Factors associated with prevalence and types of ‘may be fit’ advice on fit notes: a cross-sectional primary care analysis

Abstract

Background

The ‘fit note’, with the opportunity for the GP to advise that a patient ‘may be fit’ to do some work, was introduced in April 2010.

Aim

To estimate numbers of fit notes with ‘may be fit’ advice, the types of advice, and factors associated with any inclusion of such advice in the fit note.

Design and setting

Cross-sectional analysis of fit note data from 68 general practices in eight regions of England, Wales and Scotland.

Method

Collection of practice fit note data via GP use of carbonised pads of fit notes for a period of 12 months.

Results

The ‘may be fit’ box was ticked on 5080 fit notes (6.4% of all fit notes in study). But there was a wide variation in completion rates across the 68 practices (from 1% to 15%). The most prevalent individual item of advice was to ‘amend duties’ of patient as a prerequisite for return to work (included in 42% of all notes containing any ‘may be fit’ advice). Advice was often incomplete or irrelevant, with some GPs failing to comply with official guidance. Inclusion of any ‘may be fit’ advice was independently associated with the patient being female, less socially deprived and having a physical health reason for receiving a fit note.

Conclusion

Unlike other studies that have relied upon eliciting opinion, this study investigates how the fit note is being used in practice. Findings provide some evidence that the fit note is not yet being used to the optimum benefit of patients (and their employers).

INTRODUCTION

Following a recommendation in the review of the health of the working age population in Great Britain, and a subsequent period of consultation and discussion,^1–5 the ‘fit note’ replaced the MED3, 4 and 5 certificates in 2010. (The MED3 medical statement had been the main note used for certifying prospective periods of sickness absence, the MED5 for retrospective periods and the MED4 was issued as a first stage in assessing the patient’s eligibility for Incapacity Benefit after 28 weeks of sickness absence). Although employees and those out of work and on benefits can self-certify for up to 1 week of sickness absence, longer periods require a certificate recommending absence or work modifications from their GP. The new medical statement was a response to increasing evidence of the benefits of work for an individual’s physical and psychological wellbeing,⁶ and the recognition that an ill person, with appropriate support, may be able to do some work before being fully recovered. Unlike the previous medical statement, the fit note enables the GP to advise that the patient ‘may be fit’ to return to work with appropriate support from the employer. Common methods of support (phasing the return to work, altering work hours, amending normal work duties and making workplace adaptations) are presented in structured form as tick-box options on the fit note. A free-text comments section is also available in order to allow the GP to elaborate on the structured advice or suggest an alternative method of support.

In the first 3 years of fit note use there have been a number of evaluations providing some evidence of effectiveness. A qualitative study of 45 GPs found that the fit note was often used as a justification for the GP to initiate discussion with the patient about potential return to work.⁷ A 2012 survey of GPs (following up a baseline 2010 survey) reported an increase in the proportion of GPs who were positive about the impact of the fit note on patient outcomes and the quality of work-related discussion. In the follow-up survey, a majority of GPs believed that return to work had become more frequent in the previous 2 years.^8,9 A survey of employees, who had been issued a fit note in the previous 12 months, found that 71% agreed that the fit note had been helpful and 68% felt that it had facilitated discussion with their employer about changes in their role.¹⁰

How this fits in

An important change to the sickness certification system was made in 2010, with GPs given the opportunity to advise that, with appropriate support, the patient ‘may be fit’ to return to work. This study uses the largest sickness certification dataset in the UK in order to evaluate GP use of ‘may be fit’ advice in fit notes issued to patients. The findings provide the first indication of how far the new system of certification is meeting stated objectives.

However, the same studies reported some problems in the adoption of the new certification system. Some GPs reported difficulties using the ‘may be fit’ options, and others expressed reluctance to compromise their relationship with the patient by refusing to issue a fit note (even for a ‘non-medical’ reason). A number of GPs complained about the lack of feedback from employers about the efficacy of the advice on the fit note.^8,9 From the employer perspective, there has been concern expressed that GPs are not issuing enough ‘may be fit’ notes.¹¹ This study uses a large national dataset to:

• estimate the prevalence of ‘may be fit’ advice on fit notes;

• explore the types of advice given;

• look at variation across patients, GPs and general practices; and

• estimate the independent effect of a range of factors on the likelihood of a ‘may be fit’ note being issued to the patient.

