The association of patient's family, leisure time, and work situation with sickness certification in primary care in Sweden

Abstract

Objective. To investigate associations between patients’ family, leisure time, and work-related factors and physicians’ measure as to whether or not to sickness certify the patient in connection with the consultation. Design. Questionnaire survey to physicians in general practice and their patients. Setting. General practitioners (GPs) and their patients in Örebro county, Sweden. Subjects. A total of 474 patient–physician consultations from 65 physicians with up to 10 patients each. Main outcome measure. Whether or not a sickness certificate was issued. Results. Among work-related factors, high “authority over decisions” and high “social support” correlated with 30% or more reduced sickness certification probability. Worrying about becoming ill or being injured from work correlates with almost doubled sickness certification risk. Among family and leisure-time variables, only living with a common law partner and having no children correlated with increased sickness certification risk. In addition to analyses of the whole group (all diagnoses), the two largest diagnostic subgroups, infectious diseases and musculoskeletal diseases, were examined. For the infectious diseases subgroup, high demands in work correlated with increased sickness certification risk, while in the musculoskeletal diseases subgroup, worry about work-related injury or illness was the main factor correlating with increased risk for sickness certification. Conclusions. Work-related factors were the most important factors related to sickness certification in this study. Determinants for sickness certification risk differed between diagnostic subgroups.

Key Words::

• Authority over decisions
• family medicine
• family practice
• sick leave
• social support
• work capacity
• work demands
• work strain

Many authors have discussed the relationship between social, socioeconomic, or psychological stress and sickness or sickness certification, given the notion that sickness is as much a social as a medical problem.

• High social support and high authority over decisions at work correlates with lower sickness certification probability.

• Patients' worry about work-related injury or illness correlates with higher sickness certification probability.

• The relations of social determinants to sickness certification risk differed somewhat for the two main diagnostic groups, infectious diseases and musculoskeletal diseases.

The social welfare system in Sweden is multifaceted, as in most Western countries. Unemployment, social distress over inability to earn one's living, and sickness causing reduced work capacity may entitle individuals to benefits from the general welfare system, of which the strictly regulated National Social Insurance is one part. According to the insurance scheme benefits are provided for income loss due to reduced work capacity owing to illness or injury only, but not due to social distress or unemployment [1]. Theoretically this is quite a clear distinction, but in general practice, it is not obvious where to draw the line [2–3]. Several scientific models may be applied when sick leave is analysed [4].

When discussing the term work capacity, the three dimensions physical, psychological, and social should all be taken into account [5]. Similar reasoning may be used when discussing sick leave and sickness certification. Many authors have discussed the relationship between social, socioeconomic or psychosocial stress, and sickness or sickness certification [6–13], given the notion that sickness is as much a social as a medical problem [14–15]. Important research shows a strong relationship between work environmental circumstances on the one hand and sickness and sick leave on the other [16–22]. In a previous report from this study, non-somatic consultations were related to increased sickness certification probability [23], as did a general practitioner (GP)–patient consensus on the patient's reduced work capacity due to illness, the GP's experience in family medicine, and work hours [24].

The aim of this report was to analyse the relationship of patients’ view on their family, leisure time, and work situation to the GPs’ measure to issue sickness certificates.

Material and methods

Setting

The study was performed in Örebro county (270 000 residents), central Sweden, in 1996. GPs in all 26 County Council operated and all 11 privately operated primary healthcare centres (PHCC) were invited to participate. Fourteen of the county council and two of the privately operated PHCCs took part in the study. Half of the PHCCs were located in municipalities with 20 000 or more residents, and the rest in smaller communities. The distribution of the non-participating PHCCs was similar in this respect. No other differences were expected. Since there was no individual identification, no detailed comparison of participants and non-participants was possible.

Study population and data collection

The study design was a cross-sectional questionnaire study of appointments between GPs and eligible patients, with details presented previously [23–24]. Briefly, the participating physicians were supposed to provide data on up to 10 consultations. Patients had to be 18–64 years old, not already on sick leave or retired, seeing the GP for whatever reason, and able to fill out a questionnaire in Swedish. Consecutive patients fulfilling the inclusion criteria were asked in a preset order about willingness to participate, and those who gave oral informed consent were included in the study. The GPs filled in one questionnaire about themselves and one questionnaire about each consultation (all questionnaires available at [23]). Reports from 474 consultations were obtained from 65 physicians. Information on sickness certificate issued or not and on diagnoses assessed was collected from physician questionnaires. All other information was obtained from the patient questionnaires.

