to 6.11)
Mixed practices	113 (62.8%)	104	100	99	103	119	127	133	146	172	190 (54.7%)	8.03% (6.61 to 9.44)	14.63% (12.7 to 16.55)
Total number of practices with a vacant GP post	180	164	158	156	179	193	218	237	269	301	347	2.8% (2.4 to 3.2)	5.46% (4.91 to 6.02)
Number of practices with a vacant GP post for more than 1 year	81	93	83	86	93	108	131	142	160	185	240	1.26% (0.99 to 1.53)	3.78% (3.31 to 4.25)
Average age of GPs by type of practice												Change (p for trend)
Adult practices	53.3	53.6	53.9	54.1	54.6	55.0	55.2	55.7	56.3	56.6	57.0	3.7 years (p<0.001)
Paediatric practices	54.1	54.4	55.0	55.6	56.1	56.7	57.1	57.9	58.5	58.9	59.5	5.4 years (p<0.001)
Mixed practices	52.2	52.6	52.8	53.1	53.6	53.7	54.2	54.7	55.3	55.7	56.4	4.2 years (p<0.001)
Sex distribution among GPs (number)												Change (p for trend)
Number of female GPs	3265	3291	3317	3344	3349	3359	3377	3383	3390	3383	3373	3.3% (p<0.001)
Number of male GPs	2982	2960	2942	2908	2867	2830	2788	2752	2711	2672	2630	−11.8% (p<0.001)
Sex distribution among GPs (%)												Proportion of sex distribution (95% CI)
Representation of female GPs	52.3	52.6	53.0	53.5	53.9	54.3	54.8	55.1	55.6	55.9	56.2	52.27% (51.03 to 53.50)	56.19%(54.93 to 57.44)
Representation of male GPs	47.7	47.4	47.0	46.5	46.1	45.7	45.2	44.9	44.4	44.1	43.8	47.73% (46.5 to 48.97)	43.81% (42.56 to 45.07)

Bold values are of statistical significance.

The number of GPs in all types of practices decreased during the decade, especially the number of GPs in mixed (by 7.7%) and paediatric (by 6.5%) practices. Despite the administrative measures resulting in a reduction of the number of practices, the number of practices with a vacant GP post increased dramatically, especially in the second half of the period examined. At the beginning of the decade, the number of practices with unfilled GP posts was 180, while at the end of the period it was 347; these shifts occurred across 519 settlements. By the end of the decade, the number of adult practices with unfilled GP posts doubled, while the number of paediatric practices with a vacancy for a paediatrician more than tripled. Although the relative change in the case of vacant mixed practices was less (68.1% increase), at the end of the decade among the practices with a vacant GP post this type of practice was dominated (adult: 24.8%, paediatric: 20.5%, mixed: 54.7%), as at the beginning of the period (adult: 25.0%, paediatric: 12.2%, mixed: 62.8%). The most serious findings were that 69.2% of the practices with a vacancy for GP in 2017 had had the vacancy for more than a year, and that number was three times higher than it was in 2007 (240 vs 81).

Change in age and sex distribution of GPs

The average age of GPs was high even at the beginning of the decade, especially the age of the paediatricians (54.1 years), but the average age of GPs serving adult and mixed practices was also over 50 years (53.3 and 52.2 years, respectively). During the decade examined, the average age of GPs was increased by 3.7 years in adults, by 5.4 years in paediatric and by 4.2 years in mixed practices (table 1). Concerning the balance by sex, in 2007, 52.3% of the GPs were women, and this rate increased to 56.2% by the end of the decade. The age structure graph showing the distribution of female and male GPs by age in 2007 and 10 years later in 2017 (figure 1) clearly indicates both ageing and feminisation of the GPs population.

Figure 1

The age and sex distribution of general practitioners (GPs) in 2007 and 2017 in Hungary.

Association between deprivation and practices with an unfilled GP post

Mapping the DI values shows that the most-deprived areas were found in the north-eastern and south-western parts of Hungary. The least deprived sections were areas in the north-western part of the country and in the capital city of Budapest and its neighbouring areas (figure 2A). The areas with the highest relative vacancy rates were found along the north-eastern border of Hungary and in the mid-eastern and south-western parts of the country (figure 2B). Maps for DI and practices with vacant GP posts show remarkable overlap; the resultant pattern of excess vacancy rates showed a correlation with a spatial pattern of deprivation (figure 2C), as is verified by the results of association analysis for DI quintiles and practices with vacant GP posts. The results of the risk analysis showed an exponential association between relative vacancy rate and deprivation (χ2 homogeneity=232.18, p<0.001, χ2 linearity=168.87, p<0.001).