METHOD

Data collection and processing

The study uses combined data from two separate evaluation projects commissioned and funded by the Department for Work and Pensions (DWP): the national evaluation of the fit note and the evaluation of Fit for Work Service (FFWS) pilots. The first evaluation recruited 49 general practices from five geographical areas of the UK (Scotland, Wales, Derbyshire, North West and South East England). The second project involved 19 practices sited in three FFWS pilot sites (Greater Manchester, Leicestershire and North Staffordshire). Although the evaluations had different objectives, both used a similar method of collecting fit note data from participating general practices. There were no significant differences between the fit note evaluation and FFWS practices in relation to size, location (urban/rural) and social deprivation of patient population. In both evaluations GPs in practices were requested to use ‘carbonised’ pads of fit notes for a period of 12 months. Using the specialised pads ensured that details of every fit note issued in the period could be retained on duplicate sheets. The FFWS evaluation practices started recording fit note data in July/August 2011 and the national fit note evaluation practices in November/December 2011. All fit note data collection at the 68 practices had been completed by January 2013.

In addition to the details on the note itself (date of issue, diagnosis, period to abstain from work, whether the patient ‘may be fit’ to do some work, whether the patient needed to be re-assessed at the expiry of the note and the certifying GP) a number of additional items were collected from the patient practice record. These included sex, year of birth, and postcode. The latter was transformed by practice staff into a neighbourhood deprivation score for the patient. Deprivation scores were based on lower-level Super Output Area and Data Zone scores in the most recent Indices of Multiple Deprivation (IMD) for England, Wales and Scotland. At the end of data collection, practice managers were requested to provide some basic information about their certifying GPs (sex, age, partner status, and whether classed as ‘full-time’). Routine information about the practice itself (list size, location) was also collected. Practices were assigned a deprivation status based on the proportion of their patients living in the most socially deprived areas of the country.

Statistical analysis

For estimating proportions of fit notes with ‘may be fit’ advice, types of advice and group proportions, simple percentages are reported.

Multivariate logistic regression models were constructed in order to estimate the independent effect of patient, GP- and practice-related factors on the likelihood of a patient receiving a ‘may be fit’ note from their GP. In order to account for the hierarchical nature of the data (patient/GP/practice), multilevel (random-intercept) models were deemed to be appropriate. For reporting of fixed effects, covariate estimates (odds ratio [OR], 95% confidence intervals [CIs] and P-value) are included in the tables of results. A conventional criterion of statistical significance (P<0.05) is assumed. For random effects, the variance and median OR (MOR) are reported as estimates of heterogeneity of outcome at each stage of model development. Data were analysed using SPSS for Windows 20 and Stata IC10.

RESULTS

Fit notes issued in practices

A total of 68 general practices were represented in the dataset containing merged fit note data from the two evaluation studies. All but two practices met the obligation to provide details of every fit note issued within a 12-month collection period. Twenty two practices (32.4%) were small (under 5000 registered patients) and 19 (27.9%) had a list size exceeding 10 000 patients. Twenty-three practices (33.8%) reported serving a largely rural catchment area. Nineteen practices (27.9%) were classified as having ‘very high deprivation’, with over 70% of their patients residing in the most deprived 40% of neighbourhoods in the respective country of residence (England, Scotland, or Wales). A total of 744 GPs (including partners, salaried GPs and locums) issued at least one fit note in the data collection period. Over 73% (n = 43) issued more than 20 fit notes in the 12-month period. Some basic information was available for 602 GPs. Over 51% (n = 307) were male, 26% (n = 157) were aged >50 years, 56% (n = 337) were partners in their practice and 51% (n = 307) worked full-time.

The 68 practices submitted details of 79 815 fit notes issued to 33 768 patients in the data collection period. The diagnostic category accounting for the highest proportion of certified sickness was the mild-moderate mental disorder category (that included common psychological causes of sickness absence such as ‘depression’, anxiety’ and ‘stress’). This category accounted for 35% of all fit notes in the dataset.

Fit notes with ‘may be fit’ advice

A total of 5080 fit notes, representing 6.4% of all notes, had the ‘may be fit’ box checked (7.7% of all notes issued in first month of data collection, falling to 6.1% in final month). Nearly 58% were issued to female patients, 31% to patients aged >50 years, and 22% to patients living in one of the 20% most deprived neighbourhoods in the country of residence.

The physical illness categories tended to have the highest proportions of ‘may be fit’ notes (Table 1). Over 10% of fit notes issued to a patient incapacitated due to an injury included ‘may be fit’ advice. Nearly 10% of notes issued to assist the patient recover from a recent surgical intervention and over 8% of musculoskeletal fit notes included this type of return to work advice. While fit notes issued for common mental health problems accounted for over a third of all notes issued, only 4% of these notes included ‘may be fit’ advice.

Table 1.

