Rates of sickness certification in European primary care: A systematic review

Abstract

Background: General practitioners (GPs) are responsible for assessing a patient's capacity for work and issuing a sickness certificate, enabling a patient to receive statutory sick pay and take time away from the workplace. The management of sickness absence across Europe varies considerably, and there is a need for comparable rates of certification to facilitate appropriate health and economic planning. Objective: To systematically review the literature reporting rates of sickness certification in general practice settings. Methods: Electronic databases were searched from their inception to November 2007. Inclusion criteria were reporting a measure of sickness certification, conducted in European primary care. Results: 298 citations were identified from the literature search, of which 11 met the inclusion criteria. These studies demonstrated that the rates of sickness certification are not routinely recorded. The certified rates were subject to wide variation, ranging from 18 per 100 person years in Norway to 239 per 100 person years in Malta.

Conclusion: There is large variability in sickness certification policy and hence sickness certification rates across Europe. A system that enables comparisons across countries would be beneficial in ensuring health and economic planning. To enable a baseline rate of certification to be established and compared across countries, standardized reporting of sickness certification is needed.

• Systematic review
• sickness certification
• Europe
• primary care
• general practice

Introduction

General practitioners (GPs) in the United Kingdom (UK) and in many other European countries act as the gatekeepers to both secondary care and some social security benefits including statutory sick pay [1]. GPs are responsible for assessing a patient's capacity for work and issuing a sickness certificate, enabling a patient to take time away from the workplace if the GP considers this necessary. Although all doctors are able to issue sickness certificates, GPs are the group who most often issue them [2], with Tellnes et al. reporting that 80.9% of all initial certificates were issued by GPs [3]. It has been estimated that, on average in the UK, GPs will issue approximately 20 sickness certificates per week, the majority of which will be for short spells of incapacity [4]. The legislation, and consequently the management, of sickness absence across Europe and the rest of the world varies considerably, with different countries enforcing different periods of time available for self-certification (the duration of sickness absence before which a sickness certificate is needed).

Although there have been a number of studies assessing sickness certification, a comparison of the rates reported from differing countries and differing healthcare systems has not been carried out. A comparison of certification rates across Europe would provide useful information from which to evaluate the degree of certified sickness absence that is occurring, and enable the methods currently employed to manage certification to be examined. A review of research estimating the rate of sickness certification would be beneficial in planning appropriate healthcare services by understanding which conditions lead to the greatest impact on work absence, and in estimating the costs associated with sickness absence from the workplace across Europe.

The objective of this study was to systematically review the literature reporting rates of sickness certification in general practice settings.

Methods

Search strategy

Systematic searches of online electronic bibliographies (AHMED, BNI, CINAHL, DHData, EMBASE, Kingsfund, PsycINFO, and MEDLINE) identified studies for potential inclusion in the review. All databases were searched from their inception until November 2007 using the following set of key words: primary care, primary healthcare, general practice, family practice, family physicians, generalists, general practitioners, primary care physicians, sickness certification, sickness certificate, sick note, sickness absence, sick leave, absenteeism, work absence, illness days, disability leave, and sick days (the full search is available on request). The search strategy was developed with clinicians (general practice and nursing) and epidemiologists experienced in literature searching and systematic reviewing. The reference lists of papers selected for inclusion in the review were also hand searched to check for additional papers. All available abstracts from the European Public Health Association (EUPHA) annual meeting were also reviewed (from the year 2006) as were all listed publications by EUPHA members.

Study selection

To be included in the review, studies had to report a measure of sickness certification and be conducted in primary care; studies had to be conducted within Europe, and participants had to be adults; studies carried out in other languages were included where translation was possible. A search for duplicates was conducted at this stage to identify unique citations only.

The titles of all studies identified from the search strategy were screened, and those clearly not meeting the inclusion criteria were excluded at this stage. The abstracts and full texts of the remaining studies were then appraised by two reviewers to assess eligibility; those not meeting the inclusion criteria at this stage were again excluded. Figure 1 shows the flow of citations from the initial search to inclusion in the final review.

