Does socioeconomic status of list populations affect GP practice? A register-based study of 2201 Norwegian GPs

Abstract

Background: Mortality and morbidity rates differ markedly across social strata, resulting in different needs for health services. The utilization of GP services may be higher in groups with lower socioeconomic status (SES), although findings differ when taking health needs into account.

Objectives: The aim of this study was to assess the association between the SES of list populations and the characteristics of GP practices, consultation rates, and income per patient in a fee-for-service financing model.

Method: A cross-sectional register-based study, including all Norwegian specialist GPs practising in 2008. After grouping GP lists into five levels based on a constructed index of SES, associations between SES and GP practice characteristics were analysed by analysis of variance and linear regression.

Results: GP lists with the lowest SES had higher consultation rates (regression coefficient, 0.31; P < 0.001) and a higher total fee-for-service (regression coefficient, 104; P < 0.001) than lists with the highest SES. Laboratory use in consultations was less frequent in the lowest SES group (regression coefficient, –3.1; P < 0.001). No differences were found in the frequency of long consultations or fee-for-service per consultation. The frequency of multidisciplinary meetings was 2.5 times higher in the lowest SES group compared to the highest SES group.

Conclusion: The findings indicate a markedly higher utilization of GP services in list populations with a lower SES, compensated by a higher annual GP income per patient. However, consultation characteristics, such as time spent with patients and use of laboratory tests, did not increase with lower SES lists.

Key words::

• General practice
• practice profile
• socioeconomic factors
• fee-for-service
• health service research

KEY MESSAGE(S)

• GP consultation rates and fee-for-service per patient per year vary markedly across social strata of the GP list populations

• The use of time and medical procedures in the consultations were not affected by the socioeconomic differences on the GP lists

Introduction

Mortality and morbidity rates differ markedly across social strata, even in developed countries (1,2). One aim for health services should be to reduce this gradient by effective health care with a preference to disadvantaged groups (3). Secondary health care likely works in the opposite direction, with greater utilization in population groups with higher socioeconomic status (SES) (4). Primary health care (PHC) appears to be more equitable and people from lower SES groups use GP services more frequently (4). However, the use of GP services might still be low given greater needs (5,6). Indeed, high quality in PHC may reduce inequalities in health (7,8).

Aggregating SES characteristics of the individuals on a GP list may serve as an indicator for the actual need for health care in a list population, and could also give information about the social context of the patient and the GP. This context may have an independent impact on health beyond what is found based on individual SES markers (9).

The GP workload and practice characteristics should reflect the medical needs of their patients (10,11). However, studies have shown great variations among GPs with respect to prescription rates, use of medical procedures, and time spent with patients that are only partly explained by differences in the practice populations (12,13).

Various models for financing GP services are used to improve the services offered to disadvantaged groups. One model is capitation adjusted to the SES composition of the GP list population to compensate for a presumed increased workload in underprivileged areas (14,15). However, this model does not ensure that the extra resources are actually used according to needs.

In fee-for-service models, GPs are paid for actual patient contacts and medical procedures that are carried out. This financing model has been criticized for stimulating GPs to maximize their income by prioritizing their own economic interests irrespective of medical criteria. GPs might also increase their activities per patient if they experience a shortage of patients (16). However, in a fee-for-service model GPs may be more responsive to the need for services in the list population through increased activity (output) because they are rewarded for treating medical problems.

In the list-based Regular GP Scheme, which was introduced in Norway in 2001, capitation constitutes one-third of GP income and is based solely on the number of patients on a GP's list and two-thirds of the income is generated from fee-for-service. An underlying assumption is that this model gives the GP stronger incentives to work with the patients’ health problems than to aim for a large patient list.

In an equitable GP system, the workload and fee-for-service would be expected to increase with an increasing proportion of patients from lower SES groups on the list given a fixed list size. Examining practice variations across social strata at the list level may indicate to what degree health care provision is allocated fairly.

The aim of this study was to determine how differences in the SES level in list populations predicts the GP rates of consultations, use of time, medical procedures, cooperation with other health and social services, and income from fee-for-service.

Methods

Cross-sectional, register-based study among Norwegian GPs.