In the patient questionnaire, which was given to them after the consultation, information on personal data, work, health, family and leisure-time situation, and consultation-related matters were sought. To capture social factors not covered by the questions posed, information could be given in a free format. The questionnaire was returned to the researchers in a closed envelope.

Statistical analysis

The analyses were performed with JMP 5.0 and SAS 6.12 software (SAS Institute, Cary, NC, USA). The proportion of missing values in returned questionnaires was less than 5%. A “leisure time score” with the range 0–10 was constructed as the sum of positive answers to questions on leisure-time activities. Scores on “demands”, “decision latitude” (divided into “intellectual discretion” and “authority over decisions”) and “social support” were constructed in accordance with the Karasek-Theorell model [16–17].

The analyses were performed in two steps. To find candidate determinants for the final analysis model, a set of screening bivariate analyses (Spearman correlation) of possible relations to outcome (issuing of a sickness certificate) of GP or patient responses (determinants) was performed. In the final model (multivariate logistic regression) the outcome was entered as dependent variable and all significant determinants from the screening analyses entered as independent variables. The procedure provides odds ratios (OR), confidence intervals (CI), Wald's chi-squared estimates (measure of variable impact), and receiver operator characteristics (ROC) based estimates of degree of explanation. All analyses were performed on the total study population, and separately for the two largest diagnostic groups (infectious and musculoskeletal diseases). All tests were two-tailed. In the screening analyses p < 0.10 and in the final model p < 0.05 was accepted as indicating statistical significance.

Results

Patient characteristics

There were no significant differences between patients sick certified or not regarding age, mother tongue, sex, educational level, employment status, or working full time (see Table I). Patients with musculoskeletal diseases were sick certified to a significantly greater extent than patients with other diagnoses.

Table I. Characteristics of patients sick certified or not sick certified in Örebro County primary healthcare.

	Sick certified		Not sick certified
	n	mean or %	n	mean or %	p
n	116		358
Age, years (SD)		42.7 (11.2)		41.0 (11.8)	n.s.
Women, %	76	65.5	229	64.0	n.s.
Swedish speaking, %	109	94.0	341	95.3	n.s.
Educational level, %					n.s.
Low	41	35.3	93	26.3
Medium	57	49.1	177	50.0
High	18	15.5	84	23.7
Employment status, %					n.s.
Permanently employed	82	70.7	244	68.2
Temporarily employed	18	15.5	49	13.7
Self-employed	7	6.0	21	5.9
Unemployed	8	6.9	30	8.4
Home worker	1	0.9	1	0.3
On parental leave	1	0.9	14	3.9
Working full time, %	83	74.1	254	73.0	n.s.
Largest diagnosis groups
Infectious diseases^1	37	30.2	136	38.0	n.s.
Musculoskeletal diseases^1	50	43.1	87	24.4	0.001

Note: ¹One person was reported for both infectious and musculoskeletal diseases and is described as both in this table.

Bivariate analyses

Patients’ life situation. Leisure-time activities, including the “leisure time score”, had no association with sickness certification probability (Table II). In those with infectious diseases the item “spending leisure time with others outside the family” and among those with musculoskeletal diseases the item “read, listen to music, watch TV/video, etc.” indicated significantly fewer sickness certifications (data for subgroups not shown). Regarding the family situation those living with parents had less sickness certification and those living with partner but with no children had more sickness certification.

Table II. Leisure-time activities and family situation among patients who were sick certified during the consultation versus patients who were not.

	Sick certified		Not sick certified
	n	%	n	%	p
Leisure-time activities
Exercise heavily for at least 20 min^1	49	44.1	170	50.6	0.24
Exercise at a quick pace for at least 20 min^1	60	56.1	189	56.3	0.97
Active outdoors at a calm pace^1	91	79.8	280	80.2	0.93
Spend time with my family/children^1	101	91.8	312	91.5	0.92
Spend time with people outside my family^1	85	75.8	285	81.0	0.25
Read, listen to music/TV/video, etc.^1	110	95.7	344	97.2	0.42
Hobby activities at home^1	51	44.7	184	52.6	0.15
Participate in group activities outside the home^2	35	31.3	123	35.0	0.46
Participate in clubs, church activities, etc.^2	31	27.7	119	34.6	0.18
Cinema, theatre, dancing, watching football, etc.^2	72	64.3	220	62.7	0.76
Total leisure time score (range 0 – 10)	116	(4.96)^3	358	(5.26)^3	0.24
Family situation
Live alone	16	14.3	55	15.4	0.72
Live with parent(s)	0	0	14	3.9	< 0.05
Partner, no children	41	36.6	100	28.1	0.09
Partner and children	49	43.8	174	48.9	0.34
Single parent	6	5.4	13	3.7	0.43

Note: ¹= Once a week or more often; ² = once a month or more often; ³ = mean value.