Fit notes with ‘may be fit for work’ advice

	Fit notes in category n	Fit notes with ‘may be fit’ advice n (%)
Injury (including fracture)	5504	563 (10.2)
Post-operative recovery	6737	653 (9.7)
Musculoskeletal	11 147	928 (8.3)
Circulatory	2234	179 (8.0)
Skin	770	61 (7.9)
Symptom	7506	541 (7.2)
Blood disorder	218	15 (6.9)
Pregnancy-related	1221	84 (6.9)
Nervous system/sense organ	1960	127 (6.5)
Cause of injury	475	30 (6.3)
Procedure/investigation/treatment	859	54 (6.3)
Cancer	1197	75 (6.3)
Viral illness	2226	113 (5.1)
Digestive	2050	99 (4.8)
Genitourinary	1060	51 (4.8)
Mild-moderate mental disorder	27 792	1219 (4.4)
Endocrine/nutrition/metabolic	622	27 (4.3)
Congenital	271	10 (3.7)
Severe mental disorder	788	29 (3.7)
Respiratory	4872	140 (2.9)
All	79 815a	5080 (6.4)

^aIncludes 306 fit notes not allocated to diagnostic category.

Inter-practice variation

Although the overall rate of ‘may be fit’ note completion was 6.4%, the proportions varied widely across the 68 practices (Figure 1). Two practices (on Merseyside and in North Staffordshire) only included ‘may be fit’ advice on approximately 1% of all their fit notes issued in a 12-month period. At the other end of the range, three Derbyshire practices included this type of advice on 13–15% of their fit notes.

Figure 1.

Proportion of all fit notes issued by study practices that included ‘may be fit ’ advice

Types of ‘may be fit’ advice

Of the four structured advice items on the fit note, ‘amended duties’ was most often indicated. This advice box was ticked on 2119 fit notes (42% of all ‘may be fit’ notes), either alone or along with other items of advice. Proportions of ‘may be fit’ notes indicating ‘phased return’, ‘altered hours’ and ‘workplace adaptations’ were 35%, 20%, and 9% respectively. The nature of advice was also associated with the health problem causing the sickness certification. Over 55% of ‘may be fit’ notes within the mild-moderate mental disorder category included advice relating to a ‘phased return to work’. However, for fit notes issued to patients with an injury or a musculoskeletal problem ‘may be fit’ advice tended to recommend the employee’s normal work duties be amended (61% of all ‘may be fit’ notes in these physical health categories). In terms of combinations of advice, a ‘may be fit’ note with the ‘amended duties’ box ticked, along with a free-text comment (usually specifying the duties the patient could do), was the most prevalent (Table 2). Nearly 23% of notes had this combination of advice.

Table 2.

Combinations of advice provided by GPs on ‘may be fit for work’ notes

Items of advice indicated on fit note	n	Column %
Amended duties/comment	1158	22.8
Phased return/comment	661	13.0
Comment only	561	11.0
Phased return only	536	10.6
Altered hours/comment	340	6.7
Amended duties only	330	6.5
Altered hours only	146	2.9
Phased return/amended duties/comment	127	2.5
Phased return/altered hours/comment	121	2.4
Workplace adaptation/comment	119	2.3
Amended duties/workplace adaptation/comment	90	1.8
Altered hours/amended duties/comment	86	1.7
Phased return/altered hours	75	1.4
Phased return/altered hours/amended duties/workplace adaptation	65	1.3
Phased return/altered hours/amended duties/comment	54	1.1
Phased return/amended duties	51	1.0
Workplace adaptation only	46	0.9
Altered hours/amended duties	41	0.8
Phased return/altered hours/amended duties	28	0.6
Amended duties/workplace adaptation	26	0.4
Phased return/altered hours/amended duties/workplace adaptation/comment	26	0.4
Other combination of advice	76	1.7
No advice given	317	6.2
Total	5080	100

A substantial number of the 5080 ‘may be fit’ notes completed by GPs did not comply with DWP guidance. Over 6% of them had the ‘may be fit’ box ticked, but did not indicate any of the four structured items or include any free-text advice to the patient and/or employer. Over 26% had one or more structured advice items checked, but included no free-text advice to clarify what changes should be made to facilitate a return to work. The GP should always include a period of certification on the fit note. If the employer cannot comply with GP advice, the sickness certificate then becomes a conventional ‘sick note’. However, on 1014 (19.9%) ‘may be fit’ notes no certification period was specified.

Independent predictors of ‘may be fit’ note issue

A multilevel logistic regression model was developed by entering patient-, GP- and general practice-related variables into the model at different stages. Model A is the null (empty) model, model B has independent patient factors only, model C has both patient and GP factors and model D adds general practice variables to complete the overall model (Table 3).