Participants and inclusion and exclusion criteria

All GPs who were approved specialists in family medicine with a list size ≥ 500 patients and who had practice activities in all quarters of 2008 were included in the study (n = 2201).

Data

The data were obtained from three encrypted national databases linked by Statistics Norway, as follows:

• The national GP database, including the age, gender, and practice municipality of the GP, the size of the patient list, and the identity of the listed patients.

• A national research database (FD trygd), including data on income, educational attainment, and use of social insurance benefits of all residents. This was linked to the individual patients in the GP database.

• A database, including all invoices from GPs (claims for fee-for-service), with data on the practice- related activities of each GP.

The fee-for-service system

For each patient contact, GPs send an invoice to The National Health Insurance (HELFO) with the patient's identity, diagnosis code according to ICPC-2, and codes from a detailed tariff with a fixed patient co-payment for consultations and short contacts. Five per cent of GPs is on fixed salaries. The fee-for-service is paid to the employers (municipal authorities).

The tariff includes fees for different procedures and consultations lasting >20 min. There are also fees for short patient contacts and multidisciplinary meetings, letters, and telephone contacts with other health and social services concerning the patients.

In the current study, fee-for-service is used mainly as a marker of activity. The average fee-for-service per patient per year was calculated as an indicator of workload. In addition, all fees related to consultations were summarized separately and the average fee-for-service per consultation was estimated. This was used as an indicator of the GP consultation style.

Co-payments from patients are not included when referring to fee-for-service.

Outcome variables

The focus of the current study was the association between the SES of the list populations and the GP patient-centred activities (output), as measured by (a) the annual rate of consultations per list patient; (b) the average fee-for-services per list patient per year; (c) the proportion of consultations lasting >20 min, (d) the proportion of consultations with laboratory tests; and (e) average fee-for-service per consultation.

Explanatory variables and measurements

After aggregating individual SES variables in each list population, the GP lists were ranked in 10-percentile groups for each of the following five variables: (a) average annual income of list patients ≥ 20 years; (b) proportion of the list population ≥ 20 years with only basic education; (c) proportion of list patients 20–40 years of age receiving social assistance; (d) proportion of list patients 20–67 years of age receiving disability pension; and (e) proportion of list patients 20–67 years of age receiving unemployment benefits. The rankings (1–10) were summarized for each GP, resulting in a SES score with a possible range of 5–50. Based on this score, the GPs were divided in quintiles and these five SES groups were used as the main explanatory variable. The following variables were included in regression analyses as possible confounders: list size; the proportion of patients ≥67 years of age; the proportion of male patients on the list; the age, gender, and salary arrangements of the GP; and size of the practice municipality.

Statistical analyses

The five SES groups were compared with respect to list, GP, and practice characteristics using the Kruskal–Wallis test for continuous variables, as the variance in the subgroups differed. The Pearson chi-square test was used to compare categorical variables.

Linear regression models were used to analyse the association between the SES level in the list population and the outcome measures. P for trend across strata of SES of GP lists were estimated using the SES groups as a continuous variable in the models. A P-level ≤ 0.05 was considered statistically significant.

The statistical package STATA 11 was used for analyses.

Ethical approval

The study was approved by the Norwegian Data Inspectorate and the authorities responsible for each register.

Results

Descriptives

The study population of 2201 GPs is compared to all Norwegian GPs in Table I.

Table I. Descriptive data of the study group of 2201 Norwegian specialist GPs; the GP lists and practice characteristics compared to the data from all GPs in Norway.

	Study population of specialist GPs (n = 2210)	All GPs in Norway (n = 3901)
Characteristics of GPs
Male GPs, %	71.0	67.8
GP age, mean (SD)	51.0 (7.7)	48.2 (9.9)
List size, mean (SD)	1290 (356)	1195 (400)
GP on fixed salary, %	5.1	7.4
Characteristics of list populations:
Proportion of men in list, %	49.4	49.7
Proportion >67 years of age; %	12.5	12.4
List in municipality with < 10 000 inhabitants; %	23.6	28.5

Table II shows that 31% of the working age population within the lowest SES group was receiving a disability pension, social assistance, or unemployment benefits. List sizes were smaller and GPs were slightly older in the lower SES groups.

