541019 research-article2014

SJP0010.1177/1403494814541019Variation in patients’ sick leave between general practitioner practicesM. Rudbeck

Scandinavian Journal of Public Health, 2014; 42: 621–626

Original Article

Variation in patients’ sick leave between general practitioner practices

Marianne Rudbeck Department of Social Medicine, Aalborg University Hospital, Denmark

Abstract Background: General practitioners (GPs) assess the existence of the patient’s disease, decide whether the disease affects the patient’s ability to work and if necessary, recommend sick leave. Our aim was to describe correlations in patients’ sick leave between GP practices (GPPs) in a 5-year period. Method: The study included 253 GPPs, from 2007 to 2011. The personal numbers of patients from each GPP were connected to DREAM, a registry at the Danish Ministry of Employment that includes social welfare payments, including sick leave benefits. We adjusted for patient age, gender, ethnicity and social differences. Spearman’s rank correlation coefficients (2007 – 2011) were used for calculating the correlation in adjusted sick leave. Results: The number of patient sick leave weeks between GPPs varied from 36 to 2,704 sick leave weeks per 1000 patients (18 – 65 years). The correlation coefficients for adjusted sick leave weeks varied from 0.90 to 0.94 (P < 0.05). Correlations for the 10 GPPs with the highest number of sick leave weeks and the 10 GPPs with the lowest number of sick leave weeks were almost as high as the correlations of the total population of GPPs. Conclusions: The study showed great differences in sick leave between GPPs; however, significant correlation for adjusted sick leave in each GPP was demonstrated. This may indicate that GPs play an important role in their patients’ sick leave. The study provides a method to distinguish between GPPs with low patient sick leave and high patient sick leave in causal studies of sick leave differences among GPs. Key Words: Employment, general practitioner, medical practise, registry study, sick leave

Background General Practitioners (GPs) are gatekeepers in the sick leave benefit system in most western European countries. Every inhabitant in Denmark has a right to a GP nearby. Patients select their GP based on personal wishes and generally use the same GP practise (GPP) for years (their GPP is mostly situated in their neighbourhood). GPs assess the existence of disease, decide whether the patient’s disease affects their ability to work, and if necessary, recommend sick leave as well as the duration and grade (full or part time) of the sick leave. The GP will fill in a sickness certificate for the municipality on request, stating the medical diagnosis and duration, along with the recommended rehabilitation or activities. The municipality claims, in general, a certificate after 4 – 8 weeks of sickness

absence. The employers generally pay the sick leave benefit for the first 3 weeks, and from then on the municipality pays the sick leave benefit. Sickness absence may be issued by physicians other than GPs; however, sickness absence certificates are mostly issued by the GP. The main predictive factors for length of sick leave have been described to be: diagnosis, age, gender, family circumstances, economic incentives and restrictions in insurance legislation, type of work, social norm, and the functioning of the labour market [1,2,3,4]. Individual patient factors have been described to be the most important factors in sickness absence certification [1,5], and 66% of GP assessments of work ability are described as based

Correspondence: Marianne Rudbeck, Department of Social Medicine, Aalborg University Hospital, Havrevangen 1, DK- 9000, Aalborg, Denmark. Email: [email protected] (Accepted 26 May 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814541019

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

622    M. Rudbeck on patients’ statements and 34% are based on medical findings [6]. Diagnosis has been found to explain 18% of long-term sick leave (< 28 weeks), while clinician and general practice explain 3.4% and 2.3%, respectively [3]. Variance in sick listings is described to be partly (23%) due to GP factors [7]. It is reported that GPs highly value their relationships with their patients [8] and they take the patients’ input into consideration during the sickness certificate process, for a number of reasons [9]. On the other hand, it has been demonstrated that patient demands do not influence the sick leave certification [10]. About 25% of GPs have patients whom want to be absent due to sickness for other reasons than medical work incapacity [11]; and previously, GPs allowed sickness certification on purely social grounds, even in countries with high social security [2]. Previous studies were seen to be diverse about GP factors, including GP behaviour and attitude [12,13]. Patients who want sick leave for non-medical reasons can be difficult for the GPs to handle, if the GPs cannot stick to a purely medical recommendation of sick leave, and it may result in differences between the sickness practices among GPs. We have not found any studies that have reported sick leave over years in connection with GPPs. Aim The aim of the present study was to describe the variation in patients’ sick leave between GPPs in a 5-year period. If we found consistency in sick leave in each GPP, we could use a similar method to identify the GPPs with low patient sick leave and high patient sick leave, for further studies of GP factors leading to assignment of low and high amounts of sick leave, respectively. Materials and methods The present study included all GPPs (N = 253) in a region of about 600,000 inhabitants over a 5-year period, from 2007 to 2011. The total amount of patients (18 – 65 years of age) went from 403,506 in 2007 to 411,495 in 2011. GPPs with less than 500 patients between 18 and 65 years of age were excluded. GPs usually have a unique provider number, though a few GPs in the same GPP share a provider number. The provider numbers and the patients’ personal numbers were combined to identify the group of patients in each GPP by the first of July, in each year. We extracted data on sick leave, social benefits and ethnicity from each patients’ personal number in the DREAM registry [14].