Figure 1.  Flow chart showing the results of the systematic search and selection of studies.

Data abstraction and quality assessment

The following information was abstracted from each eligible study: the country the study was conducted in, the population included, the data source and the data collection period, the number of days of work absence from when a certificate is required, the number of certificates issued, the rate of certification, and any other results reported in relation to the rates of certification in specific groups or patients.

Where studies did not report a rate of sickness certification, but the data presented could be used to calculate a rate, this was calculated by dividing the reported number of certificates issued per year by the number of patients in the population. This calculation provided the crude rate of sickness certification for that population.

The quality of each study was assessed using a checklist consisting of 11 items (Table I). This checklist was based on the Newcastle Critical Appraisal Tool developed by Darzins et al. [5]. Although there are a number of critical appraisal tools available, this tool was chosen as it is not study specific, and could therefore be used to appraise the range of different methods it was expected this review would include. Two reviewers (G.W.-J., C.D.M.) quality appraised all the papers. Each paper was scored according to its quality; assessment of each item identified in Table I generated a positive score “+” if the item was clearly presented, or a negative score “−” if the item was missing or not discussed fully enough to judge the quality. Where it was unclear whether an item was adequately reported, the reviewers discussed the paper until a consensus could be reached. By adding together the positive scores of each paper, an overall quality assessment was calculated. Paper scores could range from 0 to 11, with a higher score indicating higher-quality reporting.

Table I.  Items used to assess the quality of studies.

Item	Assessment
1. What is the research question?	Is the research question clearly presented? Does the research question follow from the rational?
2. What is the study type?	Is the study type appropriate to the research question?
3. What are the outcome factors, and how are they measured?	Are all the relevant outcomes assessed? Is there measurement error? Is measurement error a source of bias?
4. What are the study factors, and how are they measured?	Is there measurement error? Is measurement error a source of bias?
5. What important confounders are considered?	Are potential confounders controlled for? Is confounding a potential source of bias?
6. What are the sampling frame and sampling method?	Is there selection bias? Does the way participants are selected threaten the external validity of the study?
7. Issues of internal validity	How many participants reached follow-up? Are control participants appropriate? Does this threaten the internal validity of the study?
8. Are statistical tests considered?	Are the tests appropriate to the data? Are power calculations and confidence intervals reported?
9. Are the results clinically/socially significant?	Was the sample size large enough to detect a clinically/socially significant result? Is the study inconclusive?
10. Is the study ethical?	Is ethical approval reported?
11. What conclusions did the authors reach about the study question?	Are the conclusions supported by the results?

Adapted from Darzins et al. [5].

Results

A total of 298 citations were identified after searching the bibliographic databases; over three-quarters of these citations were retrieved from the predominantly medical databases MEDLINE (45% of studies identified) and EMBASE (32% of studies identified). After screening by title, 194 abstracts met the inclusion criteria for review; of these abstracts, 26 met the eligibility criteria and full papers were obtained (Figure 1). After reviewing all 26 full-text papers, 11 were retained for inclusion in the review (10 did not report data on the rate of certification, one included children, two were literature reviews, one was a qualitative study, and one was a vignette study).

Quality assessment

Quality assessment scores ranged from 5 to 9 (Table II). The majority of the studies did not report ethical approval; although this may be a function of the study type included—record review—analyses of which do not always require specific ethical approval. All studies stated clear research questions or objectives, reported appropriate study designs, and clearly reported outcome measures.

Table II.  Results of quality assessment.

