Change in quality of life in relation to returning to work after acquired brain injury: a population-based register study

ABSTRACT

Purpose: This study investigated changes in quality of life (QoL) in relation to return to work among patients with acquired brain injury (ABI).

Method: The sample consisted of 1487 patients with ABI (63% men) aged 18–66 years (mean age 52) from the WebRehab Sweden national quality register database. Only patients who worked at least 50% at admission to hospital and were on full sick leave at discharge from hospital were included. QoL was measured by the EuroQol EQ-5D questionnaire.

Results: Patients who returned to work perceived a larger improvement in QoL from discharge to follow-up one year after injury compared to patients who had not returned to work. This difference remained after adjustment for other factors associated with improved QoL, such as having a university education, increased Extended Glasgow Outcome Scale scores and getting one’s driving licence reinstated.

Conclusion: Return to work is an important factor for change in QoL among patients with ABI, even after adjusting for other factors related to QoL. This is consistent with the hypothesis that having employment is meaningful, increases self-esteem and fosters participation in society. Thus, helping patients with ABI return to work has a positive influence on QoL.

KEYWORDS:

• Stroke
• traumatic brain injury
• vocational rehabilitation
• rehabilitation
• life satisfaction

Introduction

In Western countries, it is estimated that about 100 to 300 per 100,000 adults of working age live with an acquired brain injury (ABI) (1). In Sweden, approximately 5000 adults of working age acquire a brain injury every year, giving an incidence of about 90 per 100,000 people between 18 and 65 years old (2). Acquired brain injury is an umbrella term that includes brain damage from cerebrovascular accidents, infections, toxins, tumours and traumatic brain injury (3) and it can affect all aspects of a patient’s life (3). Physical impairment (3) as well as reduced cognitive function such as fatigue, memory and concentration problems are common, affecting the patient’s functioning in everyday life (4,5). With reduced cognitive function, it becomes harder to manage social life, living independently (6,7), getting a job or returning to work (8–14).

Overall, people acquiring a brain injury experience reduced mental health (10,15) and quality of life (QoL) (16,17). Studies show that patients’ perceived QoL is strongly related to the severity of the brain injury and the degree of injury (18–21), but even a mild ABI can negatively affect a patient’s perceived QoL to a great extent (22). Factors such as preserved cognitive ability (12), physical functioning (8), motor independence (9), ability to walk (12) and absence of depression (8,9,11) have an impact on QoL. Furthermore, higher QoL has been found among patients with ABI who avoided comparing life with how it was before the injury, did not rate health status as very important, placed increased value on the family (23), had an optimistic life orientation, good social support and task-oriented coping styles (7).

One of the most important factors for improving QoL is being able to return to work (8–13). Studies show that people who had returned to work rated their QoL higher than those who had not returned to work (8,9,11,12). These studies measured QoL 1–5 years after ABI and found associations between a healthy and productive lifestyle and QoL (8,9,11). However, previous research has seldom measured the influences of RTW on QoL using change scores, despite the fact that rehabilitation is a process over time (24), thus change is the ultimate aim of rehabilitation. Usually, studies have a cross-sectional design and correlate QoL with other variables at a certain point in time. Thus, measuring change scores can reveal factors important for QoL changes. Since brain injury rehabilitation is about changing the condition of the patient, studying the relationship between factors over time should be of central importance. To the best of our knowledge there are no such studies exploring the change in QoL from hospital discharge to follow-up (e.g. one year later) for patients with ABI and what factors could either hinder or facilitate QoL change. Therefore, the aim of this study was to investigate the change in QoL in relation to return to work among patients with ABI.

Method

In Sweden, national quality registers with detailed clinical data are well established (25). These registers provide an important source for population-based research. One such register, the WebRehab Sweden, is directed at brain injury rehabilitation (26). The register was established in 1997 and includes all 21 county councils in Sweden that have at least one brain injury rehabilitation unit. The register covers about 75% of the existing rehabilitation clinics in Sweden. It has a steering group that has produced a manual for the registrations, makes a yearly report about the data and discusses issues about the instruments. Each rehabilitation unit that participates in the register also has a local coordinator (26). This helps to increase the reliability for the data in this study. Patients are assessed at three time points: (I) at admission to the hospital, (II) at discharge from the hospital and (III) at follow-up one year after injury (26). Importantly for the present study, the register measures, among other variables, return to work and QoL. Thus, WebRehab Sweden is well suited to exploring the relationship between returning to work, and change in QoL among people with ABI.

Ethical approval was given by the Regional Ethical Review Board in Uppsala, Sweden (ref. 2016/055). After application, the responsible administrator for WebRehab Sweden agreed to extraction of data and a statistician from the register assembled the data files and sent them to the researchers.