Table II. Characteristics of list populations after grouping GP lists in quintiles according to level of socioeconomic status (SES) of the list population. n = 2201 GPs lists.

	SES level in the GP list population
	Highest	Medium high	Medium	Medium low	Lowest	P-value
Number of GPs	501	441	435	426	398
Characteristics of GPs:
Male GPs, %	68.2	67.6	71.5	74.2	74.4	0.070^a
GP age, mean	50.6	50.2	51.5	51.2	51.7	0.017^b
List size, mean	1427	1340	1271	1249	1130	< 0.001^b
GP on fixed salary, %	3.4	3.0	4.4	4.2	11.6	< 0.001^a
Characteristics of list populations:
Proportion of men in list, %	49.3	48.7	49.1	50.1	49.9	0.15^b
Proportion > 67 years of age, %	10.4	11.3	12.5	13.8	14.9	< 0.001^b
Yearly income among persons ≥20 years, in NOK 1000, mean^c	317	255	235	223	204	< 0.001^b
Proportion with basic education only, ≥20 years, % of list population^c	9.5	12.4	14.4	16.4	20.0	< 0.001^b
Proportion with social assistance recipients, % of list population 20–40 years^c	2.2	3.1	4.0	5.0	7.1	< 0.001^b
Proportion disability pensioning, % of list population 20–67 years^c	6.7	9.3	11.0	13.2	15.3	< 0.001^b
Proportion with unemployment benefit, proportion, % of list population 20–67 years^c	4.5	5.5	5.7	6.4	8.3	< 0.001^b
Proportion of lists in municipalities with <10 000 inhabitants, %	7.2	19.1	25.1	29.1	41.7	< 0.001^a

^aPearson chi-square test.

^bKruskal–Wallis test.

^cList population characteristics used when constructing the index of SES level that the grouping of the GP list are based on.

Outcomes

Among all GPs the mean annual consultation rate was 2.4 per patient in the list population and the average fee-for-services per list patient per year was NOK 465 per patient (荤56.70, using an exchange rate of NOK 100 = 荤8.20).

Table III shows different practice profiling factors for the GPs grouped by the SES level. Comparing lists in the highest SES group to lists in the lowest SES group showed a gradual increase in rates of consultations and multidisciplinary meetings and the average fee-for-service per patient per year. Use of time and procedures in consultations had minor differences.

Table III. GP fee-for-service, contact rates, and practice profiling characteristics after grouping GP lists according to level of socioeconomic status (SES) in the list populations. n = 2201 GPs lists.

	SES level in the GP list population
	Highest	Medium high	Medium	Medium low	Lowest	P-value^a
Number of GPs	501	441	435	426	398
Related to whole practice:
Consultation rate per year per patient in list	2.19	2.31	2.35	2.47	2.53	< 0.001
Multidisciplinary meetings pr 1000 patient per year	14.5	21.9	25.5	28.7	37.8	< 0.001
Average fee-for-service per patient per year, NOK^b	405	443	455	501	536	< 0.001
Related to consultations:
Proportion of consultation:
• lasting >20 min, %	27.7	27.2	27.0	27.8	28.8	0.46
• with use of laboratory test, %	42.2	42.1	42.6	42.3	41.0	0.14
• with EKG taken, %	1.9	2.0	1.9	2.2	2.2	0.019
• with spirometry taken, %	1.6	1.6	1.5	1.7	1.3	< 0.001
Average fee-for-service per consultation, NOK (1)	134	133	132	136	137	0.30

^aKruskal–Wallis test

^bExchange rate average in 2008: NOK 100 = 荤12.2.

Statistical associations in regression models

In the adjusted linear regression models (Table IV), the increase in average consultation rates was approximately 13% across these strata of list SES from high- to-low. The average fee-for-service per patient per year was approximately NOK 100 ( 12.20) higher for GPs with the most disadvantaged population compared to the most well off.