Quality criteria	1 Research question	2 Study type	3 Outcome measures	4 Other study measures	5 Confounders	6 Sampling	7 Internal validity	8 Statistical tests	9 Clinical/social significance	10 Ethical issues	11 Conclusions	Score
Author*
Eade et al. [11]	+	−	+	−	−	+	−	−	+	−	+	5
Tellnes et al. [6]	+	−	+	+	+	+	−	+	+	−	+	8
Tellnes et al. [3]	+	+	+	−	−	+	−	+	+	−	+	7
Englund et al. [7]	+	+	+	+	−	+	−	+	+	−	+	7
Soler [12]	+	+	+	−	−	+	−	+	+	−	+	7
Shiels et al. [13]	+	+	+	+	−	+	−	+	+	−	+	8
García-Díaz et al. [14]	+	+	+	+	+	+	−	+	+	−	+	9
Hensing et al. [8]	+	+	+	+	−	+	−	+	+	−	+	8
Norrmén et al. [9]	+	+	+	+	−	+	−	+	+	+	+	9
Ihlebaek et al. [10]	+	+	+	+	−	+	−	+	+	−	+	8
Bollag et al. [15]	+	+	+	+	+	+	−	−	+	−	+	8

*Arranged by publication date.

Study characteristics

The characteristics of the studies included in the review are presented in Table III. The majority of studies included in the review were conducted in the Scandinavian countries (six studies [3], [6–10]). Apart from three studies published in 1979 [11] and 1989 [3], [6], all studies were published from 2000 onwards. Record review was the most commonly used method of data collection, either from National Insurance Office records, in Norway in particular [3], [6], [8], or from records of general practitioners or family physicians in Malta and the UK [11–13]. Seven of the papers collected data over multiple time points or throughout the year, to avoid potential bias of results by season [8], [10–15].

Table III.  Study characteristics.

Authors	Year of publication	Country	Population	Data source	Data collection period
Eade et al. [11]	1979	UK	All patients registered at one general practice in the UK	Consultation data – unclear on how this was obtained	Four 2-week periods in Feb, May, July, and Oct 1976
Tellnes & Bjerkedal [6]	1989	Norway	16–69-year-old employed persons in one county with an initial certificate	National Insurance Office of the county, 5042 certificates	4 weeks, 25 February to 24 March 1985
Tellnes et al. [3]	1989	Norway	16–69-year-old employed persons in one county with an initial certificate	National Insurance Office of the county, 5042 certificates	4 weeks, 25 February to 24 March 1985
Englund & Svärdsudd [7]	2000	Sweden	53 GPs in one country in Sweden	Audit/protocol completed by GP	Two-week period in March 1996
Soler [12]	2003	Malta	All 421 employees at five Maltese organizations	Employer family physician records	All certificates issued between 1997 and 1999
Shiels et al. [13]	2004	UK	All persons registered at any of nine GP practices, >50 000 employed persons	Carbon copies of sickness certificates issued to patients	Over a period of 12 months
García-Díaz et al. [14]	2006	Spain	All patients registered at one health centre	Health centre records	January 2000 until December 2002
Hensing et al. [8]	2006	Norway	All persons aged 16–66 who were entitled to sickness benefits in 1994, 1996, 1998, and 2000	Norwegian National Insurance Administration	Data were collected for a full year in 1994, 1996, 1998, and 2000
Norrmén et al. [9]	2006	Sweden	All persons aged 18–64 years registered with a GP in Örebro county	Questionnaire	1996
Ihlebaek et al. [10]	2007	Norway	All persons aged 16–67 who were working full or part time	Questionnaire survey of selected participants from a national survey	Over a period of 3 months in the spring of 1996 and spring of 2003
Bollag et al. [15]	2007	Switzerland	Physicians with primary care activity participating in the Swiss Sentinel Surveillance Network	Swiss Sentinel Surveillance Network	1 year, 1 Jan to 31 Dec 2005

Period when certification is required

The period from when certification is required, and therefore the number of days individuals can self-certify work absence, varies considerably (Table IV). The least amount of time allowed for self-certification is in Malta, where a sickness certificate must be issued from the first day of absence, although this is facilitated by organizations employing private GPs who will visit employees to assess their work ability. The UK and Sweden allow self-certification for the greatest period, 7 days in total (5 working days), with the most common period of self-certification at 3 days for the Scandinavian countries.