Patient inclusion in the study

The study comprises patients registered in WebRehab Sweden between 1 January 2007 and 15 January 2016. Based on the research question, the following inclusion criteria were used in the study: 1) having an acquired brain injury (i.e. stroke, subarachnoid haemorrhage, traumatic brain injury, postinfectious or postinflammatory brain injury, anoxic brain injury or other brain injury); 2) having only one registered injury and hospitalisation period; 3) being of working age (18–66 years); 4) working 50% or more at admission (employed or self-employed); 5) not working at all at discharge (i.e. on full sick leave) and 6) having QoL data registered both at discharge and at follow-up. Based on these inclusion criteria there were 1487 patients eligible for inclusion in the study, as shown in the flow diagram in Figure 1.

Figure 1. Flow diagram of patient inclusion in the study. (Abbreviations: QoL, quality of life; EQ VAS, EuroQol five-dimension questionnaire visual analog scale; EQ-5D, EuroQol five-dimension questionnaire.).

Measures

The EuroQol five dimensions questionnaire (EQ-5D) (27) was used to measure QoL (28). EQ-5D is a patient-reported outcome measure. It is a standardised instrument measuring status in five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression). Each dimension has three levels of severity (16,27). These five dimensions are converted into a preference-based index value which can vary between −0.594 and + 1.000 (29). The value 1 corresponds to complete health and 0 corresponds to being dead. Notably, an index value below 0 is fully possible, because the respondents compare themselves with a population-based profile and some patients may rate their QoL as worse than being dead. The EQ-5D questionnaire also includes a visual analogue scale, EQ VAS (27), which is a rating scale from 0 (worst imaginable health) to 100 (best imaginable health) (30). Respondents mark a point along a 100 mm line to indicate their estimated overall health at that moment. The EQ-5D questionnaire has been shown to be a valid measure of QoL after stroke (31) and the EQ-5D index has been validated against populations in several countries.

The Extended Glasgow Outcome Scale (GOSE) (32) is one of the most frequently used instruments to assess outcome after head injury; it is a global scale for functional outcome and describes overall function relative to before the injury (33,34). The scale is in 8 categories ranging from (I) dead, (II) vegetative state, (III) lower severe disability, (IV) upper severe disability, (V) lower moderate disability, (VI) upper moderate disability, (VII) lower good recovery and (VIII) upper good recovery. Prior to analysis, GOSE was recoded into three categories on the basis of patients’ change in GOSE from discharge to follow-up: a decrease in GOSE, no change in GOSE and an increase in GOSE.

The Functional Independence Measure (FIM) involves 18 activities, grouped into six functional areas: (I) self-care, (II) sphincter control, (III) mobility, (IV) locomotion, (V) communication and (VI) social cognition (35,36). FIM is an outcome measure rated from patient observation, often by consensus opinion in a multidisciplinary team (35). The FIM score for each functional area ranges from 1 (Total assistance), the lowest possible score, to 7 (Complete independence), the best possible score. In the present study, the levels 1–7 were rescored into three categories: total assistance, some help needed and no help needed. Motor function can be measured by combining areas I to IV (13 activities), with a score from 13–91 points, and cognitive function by combining areas V and VI (5 activities), with a score from 5–35 points. It was specifically designed to show how much help the patient needs and his or her ability to function independently, and it can measure changes among patients with different disorders over time. FIM was not measured at follow-up in WebRehab Sweden because the whole hospital team did not meet the patient at follow-up.

Working at follow-up was defined as being employed or self-employed 50% or more. This criterion for working has been used in previous studies (13,14).

Prior to analysis, some variables in the WebRehab Sweden register that were a mix of two concepts were rescored into new variables to ease the analysis, as outlined below; in other cases, categories within a variable were combined.

Diagnosis The Diagnosis variable has seven categories and was rescored into three categories: stroke (stroke and subarachnoid haemorrhage), traumatic brain injury (traumatic brain injury) and other brain injuries (postinfectious or postinflammatory brain injury, anoxic brain injury and other brain injuries).

Length of stay in hospital Has a range from admission to discharge between 0 and 357 days and was divided into following categories; short (0–24), moderate (25–68) and long (69–357).

Marital status and having children at home In the register, the Household variable has seven categories. Since this variable combined being married with having children, it was rescored into two new variables: Being single (Yes/No) and Having children at home (Yes/No), treating the ‘other’ and ‘don’t know’ alternatives as missing data. The being single variable and having children at home variables were recoded into four categories, respectively, based on the change from discharge to follow-up.

Accommodation and personal support The independent living variable consisted of four categories that mixed type of accommodation (independent or special) and personal support needs (with support or no support). This variable was rescored into a Personal support variable, disregarding type of accommodation and taking into account only whether the person was living with support or no support. The category no support was applied to those living independently without any help in their everyday life despite their brain injury. The accommodation and the personal support variables were recoded into four categories, respectively, based on the change from discharge to follow-up.