Table 3.

Independent effects on likelihood of patient receiving a ‘may be fit’ note

	Model A (Null)	Model B		Model C		MODEL D
		OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value
Level 1: Patient
Category of health problem
Respiratory problem		1.00		1.00		1.00
Mental health problem		1.86 (1.50 to 2.32)	<0.001	1.82 (1.38 to 2.44)	<0.001	1.80 (1.39 to 2.44)	<0.001
Musculoskeletal		3.27 (2.60 to 4.10)	<0.001	3.19 (2.56 to 4.16)	<0.001	3.17 (2.38 to 4.36)	<0.001
Injury		4.29 (3.38 to 5.43)	<0.001	4.19 (3.29 to 5.54)	<0.001	4.17 (3.10 to 5.71)	<0.001
Circulatory illness		3.58 (2.63 to 4.88)	<0.001	3.50 (2.55 to 5.02)	<0.001	3.45 (2.40 to 5.09)	<0.001
Post-operative recovery		3.30 (2.61 to 4.17)	<0.001	3.19 (2.53 to 4.30)	<0.001	3.15 (2.37 to 4.48)	<0.001
Other		2.20 (1.77 to 2.74)	<0.001	2.11 (1.67 to 2.85)	<0.001	2.04 (1.45 to 2.94)	<0.001
Female patient		1.11 (1.02 to 1.22)	0.020	1.10 (1.01 to 1.20)	0.030	1.09 (1.01 to 1.21)	0.032
Age (10 years older)		1.01 (0.98 to 1.02)	0.910	1.05 (0.93 to 1.08)	0.880	1.05 (0.92 to 1.09)	0.890
Living in one of 20% most deprived neighbourhoods in the country		0.83 (0.72 to 0.94)	0.006	0.85 (0.75 to 0.93)	0.005	0.86 (0.73 to 0.94)	0.011
Level 2: Certifying GP
Female GP				1.01 (0.88 to 1.15)	0.830	1.01 (0.80 to 1.24)	0.890
Age, years
<35				1.00		1.00
35–50				0.96 (0.81 to 1.14)	0.630	0.97 (0.81 to 1.15)	0.720
>50				0.92 (0.76 to 1.13)	0.440	0.93 (0.76 to 1.14)	0.490
Partner in practice				0.92 (0.79 to 1.07)	0.290	0.91 (0.78 to 1.07)	0.250
Works full-time				0.90 (0.78 to 1.03)	0.130	0.94 (0.74 to 1.08)	0.140
Level 3: Practice
List size, patients
<5000						1.00
5000–10 000						1.24 (0.82 to 1.85)	0.300
>10 000						1.23 (0.78 to 1.95)	0.380
Urban (versus rural) practice area						0.88 (0.61 to 1.27)	0.510
Practice ‘very deprived’						0.76 (0.50 to 1.15)	0.200
Random effects
Model Variance	0.42	0.37		0.36		0.34
Median OR	1.85	1.78		1.77		1.74

Values in bold denote statistically significant effects.

In the final model, GP factors (sex, age, being a partner in the practice and working full-time) and general practice variables (size of practice, location and the social deprivation of the patient population) had no statistically significant association with the likelihood of the patient receiving a fit note containing ‘may be fit’ advice. Patient-level factors, such as type of health problem causing sickness certification, gender of the patient and social deprivation, did have a significant effect on this outcome. Patient age was not associated with a higher likelihood of receiving a ‘may be fit’ note. Compared to patients in the respiratory reference category (which had the lowest rate of ‘may be fit’ advice on notes), those incapacitated due to an injury, musculoskeletal problem or circulatory illness were 3–4 times more likely to have received a ‘may be fit’ note. Female patients were 9% more likely to receive this type of note than were male counterparts. Patients living in the most deprived neighbourhoods of their respective country (England, Wales, or Scotland) were 14% less likely to receive a ‘may be fit’ note than were patients living in less deprived areas.

In terms of random effects, the variance in outcome decreased from 0.42 (MOR = 1.85) in the null (empty) model to 0.34 (MOR = 1.74) in the final model including all patient, GP, and practice factors.