Patients’ work situation. Worry about work-related injury or illness, not getting on well with work, high job demands, low intellectual discretion in the job, low authority over job decisions, and low job social support all related to increasing the probability of sickness certification (see Table III). Within the job demand dimension, the items physical or mental exhaustion were positively associated with sickness certification. Within the intellectual discretion dimension the items “stimulating work” and “learning new things at work” were negatively and “monotonous work” positively associated with sickness certification. Social support except “good atmosphere at my workplace” was negatively associated with sickness certification.

Table III. Work-related variables among patients who were sick certified after the consultation and patients who were not.

	Sick certified		Not sick certified
	n	Score^3%	n	Score^3%	p
Worry about work-related injury or illness^1	61	57.0	127	38.0	< 0.001
Get on well at my work					< 0.01
Lowest quartile (= not very well)	41	36.3	74	21.4
Highest quartile (= well or fairly well)	23	20.4	92	26.6
Work score dimensions
Demands (low = 1; high = 5)	108	2.80	338	2.61	< 0.001
Work is done on time^2	95	88.0	297	88.1	0.96
Work often involves conflicting demands^2	17	16.2	39	12.0	0.27
Work results in physical exhaustion^2	43	39.8	83	24.8	< 0.01
Work results in mental exhaustion^2	44	40.7	93	27.8	< 0.05
Intellectual discretion (low = 1; high = 5)	108	3.42	339	3.62	0.001
Work is stimulating^2	64	60.4	240	71.2	< 0.05
Work is important to society^2	63	60.0	211	62.8	0.61
Work requires skill or expertise^2	86	81.1	266	78.9	0.63
Work requires ingenuity^2	69	63.9	228	67.7	0.47
Learn new things at work^2	39	36.4	158	46.7	0.06
Monotonous work^2	68	63.0	169	50.3	< 0.05
Authority over decisions (low = 1; high = 5)	108	3.16	339	3.45	< 0.001
Can influence work pace^2	38	35.5	161	47.6	< 0.05
Can decide what to do during the working day^2	38	35.5	157	46.7	< 0.05
Can influence how work is carried out^2	56	52.8	236	70.2	<0.001
Can alternate between easy and difficult tasks^2	40	37.0	158	47.3	0.06
Social support (low = 1; high = 5)	108	3.89	338	4.19	< 0.001
Can confer with co-workers on work tasks^2	78	72.9	281	83.4	< 0.05
Supported by supervisor/manager on work tasks.^2	55	51.4	234	70.2	< 0.001
Being together with co-workers is important^2	79	74.5	278	83.7	< 0.05
The atmosphere at my workplace is good^2	90	84.9	300	90.4	0.12
Work part-time	29	25.9	94	27.0	0.82
Worried about changes in work situation^1	71	65.7	207	61.8	0.46

Notes: ¹ = always, mostly or sometimes; ² = always or mostly; ³ = mean value.

In the subgroup with infectious diseases, some individual items were significantly related to the outcome but none of the work score dimensions (data not shown). In the subgroup with musculoskeletal diseases the situation was similar except that “worry about work-related injury or illness” was associated with more sickness certifications.

Multivariate analyses

When significant variables from the screening analyses were entered in the final multivariate analysis model, scores indicating high social support and authority over decisions correlated with lower frequency of sickness certifications for all diagnoses, while worry about work-related injuries or illness, and living with partner and having no children correlated with higher probability of sickness certification. For infectious diseases only a high demand score had a higher sickness certification probability, and spending time with people outside the family had a lower, and for musculoskeletal diseases worry about work-related injury or illness correlated to significantly increased risk of sickness certification.

The degrees of explanation of all variables in the models for all diseases, infectious, and musculoskeletal diseases were 67.6%, 68.9%, and 60.5%.

Effects of non-response

In 168 consultations, GP data but no patient data and in 47 consultations patient data but no GP data were obtained. The 215 subjects affected by these data losses were somewhat younger than other subjects, mean age 38.3 years versus 41.4 years, 55.4% versus 64.3% women, and the proportion of sickness certified 24.7% versus 24.5%. However, since neither patient age nor sex influenced outcome, the effect of non-response on the results seems to be negligible.