Figure 2

The distribution pattern of (A) deprivation and of (B) the relative vacancy rates in Hungary and (C) the relationship between deprivation and the relative vacancy rate.

Based on DI values, in the areas of highest deprivation (quintile V), nearly a fifth (18.31% (15.83–21.16)) of the practices had a vacancy for a GP; this rate is nearly triple that of the Hungarian average vacancy rate (6.39% (5.88–6.94)). Even in quintile IV of the municipalities, 8.21% of the practices had vacant GP posts, that is, the vacancy rate exceeded the national average by 29%. In the least deprived areas, the vacancy rate was 4.07%, which was only 64% of the national average. The vacancy rates between the areas of deprivation quintiles, I, II and III, did not differ significantly (table 2).

Table 2

Distribution of settlements with vacant general practitioner (GP) posts and the vacancy rates compared with the national average by deprivation index quintiles on 1 January 2017

Deprivation quintiles	Total number of settlements with vacant GP post, 2017	Vacancy rate (%) (95% CI)	Relative vacancy rate (95% CI)
D I (least deprived area)	51	4.07 (3.05 to 5.32)	0.64 (0.47 to 0.84)
D II	84	3.95 (3.17 to 4.88)	0.62 (0.49 to 0.77)
D III	107	4.52 (3.76 to 5.44)	0.71 (0.59 to 0.85)
D IV	128	8.21 (6.96 to 9.69)	1.29 (1.08 to 1.53)
D V (most deprived area)	149	18.31 (15.83 to 21.16)	2.86 (2.44 to 3.36)
Hungary	519	6.39 (5.88 to 6.94)	1

Bold values are of statistical significance.

Moving GPs to another PC practice

In the period examined, a definitive movement of GPs from the more deprived municipalities to the least deprived ones could be detected (table 3). In the least deprived (DI quintile I) areas, the number of GPs increased by 2.2%, while in the most deprived areas (DI V) it decreased by 8.4%. Municipalities in the DI IV and V areas were losing GPs while those in DI I, II and III were gaining them. The resultant outcome of this trend is that the distribution of GPs, as well as the access to PC services by DI quintiles, became more inequitable (table 3).

Table 3

Migration of general practitioners (GP) among areas with different deprivation statuses (DI–DV) between 1 January 2007 and 1 January 2017

Deprivation quintiles	Total number of GPs, 2017	Number of outgoing GPs	Number of incoming GPs	Gain/loss in number of GPs	Change between 2007 and 2017 (95% CI)
D I	809	53	71	18	2.22% (1.21 to 3.24)
D II	1802	98	118	20	1.11% (0.63 to 1.59)
D III	2029	120	142	22	1.08% (0.63 to 1.53)
D IV	1259	123	101	−22	−1.75% (−1.02 to −2.47)
D V	451	93	55	−38	−8.43% (−5.86 to −10.99)
Total	6350	487	487

Bold values are of statistical significance.

Change in rate of population underserved

Parallel with the increasing vacancy rate, the rate of people with no PC service provision by their own GP was also increasing during the period examined in strong association with the deprivation. The rate of people (number of persons/100 000 people) with no PC service provision was increasing in time, especially in the most deprived quintile (table 4).

Table 4

Rate of underserved population by year and deprivation quintile

Year	Underserved population per 100 000 people (95% CI)
	D I	D II	D III	D IV	D V
2007	581.09 (571.21 to 591.11)	1766.01 (1748.4 to 1783.76)	1724.06 (1707.86 to 1740.38)	3480.63 (3453.58 to 3507.83)	6075.86 (6035.00 to 6116.94)
2008	687.24 (676.53 to 698.07)	1465.92 (1449.89 to 1482.09)	1469.49 (1454.53 to 1484.56)	2690.60 (2666.76 to 2714.60)	6296.72 (6254.98 to 6338.68)
2009	1066.93 (1053.62 to 1080.36)	1377.62 (1362.07 to 1393.31)	1258.64 (1244.78 to 1272.61)	2979.71 (2954.54 to 3005.03)	4893.96 (4857.04 to 4931.09)
2010	1454.05 (1438.55 to 1469.68)	1205.07 (1190.51 to 1219.77)	1338.9 (1324.58 to 1353.33)	2628.33 (2604.62 to 2652.2)	5023.85 (4986.32 to 5061.59)
2011	1029.26 (1016.25 to 1042.40)	2160.00 (2140.46 to 2179.66)	1355.47 (1341.05 to 1370.02)	2931.45 (2906.32 to 2956.75)	5379.83 (5340.88 to 5419.00)
2012	1348.85 (1333.96 to 1363.86)	1779.92 (1762.16 to 1797.80)	1483.27 (1468.15 to 1498.50)	2929.17 (2903.97 to 2954.54)	6028.88 (5987.54 to 6070.44)
2013	1466.16 (1450.69 to 1481.75)	2059.88 (2040.76 to 2079.14)	1723.64 (1707.32 to 1740.09)	4174.78 (4144.47 to 4205.26)	6535.45 (6491.99 to 6579.14)
2014	1520.52 (1504.78 to 1536.37)	2173.38 (2153.71 to 2193.19)	1843.66 (1826.76 to 1860.68)	5012.76 (4979.45 to 5046.23)	6174.78 (6132.51 to 6217.26)
2015	1397.14 (1382.05 to 1412.36)	2160.64 (2141.00 to 2180.41)	2339.49 (2320.44 to 2358.66)	5444.36 (5409.71 to 5479.19)	7303.19 (7257.56 to 7349.03)
2016	2145.78 (2127.10 to 2164.59)	2251.45 (2231.38 to 2271.66)	3072.88 (3051.00 to 3094.88)	6116.59 (6079.75 to 6153.61)	7186.53 (7141.19 to 7232.08)
2017	1948.73 (1930.92 to 1966.66)	3331.61 (3307.18 to 3356.18)	3278.79 (3256.18 to 3301.51)	7057.39 (7017.81 to 7097.14)	8964.66 (8914.02 to 9015.51)