DISCUSSION

Summary

Of 79 815 fit notes analysed, only 6.4% had the ‘may be fit’ box ticked by the certifying GP. This proportion varied widely across the 68 practices providing data. Of the four structured advice items on the fit note, a recommendation to the employer to amend normal work duties of the patient was the most popular (42% of ‘may be fit’ notes). Higher proportions of ‘may be fit’ notes were found within the physical health categories of patient health problem (such as musculoskeletal problems and injuries). The types of advice offered by the GP also differed depending on whether the patient health problem was physical or psychological. A substantial number of fit notes were not completed according to DWP guidance.^12–13 In multilevel modelling, GP and general practice factors had no significant impact on likelihood of a ‘may be fit’ note being issued to a patient. Patient-level factors, such as their main health problem, female sex, and living in less socially deprived areas, significantly increased the likelihood of a ‘may be fit’ note.

Strengths and limitations

The main strength of the study lies in its use of the largest sickness certification database yet compiled in the UK in order to evaluate the effectiveness of a recent radical change in sickness certification (the opportunity for GPs to advise the patient and employer that a return to work may be possible before the patient has fully recovered). Other evaluations in the previous 3 years have interviewed or surveyed interested parties to elicit opinion. This study investigates what is taking place in practice.

The study is limited in that there were a number of key explanatory variables that were not able to be measured and included in the study. For instance, it was not possible to discern whether the patient receiving a fit note was in paid employment or not. Only patient items routinely recorded in the practice record were available to supplement details on the fit note itself. Knowledge of the patient’s employment status may have informed interpretation of the association that was found between social deprivation and receiving a ‘may be fit’ note. The regression model itself had a large degree of specification error. Under 20% of variance in outcome was explained by all variables in the completed model. In particular, the large inter-practice variation was inadequately explained.

Comparison with existing literature

As the inclusion of the ‘may be fit’ section in the medical statement is a relatively recent development it is not possible to directly compare our findings with those from previous studies.

This study found that ‘may be fit’ advice was more common on fit notes issued to patients for physical health problems and that those with mental health problems were less likely to be assessed as fit for some work. Under the previous system of sickness certification, a number of UK studies reported that patients with common mental health problems were more likely to be certified as (totally) unfit for work.^14–17 In a 2009 trial of the revised medical statement (that was later to become the fit note), a number of scenarios based on three health conditions (depression, back pain, mixed depression and back pain) were presented to 583 GPs. It was found that 70% of back pain cases were assessed as ‘fit for some work’. Only 19% of depression cases were assessed as such.¹⁸

‘Partial sick-listing’ has been available to physicians in the Nordic countries (Sweden, Norway, Denmark, and Finland) for a number of years. The patient can return to work on a part-time basis, or work full-time with reduced work duties, and still receive partial sickness benefit in addition to a part of their salary. The rationale is very similar to that used to justify revisions in UK sickness certification in 2010; that is, performing some work is beneficial to health, and it is important to prevent the onset of long-term incapacity and potential exit from the workforce. A review of the limited number of studies that have looked at the use and effects of partial sick-listing in the four countries reported some findings similar to this current study. In particular, females were more likely to claim partial sickness benefits and musculoskeletal problems were more likely to result in physicians partially sick-listing the patient.¹⁹ In all four Nordic countries poor collaboration between the interested parties (employer, employee, physician, and insurance institution) has prevented optimum use of the partial sick leave programme.

Implications for practice

The study findings imply that the fit note might be used in a more effective manner in order to assist patients to return to work with the appropriate support. The prevalence of ‘may be fit’ advice declined over the course of the data collection period. One possible explanation is that some GPs, with an already heavy workload, decided that return to work discussion would only add to the length of the consultation. When ‘may be fit’ options are indicated, the study provides some evidence that GPs are not always providing sufficient relevant information relating to the patient’s capacity to work and the modifications in working conditions that would facilitate a return to normal employment. The introduction of the electronic fit note will improve legibility, but not necessarily the content of the advice to the patient and employer. ‘May be fit’ advice on fit notes is more likely when the reason for sickness absence is a physical health problem, such as an injury. However, over a third of all fit notes issued in the study were for common mental health problems (such as anxiety and depression), and there is a need to understand why ‘may be fit’ advice was largely felt to be inappropriate for patients incapacitated due to psychological morbidity.

Funding

Both evaluations providing data for the study were funded by the Department for Work and Pensions (DWP). Evaluation of FFWS: (RF1153), evaluation of Fit note: (RF1172).

Ethical approval

The construction of a fit note database was approved by the National Research Ethics Service (NRES) in August 2011 (REC ref: 11/NW/0339). Section 251 status was sought and obtained from the National Information Governance Board (NIGB) in order to enable non-identifiable data to be collected without individual patient consent.

Provenance

Freely submitted; externally peer reviewed.

Competing interests

The authors have declared no competing interests.

Discuss this article

Contribute and read comments about this article: www.bjgp.org/letters