The DREAM registry is run by the Danish Ministry of Employment and contains weekly information on all social welfare payments for the Danish population, since 1991. The DREAM database was proven useable for follow-up analysis of the social and economic consequences of disease [15]. The type of transfer payment is recorded for a full week, even if a person received the benefit for just 1 day. Part time sick leave benefit is recorded as sick leave. The registry contains only data on sickness absence, if the duration was longer than 3 weeks since 2008 (as the first 3 weeks are paid by the employer); however, if a person got unemployment benefits or temporary social benefits before his/her sickness absence, the sick leave benefit is registered from day one. If ill, everybody has the opportunity to receive sick leave benefit for 1 year; however, under certain circumstances that benefit can be extended. Temporary social benefits are for people without a job whom are unable to support themselves; temporary social benefits are, consequently, always available for people in need. We assumed that patients with high social needs have a higher sickness absence than patients with no social needs. GPPs differ in their patient base, according to location. The number of weeks the GPP’s patients were on temporary social benefits was, therefore, used as a proxy for the socioeconomical patient differences among GPPs in the present study. Gender, age and list length of the GP have been found not to be associated with a sickness absence >16 days [16,17]. Accordingly, no adjustments were made for these GP factors in the present study. Patients who belonged to a GP list of > 10% of persons on disability pension had a higher risk of being on a sick leave > 16 days [17]. A GP list with a great amount of highly-educated and elderly patients has reduced sickness absence. Men on a GP list with a high proportion of men have a low risk for sickness absence [17]. These findings justify our adjustments for patient age, gender and socio-demographic factors, when comparing sickness absences in GPPs. We included only people in the workforce (18 – 65 years of age) and excluded GPPs with less than 500 patients 18 – 65 years old. Generally, GPs have around 1600 patients. GPs in this study had, in general, from 1650 to 1750 patients and a few GPPs had more than 1750 patients [18]. GPPs consist of one to multiple GPs, but consist generally of one or two GPs. There happens to be only a few replacements of GPs over the years. The patients’ number of weeks on temporary social benefits was calculated in each GPP. The patients’ number of weeks on sick leave in each GPP was adjusted for age and gender, in Model 1; whereas in Model 2, the number of weeks on sick leave was adjusted for age, gender, ethnicity and temporary

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

Variation in patients’ sick leave between general practitioner practices    623 Table I. Basic characteristics of the GPPs, 2007 – 2011.

Number of GPPs Number of GPPs with > 500 patients (18 – 65 y) Mean number of patients aged 18 – 65 y (min-max)  Mean age Women/men (min-max)   Ethnicity: non-Danish/Danish   Mean sick leave weeks/1000 patients (min-max)  Mean social benefit weeks (min-max)  

2007

2008

2009

2010

2011

253 220

248 214

245 210

246 207

253 206

1825 (561 – 6590) 41.3 0.97 (0.38 – 4.47) 0.068 (0.008 – 0.632) 1323 (76 – 2342) 924 (30 – 4810)

1873 (570 – 6719) 41.2 0.97 (0.42 – 4.31) 0.072 (0.008 – 0.661) 1583 (41 – 2704) 988 (5 – 3863)

1916 (551 – 7307) 41.3 0.97 (0.40 – 4.08) 0.076 (0.008 – 0.719) 1559 (64 – 2599) 1140 (30 – 3739)

1942 (532 – 6782) 41.2 0.98 (0.40 – 3.18) 0.079 (0.007 – 0.954) 1474 (36 – 2403) 1184 (48 – 3425)

1991 (525 – 6921) 41.1 0.96 (0.44 – 2.55) 0.084 (0.007 – 1.176) 1413 (39 – 2225) 1286 (36 – 3513)