Bold values are of statistical significance.

Discussion

PC is considered to be the foundation of the healthcare system throughout the world. Currently, in countries with rapidly ageing populations and a high number of people living with chronic diseases, a strong and accessible base of PC providers is essential to the proper level of medical services as well to health promotion for the whole population. There is a general agreement in the scientific literature regarding a workforce crisis at the level of PC; the shortage of PC physicians now calls for attention and urgency worldwide.64

In our present report, the following trends and phenomena are clearly identified as characteristics of the Hungarian PC workforce crisis in the decade examined:

Increasing number of PC practices with a vacant GP post.

Decreasing number of GPs, especially in mixed and paediatric practices.

Increasing average age of GPs in practice.

Feminisation of the profession indicated by increasing representation of female GPs.

Strong association between deprived areas and practices with unfilled posts for GP.

Intense movement of GPs from the deprived areas to the least deprived ones.

It is evident that the GP shortage has a direct effect on the quality of PC service provision, but it is reasonable to suppose that it causes problems at the level of secondary care and in access to public health services. Although the usefulness and effectiveness of the gatekeeping function of GPs in reducing utilisation of emergency and hospital services are a subject of intense discussion with contradictory conclusions (see reviewed in 65 66) in the CEE countries with a very unfavourable health status of their population,67 68 it seems to be essential. In these countries (except the Czech Republic and Slovakia), patients do not have direct access to hospital consultants, but the GPs control access to secondary care.69 Although, currently, more outpatient specialist services can be approached in Hungary without referral, the traditional referral system is still the dominant one in practice. Consequently, the increasing number of practices with unfilled GP posts may strongly affect the health status of the populations not only by missing PC and preventive services, but through a backlog of referrals to relevant specialists, as well. It is worth mentioning that the population of the most deprived areas is the most affected by the GP shortage and the replacement of GPs in practices especially in small rural settlements is only partial, if any.

The chronic shortage of doctors in the primary healthcare sector owing to low pay, restricted opportunities for training or career progression and poor working conditions has been a well-known fact for a long time,70 71 but in the last decade, an exponential increase in vacant GP posts was observed in different countries. The Pulse magazine survey of 860 GPs in May 2017 found that 12.2% of positions were currently vacant in England, an increase from 2.1% in 2011 when Pulse started collecting these data.72 Similarly, in a survey in Scotland, 22% of responding PC practices reported current GP vacancies as of 31 August 2015, an increase from 9% of practices reporting current vacancies in 2013.73 The same phenomenon can be detected in Hungary, where the number of PC practices with an unfilled GP post almost tripled in the period of 2010–2017. The decrease is significant in the number of GPs in all types of practices, but it is more pronounced in mixed and paediatric practices. It seems that the mixed general practices are less attractive among GPs. One of the possible reasons of this phenomenon is that most of these practices are found in small settlements where the GP is the only PC service provider. The reason of the decrease in the number of paediatricians is also clear; the paediatrician medical specialisation training focuses mostly on the hospital replacement in Hungary. Hospitals in many parts of the country are currently reporting shortages in paediatricians and young medical graduates take the hospital work more attractive and the carrier paths more clear at present.

