Do GPs sick-list patients to a lesser extent than other physician categories? A population-based study

Family Practice Vol. 18, No. 4 Oxford University Press 2001 Printed in Great Britain Do GPs sick-list patients to a lesser extent than other physician categories? A population-based study Britt Arrelöv, Lars Borgquist, Dan Ljungberg and Kurt Svärdsudd Arrelöv B, Borgquist L, Ljungberg D and Svärdsudd K. Do GPs sick-list patients to a lesser extent than other physician categories? A population-based study. Family Practice 2001; 18: 393 398. Background. Large differences between physicians in their use of the sickness certification instrument have been described. There are also indications of differences between different categories of physicians, for instance that occupational health physicians and GPs are less generous with certificates of long duration. Objectives. We therefore decided to test the hypothesis that GPs and occupational health physicians issue more short-term certificates and use partial sick-listing more often than other physicians. Methods. Certificates for sickness absence during 4 months in 1995 and 2 months in 1996 were collected in eight Swedish counties; a total of 57 563 certificates. From the certificates, a number of variables were extracted. Results. Sickness certificates issued by GPs were on average for a shorter period of time than those issued by other physicians, for the individual certificate as well as for the total sickness period. Occupational health physicians had longer certification periods than GPs but used partial sick-listing more frequently. However, the patients of the various categories of physicians differed regarding age, sex, diagnosis distribution, etc. When the influence of these factors on the duration of the certification periods was taken into account, the GPs still issued significantly shorter periods of sick-leave than the other physicians, followed by the occupational health physicians. Conclusions. The results may be indicative of a different way of handling the sickness certification instrument among different categories of physicians, especially GPs. Keywords. Epidemiology, general practice, sick-leave, sick-listing, sickness certification practice. Introduction The costs for sickness absence have gradually increased over the years in the western world. Who is allowed to certify sickness absence varies between countries. In Scandinavia, all licensed doctors have the right of certification within the national health insurance system. In order to control costs in Sweden, numerous changes in the health insurance system have been made. In spite of these changes, cost control has not been achieved and physicians have been accused of not taking their responsibility as gate-keepers of the system seriously. Received 12 October 2000; Revised 2 February 2001; Accepted 12 March 2001. Uppsala University, Department of Public Health and Caring Sciences, Clinical Epidemiology and Family Medicine Sections, Uppsala Science Park, 751 85 Uppsala and Faculty of Health Sciences, Department of Medicine and Care, Primary Care, Linköping, Sweden. Large differences between physicians in their use of the sickness certification instrument have been found. 1 4 There are also indications of differences between different categories of physicians. 2 Costs due to sickness benefits generate high expenditure in the community. In general practice, certification for sickness absence is the single most expensive item, even more expensive than drug prescriptions. 1,3,5 Physicians attitudes and behaviour regarding sick-listing have a great influence on the costs, 2,3 which might vary considerably between different physicians. As certifiers, physicians have a central role as gatekeepers of the system in order to provide sickness benefit only to those who have a medically acceptable cause of sickness absence. However, the doctor is also supposed to be the patient s advocate, guaranteeing that the patient obtains his or her rightful part of the common resources. A third role is to determine priorities for allocation of limited resources among needy patients, or future patients. 6 When sick-listing a patient, the physician is influenced by a number of factors other than purely 393

394 Family Practice an international journal medical aspects. 1,4,6 8 According to Tellnes the question of sickness certification concerns the patient s subjective feelings of illness, the doctor s assessment of the diseases and the sick role patients assume when they feel unfit for work. 9 This is a conflict situation which is hard to handle. There is a disparity between the ability of various categories of physicians to deal with the complexity of sick-listing. GPs are probably the category that is most well trained to handle this type of conflict, due to their experience of gate-keeping in other situations. Occupational health care physicians are another category that might handle the situation better than average because of their knowledge of and contact with work-places, which might give them a better basis for their decision when evaluating the patient s ability to work. To test the hypothesis that GPs and occupational health care physicians handle the sickness certification instrument differently from other categories of physicians, we collected a database of unselected