GPP: general practitioner practices; min-max: minimum and maximum number range; y: years.

social benefits, by the general relative fractions in the region. Ethnicity was defined by the numbers of firstand second-generation immigrants. Sick leave weeks in Model 1 were normally distributed; however, in Model 2, they were skewed to the right. We did the statistical analyses with Stata 11.0 software (StataCorp LP, TX, USA). We used Spearman’s rank correlation coefficients for calculating the correlation coefficient matrix (2007 – 2011) in adjusted sick leave. The correlation coefficient matrix (2007 – 2011) was also calculated for the 10 GPs with highest sick leave and for the 10 GPs with the lowest sick leave in 2011. This study was approved by the Danish Data Protection Agency (2008-58-0028). A registry study does not need approval by the Regional Scientific Ethical Committee. Results Table I describes the basic characteristics of the GPPs during the years 2007 – 2011. Mean sick leave in 2008 was significantly higher than in 2007, mean sick leave in 2010 was significantly lower than in 2009 and significantly higher than in 2011 (data not shown); otherwise, there was no major variations in sick leave. The amount of the patients’ sick leave in GPPs varied within the same distributional pattern throughout the years, but sick leave in 2007 was lower than the other years (Figure 1). There were great differences between Model 1 and Model 2 in each year (Figure 2), with correlation up to 0.15 (data not shown). These model differences demonstrated the need to adjust for socioeconomic patient differences between the GPPs. Table II describes that the correlation in Model 2 between adjusted sick leave weeks in each GPP

varied from 0.90 to 0.94 between consecutive years; this was somewhat higher than the correlation in Model 1, which was only adjusted for age and gender. Table II demonstrates that the correlation coefficients were highest between consecutive years. Correlation of consecutive years between the 10 GPPs with the lowest sick leave was, in general, almost as high (0.67 – 0.99) as the correlation of the total population (Table III). Correlation of consecutive years between the 10 GPPs with the highest sick leave was lower (0.21 – 0.84) than the correlation of the total population; the year 2007 did not, in general, correlate with the other years. The year 2007 was the last year before the regression. Discussion The present study demonstrated that variation in the number of sick leave weeks between patient populations of GPPs was great, even when adjusted for patients’ age, gender, ethnicity and socio-demographic factors; however, the correlations in sickness absence between each GPP over the years were very high. The study included GPPs in a region consisting of 600,000 inhabitants, situated in Northern Denmark. This region includes both remote rural areas and urban districts, like the fourth largest city in Denmark. Sick leave in the region is about equal to and especially not higher than the average for Denmark (about 1500 sick leave weeks per 1000 inhabitants in 2012) [19]; however, GPs in this region have more patients than the average number of patients per GP, due to a lack of GPs in the region. GPs in the most remote areas have the highest number of patients. The patients each select their own GPP in their neighbourhood, within a radius of

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

624    M. Rudbeck 4000

Paents' sick leave weeks

3500 3000 2500 2000 1500 1000 500 0 1

11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171 181 191 201 211 2007

2008

2009

2010

2011

GPP

Figure 1. Ranked distribution of patients’ sick leave weeks in the study GPPs, 2007 – 2011. GPP: general practitioner practise.

10,000 9,000

Paents' sick leave weeks

8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176 183 190 197 204

0 GPP model 1

model 2

Figure 2. Variation of patients’ sick leave weeks in the GPP in Model 1 and Model 2 in 2011a). aBased on ranked distribution of Model 1. GPP: General practitioner’s practice, Model 1: patients’ sick leave weeks were adjusted for age and gender, Model 2: patients’ sick leave weeks were adjusted for age, gender, ethnicity and temporary social benefits.

15 km. Patients rarely change GPs on their own initiative, because they will be charged a fee. We are unable to explain how the patients’ choice of GP influences sick leave, but it seems unlikely that the patients’ choice can explain the large differences in

sick leave between GPPs over the years; however, it is known that patients influence the sickness certification process, but it remains unknown to what extent this hinders or helps the return-to-work process [9]. Both the patient’s and the GP’s expectations are

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

Variation in patients’ sick leave between general practitioner practices    625 Table II. Correlation coefficients in adjusted sick leave weeks between years. Model 1a