Table IV. Association between socioeconomic status (SES) in the GP list population and annual consultation rates and average fee-for-service per patient from National Health Insurance. Linear regression, unadjusted or adjusted models.^a

		Consultations per person per year				Average fee-for-service per person per year (NOK)^b
	n	Unadjusted Regression coefficient	95% CI	Adjusted^a Regression coefficient	95% CI	Unadjusted Regression coefficient	95% CI	Adjusted^a Regression coefficient	95% CI
SES level in list population:
Highest	501	Ref^c		Ref^c		Ref^c		Ref^d
Medium-high	441	0.12^f	(0.04–0.21)	0.10^e	(0.02–0.18)	38^f	(15–61)	27^e	(4–49)
Medium	435	0.16^g	(0.08–0.24)	0.13^f	(0.05–0.21)	50^g	(28–73)	33^f	(10–56)
Medium-low	426	0.28^g	(0.19–0.36)	0.22^g	(0.14–0.31)	96^g	(73–119)	72^g	(48–95)
Lowest	398	0.34^g	(0.26–0.43)	0.31^g	(0.22–0.40)	131^g	(108–154)	104^g	(79–128)
P-trend		< 0.001		< 0.001		< 0.001		< 0.001
Adj R2		0.03		0.09		0.06		0.12

^aAdjusted for GP age and gender, fixed salary or not, list size, age and gender distribution in the list population and size of practice municipality

^bExchange rate average in 2008: NOK 100 = 荤12.2.

^cReference value for consultation rate = 2.19

^dReference value for fee–for–service = 405 NOK = 49.4 荤

^eP < 0.05

^fP < 0.01

^gP < 0.001

A low SES level predicted less frequent use of laboratory tests (Table V). The SES of the list population did not affect the proportion of consultations lasting > 20 min or the average fee-for-service per consultation.

Table V. Association between socioeconomic status (SES) in the GP list population and consultation related practice characteristics among Norwegian GPs. n = 2201. Linear regression, adjusted models.^a

		Proportion (%) consultations > 20 min		Proportion (%) consultations with laboratory test		Average fee-for-service consultations (NOK)^b
SES level in list population:	n	Regression Coefficient	95% CI	Regression coefficient	95% CI	Regression coefficient	95% CI
Highest	501	Ref^c		Ref^d		Ref^e
Medium-high	441	–1.2	(–2.9 to 0.6)	–0.8	(–2.0 to 0.4)	–0.2	(–4.7 to 4.3)
Medium	435	–2.1^f	(–3.9 to –0.4)	–0.8	(–2.0 to 0.5)	–1.1	(–5.6 to 3.5)
Medium-low	426	–1.5	(–3.3 to 0.3)	–1.6^f	(–2.9 to –0.03)	3.1	(–1.6 to 7.8)
Lowest	398	–1.6	(–3.5 to 0.3)	–3.1^g	(–4.4 to –1.8)	4.4	(–0.5 to 9.4)
P-trend		0.087		< 0.001		0.035
Adj R2		0.05		0.11		0.01

^aAdjusted for GP age and gender, fixed salary or not, list size, age and gender distribution in the list population and size of practice municipality.

^bExchange rate average in 2008: NOK100 = 荤12.2.

^cReference value—proportion (%) consultations > 20 min: 27.7%

^dReference value—proportion (%) consultations with laboratory test: 42.2%

^eReference value—fee-for-service (NOK) per consultation: NOK134 = 荤16.4.

^fP < 0.05

^gP < 0.001

Discussion

Summary of main findings

GPs in the lower quintiles according to SES level of the list population had a 13% higher consultation rate and a 26% higher average fee-for-service per patient per year compared to the lists with the highest SES level. These findings indicate that Norwegian GPs, at least in part, respond to the expected greater need for health services in list populations with a lower SES. However, the use of time and the medical procedures in consultations were not influenced by the SES level in list populations.

Strengths and limitations of the study

This study was based on register data, including all Norwegian specialist GPs, which eliminated selection bias. Data was not influenced by interest of GPs to participate in research.

The invoices reporting GP activities gave reliable information because the fees constitute a major portion of GP income; reporting was, therefore, likely to be complete. A control system was applied by the authorities to ensure the correct use of tariffs. All invoices were automatically controlled, and unusual use of tariffs led to further control of the GP.