Parallel with the decrease in the number of GPs, the average age of the GPs became significantly higher. The age distribution clearly indicates that the number of male GPs under the age of 60 years was lower in 2017 than in 2007 across all ages, while in the case of female GPs, although the trend is very similar, increases can be seen at certain ages, mainly at younger ones. This fact is in harmony with the observed feminisation of the profession. In a systematic review based on 32 relevant publications to examine the evidence that quantifies the effect of feminisation on practice characteristics, Hedden and her colleagues74 note ‘that female GPs self-report fewer hours of work than male GPs, have fewer patient encounters, and deliver fewer services, but spend longer with their patients during a contact and address more separate presenting problems in one visit’, and conclude that ‘the feminization of the workforce is likely to have a small negative impact on the availability of primary health care services, and the drivers of observed differences between male and female GPs are complex and nuanced’. Considering the fact that Hungary is among the 10 countries that have the best parental leave policies in the world―the length of paid maternity leave is 24 weeks and is paid at 70% of salary, which can be extended by an additional time off paid at 40% of salary for 136 weeks75―female doctors may stay away from work for a fairly long time, which can make finding a suitable temporary replacement problematic.

Behind the fact that in Hungarian PC there is a strong association between deprived areas and practices with unfilled posts for GPs and the intense movement of GPs from the deprived areas to the least deprived ones is the probability of an existing financial background. Although the majority of GPs are private entrepreneurs, capitation-based financing is almost exclusively based on the list size of the practices, and only approximately 3% of GPs’ reimbursement is pay for performance. Consequently, GPs are financially motivated to maximise the number of persons in the practice and minimise performance.39 According to the estimated data from the Hungarian Medial Chamber, in 2017, after covering the expenses related to maintaining the infrastructure and salaries of the other health staff (at least a nurse) the remaining income that can be used to pay the salary of the GP was approximately €1000 before taxation. Although it can be—and should be—refused on the basis of ethical considerations, it remains a fact that the informal so-called ‘gratuity’ payment from the patients for GP services is more or less essential to make a living. It is reasonable to suppose that moving GPs from deprived areas to least deprived ones is―at least in part―motivated by the expected increase in ‘gratuity’ payment. These observations clearly indicate that targeted intervention from the Ministry of Human Capacities is needed to provide the NIHIFM with decision-making responsibility and financing opportunities to support effective recruitment and retention of GPs for the most deprived areas, and for the Hungarian PC system as a whole, as well.

The recruitment and retention problems are well known in the PC systems of developed countries, but information about effective strategies and interventions is very limited. Recently, a systematic review was undertaken in which 51 studies assessing 42 interventions categorised into 13 groups (financial incentives, recruiting rural students, international recruitment, rural-focused or primary-care-focused undergraduate placements, rural or underserved postgraduate training, well-being or peer support initiatives, marketing, mixed interventions, support for professional development or research, retainer schemes, re-entry schemes, specialised recruiters or case managers and delayed partnerships) were retrieved. Weak evidence supported the use of postgraduate placements in underserved areas, undergraduate rural placements and recruiting students to medical school from rural areas, and there was mixed evidence about financial incentives.76

A single report on a longitudinal small area whole country study using lower super output area data and the Index of Multiple Deprivation 2010 rankings for England described substantial inequalities in GPs’ service, expressed in full-time equivalent (FTE) GPs before the EAPMC programme was launched in December 2007. EMPMC invested £250 million towards establishing new general practices and GP-led health centres, as well as towards extending opening hours and expanding services in the 38 most underdoctored primary care trust areas and providing a guaranteed competitive salaries for GPs. Total numbers of FTE for GPs have grown much faster in the most deprived fifth of small areas in England than elsewhere, with the GP supply in the most affluent fifth growing at the slowest pace over the 10 years covered by the study.16 Our findings obtained in our longitudinal study strongly differ from the results presented as outcomes of the EAMC programme. It is reasonable to suppose that the main cause of the difference that competitive salaries for GPs are not guaranteed in Hungary. According to OECD figures, the income of a GP in Denmark is five times larger than the income of a GP in Hungary, and the average GP in Scandinavia earns 5–10 times more than the average GP in Hungary.77 In 2013, a salary increase programme for GPs started in Hungary, but the annual growth rate in nominal terms was as low as 2.3%.78 However, since 2014, a single settlement allowance has been given to GPs at the time of entering into a permanently vacant GP posts; no permanent incentives were introduced to support movement by GPs into underserved deprived areas. Although studies on the effect of financial incentives for GPs in Hungary are lacking, it can be stated that without increasing the salaries of GPs there is no chance to stop the deepening workforce crisis in Hungarian PC.

The limitation of our study is that it is an ecological one that assumes that individual members of a group all have the average characteristics of the group as whole, when, in fact, any association observed between variables at the group level does not necessarily mean that the same association exists for an individual plucked from the group. In addition, we had no opportunity to study the effects of different healthcare reforms on the PC workforce in Hungary.