Model 2b



2007

2008

2009

2010

2011

2007

2008

2009

2010

2011

2007 2008 2009 2010 2011

1.00 0.80c 0.78c 0.73c 0.56c

– 1.00 0.83c 0.80c 0.62c

– – 1.00 0.89c 0.67c

– – – 1.00 0.71c

– – – – 1.00

1.00 0.90c 0.83c 0.74c 0.72c

– 1.00 0.92c 0.82c 0.80c

– – 1.00 0.90c 0.86c

– – – 1.00 0.94c

– – – – 1.00

aModel

1: Patient sick leave weeks, adjusted for age and gender. 2: Patient sick leave weeks, adjusted for age, gender, ethnicity and temporary social benefits. cP < 0.05. bModel

Table III. Correlation coefficients in adjusted sick leave weeks between years, between the GPPs with highest and lowest sick leave. 10 GPPs with highest sick leave

10 GPPs with lowest sick leave



2007

2008

2009

2010

2011

2007

2008

2009

2010

2011

2007 2008 2009 2010 2011

1.00 0.21 0.11 − 0.14 − 0.34

– 1.00 0.79a 0.25 0.50

– – 1.00 0.76a 0.85a

– – – 1.00 0.84a

– – – – 1.00

1.00 0.90a 0.85a 0.40 0.60

– 1.00 0.99a 0.61 0.80a

– – 1.00 0.67a 0.83a

– – – 1.00 0.94a

– – – – 1.00

Calculations were based on GPPs in 2011 Model 2 (adjustment of patient age, gender, ethnicity and temporary social benefit). aP < 0.05. GPP: General practitioners’ practices.

known to influence sick leave duration; therefore, these expectations are included as a tool for some sick leave predictions. The Danish College of General Practitioners stands behind the clinical guidelines for GPs, but the guidelines rarely include instructions on sick leave. The present study is limited by being a registry study, which means we can describe the numbers, but not tell anything about the causes; and therefore, we cannot explain the GPs’ influence on the differences in sick leave shown. The study is also limited by the fact that we can only describe sick leave > 3 weeks. Furthermore, we cannot distinguish between the single GPs, because some GPPs consist of two or more GPs. We have adjusted for the most-known patient factors that are known to correlate with sick leave. GP factors seem to be of diverse importance in the literature and have therefore not been accounted for in the study. National differences, such as legislation and local structural factors, can also influence sick leave certification. Even in quite similar nations like the Scandinavian countries there are differences in sick leave certification [4,20,21]. A resent Scandinavian study describes similar sick leave decisions in Norway, Sweden and Denmark. The study did not investigate nor question the international GP differences in some of the patients’ sick leave [21].

Their non-commented results of these general differences in sick listing match the differences in sick leave between GPPs found in this study. We may therefore hypothesise that the big difference in sick leave between GPPs in this study maybe could be due to lack of knowledge in unspecified medical conditions and/or medical recommendations on sick leave. The present study demonstrated a high correlation over the years of 2007 to 2011, even in small groups of GPPs with high or low patient sick leave. There was no correlation in the group with high sick leave in the last year of the economic boom (2007), whereas there were correlations the following years of recession, which may confirm that sick leave is influenced by structural economic conditions. GPPs with generally low sick leave showed no differences between the years, which may be due to the already low number of sick leave weeks given. This study benefits from being based on registry data, a long study period of 5 years and a high number of GPPs. The study demonstrated that GPs may have an important impact on sick leave; however, the study is limited by being a registry study and can therefore not present causes to the findings. The study displays opportunities for further studies in sick leave and GPs, by demonstrating high correlations in sick leave over years.