Discussion

Our main findings were that patients’ work situation variables correlated with sickness certification in several respects, while family and leisure-time circumstances had a rather low correlation with sickness certification.

The strengths of this study include the fact that data were collected from patients and GPs during real-life consultations when patients had not yet positioned themselves as able or unable to work. No prior power analysis was done, but a post hoc power analysis based on differences between those who were and were not sick certified, shown in Table IV, illustrated a statistical power of more than 90% with the actual sample size. The degree of explanation was based on the ROC method, superior to the usually used R² method, which severely underestimates the true degree of explanation. The high levels in this study therefore indicate the substantial impact of the determinants on the outcome.

Table IV. Final multivariate model of the relationships of patient's family, leisure time, and work situation to sickness certification after adjustment for the effects of patient age, sex, mother tongue, education, and professional status.

	All diseases			Infectious diseases			Musculoskeletal diseases
	OR	95% CI	χ^2	OR	95% CI	χ^2	OR	95% CI	χ^2
Social support score	0.60	0.44–0.83	9.64	0.69	0.39–1.20	1.74	0.94	0.54–1.64	0.05
Worry about work-related injury or illness	1.91	1.20–3.02	7.50	1.66	0.69–4.01	1.28	2.37	1.14–4.94	5.34
Authority over decision score	0.68	0.51–0.91	6.62	0.70	0.46–1.22	1.58	0.81	0.50–1.32	0.74
Partner, no children	1.68	1.04–2.72	4.54	1.37	0.56–3.32	0.48	1.77	0.80–3.91	1.99
Demand score	1.37	0.86–2.18	1.70	2.55	1.14–5.72	5.20	1.55	0.74–3.24	1.37
Spending time with people outside the family	0.77	0.44–1.35	0.84	0.38	0.16–0.90	4.88	1.17	0.46–2.96	0.11

Notes: OR = odds ratio, 95% CI = 95 % confidence interval, χ² = Wald's chi-squared estimate.

The limitations include too small a sample size for analyses of other diagnostic subgroups, such as mental disorders, and also that the data were collected more than a decade ago. However, the proportion of consultations concerning sick leave is today about the same as when our data were collected [25]. Another limitation may be that generalizations of our findings to situations where patients already have a long sick-leave record should be avoided, since iterative sickness certification determinants may differ from those influencing a first one. The reason why some GPs did not forward information on some of the responding patients is unknown. The most probable reason may be forgetfulness during hectic surgery hours.

An increased risk for patients with musculoskeletal diagnoses of being sick certified when they were worried about health risks at work was anticipated [26]. Fear of injury or illness may affect the description of symptoms and be interpreted by the doctor as more serious than they really are [27–28]. Patients’ as well as physicians’ fear-avoidance beliefs appear to influence long-term sick leave risk [29–30].

Sickness certification, inversely correlated to “authority over decisions” and “social support at work”, corresponds well with existing theories and findings about the effects of strain and control [16,18–22]. The increased risk of sickness certification in those with high demands at work, significant in infectious diseases, was expected. For subjects with functional impairment, apparent in febrile conditions, high work demands may be a major obstacle to performing work tasks.

For infectious diseases we found less frequent sickness certification related to “spending time with people outside the family”. One interpretation might be that subjects with a good social network appreciate the personal relationships formed at their workplace so much that they go to work despite symptoms of illness.

Our study adds another dimension in this field by indicating that social impact on decreased work ability also relates to type of health problem that brought the patient to the consultation. The GP should be aware of these social aspects, even though, according to regulations, they should not be considered when sickness certification decisions are made. Although information on patients’ circumstances reflects their own view and is not objective descriptions, the GP's knowledge of these matters is highly dependent on what the patients report, as objective information is seldom available at the time of the consultation.

Conclusions

Patients’ work-related and to some extent family-related factors were important for GPs’ measure of sickness certification. The impact of these factors was somewhat different depending on type of health problem, as exemplified in this study by musculoskeletal and infectious disorders.

Ethical approval

The Örebro Research Ethics Committee approved the study.

Acknowledgements

This study was supported by grants from the National Health Insurance Office, Örebro, the Family Medicine Research Centre, Örebro University, and Uppsala University.

Competing interests: The authors declare that they have no competing interests.