sick-leave certificates large enough to allow measurement of the performance of various categories of physicians taking possible and measurable confounding factors into consideration. adjustment for degree of sickness absence (number of crude days degree), expressed as net days. The physicians were classified as GPs, specialists working in hospitals, occupational health care physicians and other physicians (mostly doctors in private practice). The majority of the latter were private practitioners. Diagnoses were coded according to a simplified version of the International Classification of Diseases (ICD-9). Statistical analysis The analyses were performed with the Statistical Analysis System and the JMP program packages. Fewer than 1% of the certificate data were missing. Standard methods were used for computing means, standard deviations and confidence intervals. Analyses based on continuous data were performed with Student s t-test or analysis of variance, and analyses based on ordinal data were performed with the chi-square test. Multivariate analyses were performed with linear or ordinal logistic regression in their multivariate forms. All tests were two-tailed. P-values 0.05 were regarded as statistically significant. Very small P-values were denoted 0.0001 even when they were much smaller. Study population and methods The Swedish National Insurance Act covers all residents between 16 and 65 years of age. To obtain sickness benefits, a sickness certificate (standardized form) must be issued by a physician, except for the first week, when sickness absence may be self-certified. Eight counties (out of 24) participated in this study. At least one city or municipality with a hospital and at least one with no hospital in each county was included as a data source. All sickness certificates received by the local branch of the National Health Insurance Office in these communities during February, April, June and October of 1995 and April and October of 1996 were registered. The area covered included urban and rural districts. A total of 57 563 doctors certificates for sickness absence were received. At the time of the study, employers were by law financially responsible for the first 2 weeks of sickness benefit. The collected certificates therefore covered sickness absence after the first 2 weeks, except for unemployed persons, for whom all sickness absence was covered. Variables used for this report were identification number of the insurance office, year and month when the certificate was issued, the patient s age and sex, diagnosis, category of certifiying physician, duration of the total sickness period (from leaving work for sick-leave until going back to work), crude duration of the current certified sickness period (the total sickleave period may have contained several certification periods), degree of sickness absence (100, 75, 50 or 25%) and duration of the current certified period after Results Characteristics of the study population Characteristics of the certificates and the certified study population are shown in Table 1. Patient mean age was 45 years and 60% were women. Twenty-eight per cent of the certificates were initial ones and 80% were fulltime sick-listings. The average duration of the current certificate was 36 days, or 31 net days when the degree of sick-listing was taken into account. The total sick-listing episodes were on average 193 days long. A total of 45.4% of the initial certificates were issued by hospital physicians, 40.4% by GPs, 7.6% by occupational health physicians and 6.6% by other physicians. GPs issued 41.1% of the continuation certificates, hospital physicians 36.8%, occupational health care physicians 13.4% and other physicians 8.7%. The distribution of diagnoses is shown in Table 2. The largest diagnostic group was musculoskeletal disorders, accounting for 43%, followed by psychiatric diseases (15%) and injuries (12%). All other groups constituted 6% or less. Effects of patient age and sex The total duration of the episode, the crude duration of the current period and net days of the current period all increased with age (Fig. 1). For the total episode and crude days of the current certificate, there were no systematic sex differences. However, women had significantly more part-time sickness absence than men (P 0.0001), which for net days gave men more net days of the current certificate than women, except for the very youngest age

Sick-listing practice 395 TABLE 1 Characteristics of the study population TABLE 2 Sickness certification diagnoses n Mean or % SD Diagnoses n % Age (years) Mean 57 563 44.7 11.24 Sex Men 22 762 39.5 Women 34 801 60.5 Type of certificate Initial 16 179 28.1 Prolongation 41 377 71.9 Degree 1.0 45 777 79.6 0.75 1212 2.1 0.50 8342 14.5 0.25 2213 3.8 Other 23 0 Number of days certified Total episode of sickness 57 549 193.4 291.5 Duration of current certificate 57 536 36.0 32.1 Net days of current certificate 57 535 31.0 28.5 Issuing physician GP 23 512 40.8 Hospital doctor 22 540 39.2 Occupational medicine physician 6765 11.8 Other doctors 4662 8.1 Unspecified 84 0.1 Musculoskeletal disorders 24 568 42.7 Neck and shoulder 5326 9.3 Thoracic and lumbar spine 6895 12.0 Joint diseases 3349 5.8 Other musculoskeletal disorders 8998 15.6 Psychiatric disease 8468 14.7 Alcohol and drug abuse 684 1.2 Psychoses 402 