Table IV.  Rates of certification.

Authors	Period when certification required	Number of sickness certificates issued	Rate of certification	Comparable rate of certification*	Other results reported
Eade et al. (1979) [11]	Not stated, but UK is >7 days	53 certificates issued to 288 rheumatic patients	0.54 per person per year Calculated from data presented	54 per 100 person years	• More men received certificates, but there were more men eligible for certification, as more women were reported as not working • There was a relationship between the degree of symptoms and certification • For BP patients occupation was also associated with certification
Tellnes & Bjerkedal (1989) [6]	>3 days	5042 (4858 people)	580 per 1000 employed persons per year	58 per 100 employed persons per year	• Highest incidence in 20–29-year-olds • Highest incidence in service/manufacturing municipalities
Tellnes et al. (1989) [3]	>3 days	5042 (2257 female, 2785 male)	580 per 1000 employed persons per year (595.8 females, 567.7 males)	58 per 100 employed persons per year	• GPs issued 81% of certificates, hospital drs 13%, private consultants 2%, occupational drs 2%, and others/unknown 2% • Most frequent groups were respiratory system and musculoskeletal/connective tissue diseases
Englund & Svärdsudd (2000) [16]	>7 days	68 certificates	Calculated as 68/301=22.6 per 100 patients	22.6 per 100 consultations	• Sick-listing cases constituted 8.8% of all consultations • Median length of sick leave was 14 days, and 25% of all sick listing was partial • Pain conditions were the most common reason for sick listing
Soler (2003) [12]	>1 day	3015 episodes of certification	239 per 100 person years (23.9 per year)	239 per 100 person years	• Average duration of sickness episode was 2.9 days • The three most common episodes were upper respiratory tract infections, sprains and strains and gastro-enteritis • Very few certificates were issue for depression and anxiety
Shiels et al. (2004) [13]	Not stated, but UK is >7 days	13 127 certificates issued to 6271 patients	Calculated as 2.1 per person per year	210 per 100 person years	• Mean duration of certification of 9.9 weeks/69 days • Almost 10% of patients had a certificate duration of >28 weeks
García-Díaz et al. (2006) [14]	Not stated	1714 certificates issued to 1146 employees	14.7 per 100 persons per year	14.7 per 100 person years	• Musculoskeletal conditions accounted for the majority of certificates • Mean number of days lost per worker was 99.7 • Multivariate analysis demonstrated that longer episodes of absence were associated with being female, older age, self-employed, or a home worker
Hensing et al. (2006) [8]	>4 days	Not reported	Cumulative incidence of certification with psychiatric diagnosis was 1.7% in 1994, 4.6% in 2000 in women	—	• The incidence of certification with psychiatric diagnosis was higher in women than men • The incidence of certification with psychiatric diagnosis was higher in middle age (40–49 years) than in other groups • The incidence of certification with psychiatric diagnosis increased in all regions of Norway
Norrmén et al. (2006) [9]	>7 days	152 certificates	Calculated as 152/642×100=23 per 100 patients	23 per 100 consultations	• Experienced physicians certify patients sick more often than less experienced physicians • Physicians working part time certify patients with infectious diseases sick more often than physicians working full time • Physicians trained in social insurance medicine certify patients with musculoskeletal problems sick more often than physicians who do not have such training
Ihlebaek et al. (2007) [10]	>4 days	174 564 (in first 6 months of 1996)	0.18 per person per year calculated by data presented	18 per 100 person years	• There was poor concordance between % change in health complaints and corresponding sickness absence • The prevalence of back pain complaints and back pain related sickness absence remained unchanged over 7 years • Tiredness was the single complaint with the largest change in sickness absence, with an increase of >350%
Bollag et al. (2007) [15]	>3 days	24 847 certificates from 162 physicians	GPs issued 4.3 certificates per 100 consultations Internists issued 3.6 certificates per 100 consultations	4.3 per 100 consultations	• Psychosocial and work-related cofactors were mentioned on 19% of certificates • Longer absence from work (>3 weeks) was associated with a higher proportion of psychosocial and work-related cofactors

*Re-calculated to rate per 100 (if not already reported) to enable comparisons.