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

626    M. Rudbeck In spite of a great variation in patients’ sick leave between GPPs, the present study demonstrated a high association between GPPs and patients’ sick leave throughout the years. These results indicate that it may be appropriate to focus on the reasons for differences in sick leave among GPPs, to avoid the risk of long-term sickness absence and its implications. Conflict of interest The author declares that there is no conflict of interest. Funding This research received no special grant from any funding agency in the public, commercial, or nonfor-profit sectors. References [1] Aakvik A, Holmås TH and Islam MK. Does variation in general practitioner (GP) practice matter for the length of sick leave? A multilevel analysis based on Norwegian GP-patient data. Social Sci Med 2010;70:1590–1598. doi:10.1016/ j.socscimed.2010.01.031 [2] Tellnes G, Sandvik L and Moum T. Inter-doctor variation in sickness certification. Scand J Prim Health Care 1990;8:45–52. [3] Shiels C and Gabbay MB. Patient, clinician and general practice factors in long-term certified sickness. Scand J Pub Health 2007;35:250–256. Doi: 10.1080/ 14034940601072364 [4] Arrelöv BE, Borgquist L and Svärdsudd KF. Influcence of local structural factors on physicians’ sick-listing practice: A population-based study. Eur J Pub Health 2005;15:470– 474. Doi: 10.1093/eurpub/cki029 [5] Winde LD, Hansen HT and Gjesdal S. General practitioner characteristics and sickness absence – a register-based study of 348 054 employed Norwegians. Eur J Gener Pract 2011;17:210–216. Doi: 10.3109/13814788.2011.602060 [6] Reiso H, Nygård JF, Brage S, et al. Work ability assessed by patients and their GPs in new episodes of sickness certification. Fam Pract 2000;17:139–144. [7] Morris J and Watson PJ. Investigating decisions to absence from work with low back pain: A study combining patient and GP factors. Eur J Pain 2011;15:278–285. Doi: 10.1016/j.ejpain.2010.07.002

[8] Macdonald S, Maxwell M, Wilson P, et al. A powerful intervention: General practitioners’ use of sickness certification in depression. BMC Fam Pract 2012;13:82. www.biocentral. com/1471–2296/13/82 [9] Wrapson W and Mewse AJ. Does the doctor or the patient control sick leave certification? A qualitative study interpreting patients´ interview dialogue. Fam Pract 2011;28:202– 209. Doi: 10.1093/fampra/cmq088 [10] Campbell A and Ogden J. Why do doctors issue sick notes? An experimental questionnaire study in primary care. Fam Pract 2006;23:125–130. Doi: 10.1093/fampra/cmi099 [11] Engblom M, Nielsson G, Arrelöv B, et al. Frequency and severity of problems that general practitioners experience regarding sickness certification. Scand J Prim Health Care 2011;29:227–233. Doi: 10.3109/02813432.2011.628235 [12] Söderberg E and Alexanderson K. Sickness certification practices of physicians: A review of the literature. Scand J Pub Health 2003;31:460–474. Doi: 10.1080/14034940310005367 [13] Wynne-Jones G, Mallen CD, Main CJ, et al. What do GPs feel about sickness certification? A systematic search and narrative review. Scand J Prim Health Care 2010;28:67–75. Doi: 10.3109/02813431003696189 [14] Ministry of Employment of Denmark, Labour market board. www.dst.dk/da/TilSalg/Forskningsservice/Data/Andre_ Styrelser (2011, accessed 9 December 2012) [15] Hjollund NH, Larsen FB and Andersen JH. Register based follow-up of social benefits and other transfer payments: Accuracy and degree of completeness in a Danish interdepartmental administrative database compared with a population-based survey. Scand J Pub Health 2007;35:497–502. Doi: 1080/14034940701271882 [16] Starzmann K, Hjerpe P, Dalemo S, et al. No physician gender difference in prescription of sick-leave certification: A retrospective study of the Skaraborg Primary Care Database. Scand J Prim Health Care 2012;30:48–54. [17] Winde L, Haukenes I, Hetlevik Ø, et al. The regular general practitioner and sickness absence: A register-based study. Tidskr Nor Laegeforen 2013;133:28–32. Doi: 10.4040/tidsskr.11.1340 [18] Ryesgaard KK. Praktiserende læger har forskellige arbejdsvilkår. Ugeskr Læger 2014; 9:807. [19] Statistics Denmark, www.dst.dk/nytudg/18075, 2012 (accessed 30 March 2014). [20] Winde LD, Alexanderson K, Carlsen B, et al. General practitioners’ experiences with sickness certification: A comparision of survey data from Sweden and Norway. BMC Fam Pract 2012;13:10: 1–8. [21] Maeland S, Werner EL, Rosendal M, et al. Sick-leave decisions for patients with severe subjective health complaints presenting in primary care: A cross-sectional study in Norway, Sweden, and Denmark. Scand J Prim Health Care 2013;31:227–234. Doi: 10.3109/02813432.2013.844412

Downloaded from sjp.sagepub.com at NORTHERN ARIZONA UNIVERSITY on May 23, 2015

Variation in patients' sick leave between general practitioner practices.

General practitioners (GPs) assess the existence of the patient's disease, decide whether the disease affects the patient's ability to work and if nec...
464KB Sizes 2 Downloads 4 Views