0.7 Neuroses 7382 12.8 Injuries 6977 12.0 Neck and shoulder 1167 2.0 Thoracic and lumbar spine 289 0.5 Upper extremities 2033 3.5 Lower extremities 2604 4.5 Miscellaneous 884 1.5 Gastrointestinal and urinary tract disorders 3471 6.0 Cardiovascular diseases 3237 5.6 Respiratory diseases 3155 5.5 Diseases during pregnancy and puerperium 1914 3.3 Gynaecological diseases 1573 2.7 Dermatological diseases 865 1.5 Other infectious diseases 463 0.8 Miscellaneous diseases 2872 5.0 Total 57 563 100 group. The proportion of continuation certificates increased with age but there were no clear-cut sex differences regarding type of certificate. As expected, some of the diagnostic groups were age and sex dependent. Cardiovascular diseases and musculoskeletal disorders increased with age, whereas psychiatric disorders decreased (Fig. 2). Injuries decreased with age among men but tended to increase among the oldest women. The fairly high frequency of miscellaneous causes among young women was due to sickness during pregnancy. Effects of the certifying physician The GPs issued significantly shorter periods of sickleave than the other categories of physicians, whether measured as duration of current certificate, duration of total sickness episode or net days after adjustment for the degree of sick-listing. On the other hand, GPs had a significantly larger proportion of female patients than the other categories of physicians, and there were substantial differences between the categories regarding distribution of diagnostic groups, type of certificate and degree, which might possibly explain the differences in duration of the certificates. Multivariate analyses For this reason, a multivariate analysis was performed with net days as the dependent variable, and physician category, patient age and sex, diagnostic group and type of certificate as independent variables. All these factors were correlated independently and significantly to net days (P 0.0001 for each one). The effect of physician category, patient age and type of certificate is shown in Figure 3 when the effects on net days of patient sex and of diagnostic groups were adjusted for. GPs still issued the shortest duration certificates even when the effects of the other factors were taken into account, followed by occupational health physicians, other physicians and hospital physicians. Discussion The geographical area chosen for this study is fairly representative of Sweden. The certificates were collected during 6 months scattered over 2 years, which covers possible seasonal variation. The insurance benefit came from a common pool with a standardized basis for forming judgements. We therefore have no reason to believe that the data should be non-representative or biased in any other way to such an extent that the results would be affected. Most sick-listings were done by GPs and physicians working at hospitals or in out-patient specialist clinics. The frequency of certificates issued by GPs in different

396 Family Practice an international journal TABLE 3 Characteristics of certifier Issuing physician GPs Hospital physicians Occupational medicine Other physicians physicians Mean or % 95% CI Mean or % 95% CI Mean or % 95% CI Mean or % 95% CI No. of days certified Total episode of sickness 173.2 169.6 176.8 189.4 185.8 193.1 238.8 231.4 246.3 249.0 239.5 258.5 Duration of current certificate 30.1 29.7 30.4 39.2 38.8 39.6 40.7 39.8 41.5 43.2 42.2 44.2 Net days of current certificate 25.6 25.2 25.9 35.5 35.1 35.8 31.3 30.6 32.1 36.4 35.5 37.4 Patients (%) Women 64.8 64.2 65.4 55.7 55.1 56.4 60.8 59.6 62.0 61.2 59.8 62.6 Mean age 44.5 44.4 44.7 44.0 43.8 44.1 47.5 47.2 47.7 44.7 44.4 45.0 Diagnostic groups (%) Gastrointestinal 4.2 3.9 4.4 9.5 9.1 9.9 2.6 2.2 2.9 3.4 2.9 3.9 Cardiovascular 3.9 3.7 4.2 8.7 8.3 9.1 3.0 2.6 3.4 3.0 2.6 3.5 Respiratory 8.5 8.2 8.9 3.4 3.2 3.6 3.3 2.8 3.7 3.6 3.0 4.1 Musculoskeletal 53.5 52.9 54.1 22.4 21.9 23.0 68.1 67.0 69.2 49.4 48.0 50.8 Injuries 8.9 8.6 9.3 18.4 17.9 18.9 6.1 5.5 6.6 6.6 5.9 7.3 Psychiatric 11.2 10.8 11.6 17.4 16.9 17.9 11.1 10.4 11.9 24.7 23.5 25.9 Miscellaneous 11.8 9.4 10.1 20.2 19.7 20.7 5.9 5.3 6.5 9.3 8.5 10.1 Degree Full 78.5 77.9 79.0 85.6 85.1 86.0 64.7 63.6 65.8 76.9 75.7 78.1 Type of certificate Initial 27.8 27.2 28.4 32.6 31.9 33.2 18.1 17.2 19.0 22.8 21.6 24.0 FIGURE 1 Total length of the sickness episode (A), the crude length of the current period (B) and net days of the current period (C) according to patient age and sex

Sick-listing practice 397 FIGURE 2 Prevalence of diagnostic groups among women (A) and men (B) according to age group FIGURE 3 Certified net days of current sick-listing according to physician category and patient age. In addition, adjustments were made for the effect of patient sex and diagnostic group countries varies due to differences in the organization of health care, rules for social insurance and traditions. In a study in Norway, 81% of the initial certificates studied were issued by GPs, as compared with 40% in this study. 10 More women than men were sick-listed, whereas sicklisted men had more net days on their certificates. When categories of certifier and certificate were accounted for, the sex differences disappeared. Some studies have shown higher rates of sickness certification for women, 11 13 but most studies of sickness certification in general practice have shown higher rates for men. 9 Differences between studies might possibly be explained by differences in the proportion of employed persons included, due to a lower employment rate among women, and differences in methodology. In a number of other studies, musculoskeletal diseases were the most common cause of sickness absence, as in the present study. 9,13,14