Rates of sickness certification

The presentation of rates fell into three distinct categories; they were reported in person-time as certificates per person or per 1000 persons per year, or as certificates per 100 consultations.

Certification rates per person

Three papers reported data per person per year [10], [11], [13], and the rates of certification they present are highly variable: the lowest rate was reported by Ihlebaek et al. [10] at 0.18 certificates per person per year; the highest rate was reported by Shiels et al. [13] at 2.1 certificates per person per year.

Certification rates per 1000 person years

Just two papers reported data per 1000 persons per year [3], [6]. Both these papers were reporting data from the same cohort. A total of 580 certificates were issued per 1000 employed persons per year in this one study conducted in Norway.

Certification rates per 100 consultations

The majority of papers, five in total, reported data per 100 consultations [9], [12–16]. The findings differed considerably between papers, with the lowest rate reported by Bollag et al. [15] at 4.3 certificates per 100 consultations in Swiss GPs. The rates per 100 consultations are much higher in the studies reported by Norrmén et al. [9] at 23 certificates per 100 patients and Englund & Svärdsudd [7] at 22.6 per 100 patients.

After re-calculating the rates, so they are all based on certification rate per 100 person years, to enable more direct comparisons, it can be clearly seen that the rates of certification vary widely from 18 to 210 per 100 person years in studies conducted by Ihlebeak et al. [10] in Norway and Shiels et al. [13] in the UK, and up to 289 per 100 patient years in a study conducted by Soler et al. [12] in Malta.

One study reported the cumulative incidence of certification with psychiatric diagnoses to be 1.74% in 1994 and 4.6% in 2004 [8]. The figures to translate this into rate of certification per 100 person years were not reported.

Denominator

The denominators used by each of the studies differed slightly: six studies used employed persons [3], [6], [8], [10], [12], [14] and five used the number of consulters [9], [11], [13], [15], [16]. The variation in rates of certification by denominator is linked to the method of reporting, such that those reporting rates of certification per 100 consultations use a denominator based on number of consultations during a specified period. Those papers that report their results per 100 person years are based on employed populations. The rates of certification for each are reported above. In summary, there are wide variations in the rates of certification by denominator, and it is generally reported in limited detail, with just one paper (Tellnes & Bjerkedal [6]) describing the denominator clearly.

Discussion

Summary of main findings

This is the first paper to systematically review rates of sickness certification in primary care. Although the methods used to identify the issue of a sickness certificate are largely similar, with the majority of studies utilizing a record review, the actual rates of certification vary considerably between studies.

There are a number of possibilities for the observed differences in certification rates. The first is the permitted duration of self-certification. The study reporting the highest rates of certificates issued by GPs is conducted in the country permitting the shortest duration of self-certification, meaning that certificates would be required for short sickness absences where they would not be needed in other countries [12]. However, the countries with the longest period of self-certification (the UK and Sweden) [9], [13], [16] also reported high rates of certification when compared to countries with shorter periods of self-certification, such as Norway and Switzerland [8], [10], [15]. It would seem logical that the country with the shortest period of self-certification should have higher rates of certification, but it is unclear why the countries with the longest periods of self-certification should also have high rates of certificates issued by GPs.

The system by which sickness absence benefits are paid may also impact upon the rate of certification. Within the UK, the employer pays for sickness absence benefits for up to 28 weeks of incapacity as long as a sickness certificate is provided [13], whereas in Norway the employer pays for only the first 16 days, after which point responsibility is passed to the National Insurance Administration [10]. In Malta, the employer pays the first 3 days of sickness absence; after this period, a National Insurance certificate is required to claim benefits, which are paid jointly by the employer and the state [12]. The Swedish system requires employers to cover the first 2 weeks of absence [17]. It appears that, in general, the longer the period of sickness absence benefit paid for by employers, the higher the rate of certification, compared to those countries where the bulk of the sickness absence benefits are covered by the state.