A major weakness of the study was that no information on the health status of the list populations was available; specifically, the number of patients with severe chronic diseases was not available. Therefore, aggregated socio-demographic variables were used as indicators of the need for health care.

For most consultations, there was a patient co- payment, which may have reduced the use of GP services among the most disadvantaged population groups. This might explain why the differences in consultation rates were not even larger.

The use of an arbitrary index for SES level should be questioned. However, a similar model is used by Statistic Norway in comparing living conditions among municipalities (17). Indicators were chosen based on known associations with health problems and were combined to avoid a strong correlation to the proportion of elderly in the lists.

Associations found in cross-sectional studies should not be interpreted as proof of causality. It is reasonable to regard the SES composition of the list as a predictor of GP output, and not vice versa. However, the practice characteristic of a GP may be taken into account by the patients when choosing a GP. List size might also be reduced if the GP consider the workload too high.

Comparison with existing literature

The present findings may be valid for other countries with list-based GP services and activity-based remuneration fee-for-service. Practice differences related to SES level have been reported in other countries most often related to geographical areas and not to specific GP list composition (15,18,19).

The findings of the current study support earlier findings indicating a higher utilization of GP services in populations with a lower SES (4,18,20–22). Having more frequent consultations with persons with a low SES rather than using more time for each consultation is in agreement with how Norwegian GPs responded in a recent survey on priorities (23). However, in another survey, Norwegian GPs reported increasing the length of consultations for patients with a lower SES (24). This was not confirmed in the present real-life study.

The 13% increase in consultation rates across strata of SES was expected to be higher. The increase in consultation rates explains approximately one-third of the increased fee-for-services per patient from highest-to-lowest SES groups. Other patient-related activities contribute to the increased workload and income for these GPs, such as patient contacts by telephone, prescriptions, and contacts with other health and social services. The most striking finding is the increased frequency of multidisciplinary meetings in lists with a lower SES, which was also reported in a prior study (25). This indicates that GPs with a low SES in their lists handle more complex patient categories, demanding more cooperation with other services, as also observed previously (11,26).

However, a greater complexity of health and social problems might demand more time and more comprehensive medical investigations during consultations (10,11). Consequently, one interpretation of the present study is that the content and quality of service provided may not be commensurate with actual needs.

Previous studies have shown that a higher patient SES leads to more active treatment for a given medical condition, also in general practice, even though patients in lower SES groups probably need more services, such as longer consultations, to achieve the same health benefit from medical advice and treatment (6,10,27). In a Norwegian survey, 82% of all physicians believed that treatment for a given condition should be equal regardless of the SES. This study also indicated that consultation styles are sparsely influenced by patient SES level in real life (23).

Implication for practice and future research

To make the provision and results of health services more equitable, there may be a need for GPs to adapt the content and use of time in consultations to the SES level of patients, taking into account the different abilities of patients to benefit from the provided health services (10, 26). This should be studied further by combining data on SES and utilization of GP services at the individual patient level.

The study indicated that the fee-for-service model gives GPs possibilities to increase the frequencies of consultations in populations with a lower SES, and work with smaller lists. Pay-for-performance incentives based on quality measures might contribute to a more equitable provision of GP services, although empirical results differ (19,28,29). There is a need for further research on financial models to achieve more equitable GP services.

There are different views on basic principles of prioritizing in health care: similar treatment for a specific disease versus allocating greater resources to lower SES groups to achieve greater equity in results (23,30). However, the attitudes, awareness, and knowledge of the possible role of GPs in reducing inequities in health might influence the priorities independent of economic incentives (23,31).

Conclusion

In this Norwegian study there was a markedly higher use of GP services in list populations with a lower SES, as indicated by higher consultation rates and higher average fee-for-services per patient per year. However, the consultation characteristics of GPs are less affected by patient SES. This study suggests that a fee-for-service payment model contributes, albeit only partially, to meeting the increased need for health services among lower SES groups.

ACKNOWLEDGEMENTS

Tor Helge Holmås, PhD, contributed to the management of data and advice on statistical methods.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

The study was financed by the Fund for Research in General Practice, Norwegian Medical Association.