398 Family Practice an international journal The longer mean duration with increasing age that we found is supported by results from other studies. 9,12,13 The differences between different categories of physicians regarding the certificate duration still remained when all the other factors were taken into account. There may be several possible explanations for these differences. In an attempt to explain the variation within the GP group, Mabeck and Kragstrup divided influencing factors into external and internal. 4 External factors were demography (age, urban rural area, occupation, health belief models, morbidity, family structure, culture and tradition) and health care system (payment system, capacity, access to medical care, equipment and referral structure). Internal factors were organization of practice (list size, ancillary staff, practice facilities, practice form and equipment) and physician-related factors (sex, age, professional skills and knowledge, personality and attitudes). These findings are supported by results from other studies. 1,7,9 The fact that GPs issued shorter periods of sick-leave than other categories of physicians might be explained by: (i) confounding by indication; (ii) different views about risks of long certification periods; or (iii) structural differences. Differences regarding patient characteristics may be important confounding factors. In Sweden, patients, by law, have a free choice of physician. This choice might depend on the patient s age and sex and the kind of disease they suffer from, or other factors related to patient preference. In addition, different physicians might have preferences in different areas of medical practice, which influence the kind of patients that consult them. 1 We have tried to control for these factors, but there may still be confounding effects left, for instance disease severity or physicians preferences, which we could not account for. However, even though there might still be confounding effects left, the differences between the GPs and other categories of physicians are so large that it appears unlikely that the differences are due solely to confounding or other bias. The only way to find out would be by a randomized controlled sickness certification trial, but such a study would be more or less impossible to perform, so we have to rely on the best available information. Conclusions GPs sick-listed patients for shorter periods than other categories of physicians, and occupational health care physicians used partial sick-listing more than other physicians. The explanation might be that the various categories of physicians take different views on the pros and cons of sick-listing and use the certification instrument with different degrees of skill. Acknowledgements This study was supported by grants from the Swedish National Board of Health and Welfare, the National Health Insurance Board in collaboration with 28 of its local offices, and Uppsala University. References 1 Rutle O, Forsen L. Allmenpraksis-tema med variasjoner. [Interdoctor variation in general practice.] Report no. 8. Oslo: National Institute of Public Health, Department of Health Services Research, 1984. 2 Englund L, Tibblin G, Svärdsudd K. Variations in sick-listing practice among male and female physicians of different specialities based on case vignettes. Scand J Prim Health Care 2001; 18: 48 52. 3 Petersson S, Eriksson M, Tibblin G. Practice variation in Swedish primary care. Scand J Prim Health Care 1997; 15: 68 75. 4 Mabeck CE, Kragstrup J. Is variation a quality in general practice? Scand J Prim Health Care 1993; 11: 32 35. 5 Fugelli P, Harstad H. Helseökonomi i almenpraxis [Health economy in general practice]. Tidsskr Nor Legeforen 1983; 103: 1176 1183. 6 Stone DA. Physicians as gatekeepers: illness certification as rationing device. Public Policy 1979; 27: 227 2254. 7 Condren L, Cox J, McCormick JS, Sullivan A. Certification of unfitness for work. Ir Med J 1984; 77: 159 160. 8 Grossmark FB, Sharer P. A study of certification in general practice. Practitioner 1967; 199: 354 355. 9 Tellnes G. Sickness certification in general practice: A review. Fam Pract 1989; 6: 58 65. 10 Tellnes G, Svendsen KOB, Bruusgaard D, Bjerkedahl T. Incidence of sickness certification. Scand J Prim Health Care 1989; 7: 111 117. 11 Alexandersson K, Leijon M, Akerlind I, Rydh H, Bjurulf P. Epidemiology of sickness absence in a Swedish county in 1985, 1986 and 1987. A three year longitudinal study with focus on gender, age and occupation. Scand J Soc Med 1994; 22: 27 34. 12 Isacsson A, Hansson BS, Janzon L, Kugelberg G. The epidemiology of sick leave in an urban population in Malmo, Sweden. Scand J Soc Med 1992; 20: 234 239. 13 Stansfield S, Feeney A, Head J, Canner R, North F, Marmot M. Sickness abscence for psychiatric illness: the Whitehall II Study. Soc Sci Med 1995; 40: 189 197. 14 Tellnes G. Days lost by sickness certification. Scand J Prim Health Care 1989; 7: 245 251.