The denominators were not often clearly reported; it may be expected that the rate of certification based on healthcare consultations would be higher than that based on number of employed persons. However, the variation in the rates of certification when looked at by denominator makes it difficult to draw any conclusions other than to advocate clearer reporting.

None of the studies included in this review reported on the validity of the records that were used to calculate the rate of certification. It is important to ensure that the source of data used to calculate the rates of sickness certification is accurate and up-to-date. Tellnes et al. [3] advocated the implementation of a routine information technology system to collect data on sickness certification, allowing the evaluation of occupational health and enabling appropriate planning of healthcare. This is a system that would be useful across Europe.

The use of a standardized system to collect data on sickness certification in each European country would be of benefit to overcome one of the major difficulties in comparing rates of certification. Currently, the actual rates reported provide only limited information. For example, the rate reported by Tellnes et al., of 58 certificates per 100 employed persons per year, could refer to one patient with 58 certificates or 58 patients with one certificate. To facilitate comparisons, it should be recommended that a minimum set of data is routinely collected when a sickness certificate is issued (Figure 2).

Figure 2.  Recommendations for a minimum data set when collecting sickness certification data.

Strengths and weaknesses of this paper

This review has identified 11 papers that report the rate, or enough data to calculate the rate, of sickness certification in primary care.

This review did not specifically address all of the grey literature; abstracts from the most relevant conference (EUPHA) were reviewed, but other “grey” sources (i.e., reports, conference proceedings, dissertations, and theses) that may report sickness certification rates were not searched. However, a search of each paper's citations did not elicit any further literature that may have contributed to the review. In addition, the search from EUPHA, of both conference abstracts and listed publications, elicited just one abstract, Bollag et al. [15], the full paper of which had already been identified through the systematic search of online databases. Obtaining data relating to sickness absence in Europe has been demonstrated to be very difficult due to the insufficient availability of comprehensive, reliable, and comparable data on sickness absence [18]. Furthermore, a search of the reference lists of papers identified from online databases did not highlight reports of certified sickness absence at either a local area level or a governmental level.

No paper was excluded on the basis of the quality assessment, and the majority of studies scored in the higher end of our assessment scale (≥7). Excluding the study with the lowest score on the quality assessment [11] did not change our conclusions regarding the variability of the rate of certification. There is the potential for publication bias in a systematic review, but it is unlikely that a paper reporting exceptionally high or low rates of certification would not be published on that basis. Furthermore, the search strategy developed for this review included a broad range of databases covering medical, allied health professional, and psychological literature; a topic such as sickness certification, which is so intrinsically linked to healthcare, is unlikely to be widely reported in non-medical journals.

Implications of this paper

UK GPs are required by law to provide medical statements (sickness certificates) recording the advice they have given a patient regarding sickness absence, in the context of that patient's ability to carry out their usual job [1]. The process of sickness certification, whilst centred on an individual's ability to carry out his/her job, incorporates a range of other social and psychological aspects in the complex decision-making process that results in the issuing of a sickness certificate. High levels of sickness certificates represent an additional burden to the GP in a number of ways, including time available in the consultation to address issues relating to certification [19], the potential for conflict between the GP and the patient [2], [20], and the role of gatekeeper to social security benefits that the GP has to play [4], [20].

Conclusion

In conclusion, there is a general lack of research documenting the rate of sickness certification across European countries, making comparisons both between and within countries very difficult. It is important to put in place the systems to generate comparable rates of certification across Europe. With current practice differing in both self-certification and GP-certified absence across Europe, recording the number of certificates and some basic certification information, such as health condition and duration of the certificate, would be of benefit to both policy makers and employers. Accurate figures for the rate of sickness certification would help the planning and implementation of systems to support individuals to remain at work or return to work after a period of certified sickness absence.

Acknowledgements

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