Driver’s license Having a driver’s license was scored as Yes or No. Driver’s license was rescored into four recoded categories based on the change from discharge to follow-up.

Statistical analysis

All analysis was conducted using IBM SPSS Statistics version 22.0 (IBM Corp., Armonk, NY, USA). Before the analysis, all categorical and ordinal variables were dummy coded (0/1). Since the aim of this study was to investigate the QoL change in connection with returning to work and other relevant factors, change scores (from discharge to follow-up) were computed for all dependent and independent variables except age, gender, diagnosis, country of birth, education, aphasia/dysphasia, GOSE and FIM. Length of hospital stay was measured as days between admission and discharge.

To investigate influences of return to work and other variables on QoL, we first performed univariate linear regression analysis with the EQ VAS and EQ-5D index variable regressed on each of the independent variables. Secondly, to adjust for influences of concomitant variables we performed multiple regression analyses with EQ VAS and EQ-5D index on return to work and each of the other independent variables found to be significant in the univariate regression analyses. Because of missing data, it was not possible to adjust for influences of more than one concomitant variable at a time. Therefore only one independent variable could be adjusted for at the time in the multiple regression analysis (37). A p < 0.05 was regarded as statistically significant (37).

Results

Characteristics of the study sample are shown in Table 1. About 60 % of the patients were men and the mean age was 52 years, slightly skewed towards somewhat older patients. There was no significant difference in age between men and women. Most of the patients were born in Sweden (90%) and spoke Swedish (97%). The majority had finished upper secondary school or higher education (89%). More women (42%) than men (32%) had completed university studies (χ² = 9.91, p = 0.002). Most of the patients had a diagnosis of stroke approximately evenly distributed among women (77%) and men (75%). However, a higher proportion of men (17%) than women (10%) were diagnosed with traumatic brain injury (χ² = 13.39, p < 0.001), and more women (13%) than men (8%) were diagnosed with other brain injuries (χ² = 9.03, p = 0.003). The patients’ hospital stay ranged from 0 to 357 days, with a mean of 51.20 days and a standard deviation (SD) of 40.11. Functional outcome, measured by GOSE, showed that, at discharge, a more than half of the registered patients (67%) had a moderate disability (> 5 on a 1–8 scale) (32). At discharge, they had a mean FIM motor function score of 80.41 (SD 14.94) on a scale of 13–91. Their mean FIM cognitive function score at discharge was 30.02 (SD 5.30) on a scale of 5–35.

Table 1. Characteristics, mean and standard deviation (SD) between EQ VAS and EQ-5D Index at discharge and follow-up.

		EQ VAS		EQ-5D index
Variable	N (%)	Discharge Mean (SD)	Follow-up Mean (SD)	Discharge Mean (SD)	Follow-up Mean (SD)
Returned to work	1487
No	917 (62%)	62.79 (20.20)	62.30 (20.65)	0.58 (0.30)	0.58 (0.32)
Yes	570 (38%)	72.51 (18.26)	77.55 (17.31)	0.73 (0.25)	0.78 (0.25)
Age (at admission)	1487
18–30	78 (5%)	72.59 (18.44)	74.80 (18.45)	0.73 (0.26)	0.70 (0.28)
31–40	135 (9%)	65.27 (20.06)	69.12 (19.78)	0.68 (0.24)	0.72 (0.26)
41–50	312 (21%)	67.08 (20.95)	69.81 (19.93)	0.64 (0.30)	0.67 (0.32)
51–60	648 (44%)	65.82 (19.75)	67.15 (21.22)	0.62 (0.30)	0.63 (0.32)
61–66	314 (21%)	66.38 (19.93)	67.03 (21.40)	0.64 (0.31)	0.65 (0.32)
Gender	1487
Men	924 (62%)	67.51 (19.62)	68.92 (20.93)	0.65 (0.29)	0.66 (0.32)
Women	563 (38%)	64.88 (20.63)	67.19 (20.54)	0.62 (0.30)	0.64 (0.30)
Country of birth	915
Sweden	824 (90%)	68.13 (18.55)	69.90 (19.86)	0.66 (0.27)	0.68 (0.30)
Nordic countries (except Sweden)	25 (3%)	68.79 (20.74)	65.92 (20.48)	0.67 (0.25)	0.66 (0.25)
European countries (except Nordic)	31 (3%)	63.14 (21.91)	65.47 (21.29)	0.56 (0.30)	0.56 (0.28)
Countries outside Europe	35 (4%)	57.29 (20.73)	64.52 (21.84)	0.53 (0.35)	0.53 (0.42)
Education level (years in education)	868
Compulsory school (9)	94 (11%)	66.49 (17.27)	65.51 (19.97)	0.60 (0.30)	0.61 (0.31)
Upper secondary school (≥12)	466 (54%)	69.35 (18.39)	68.93 (20.67)	0.66 (0.27)	0.67 (0.30)
University	308 (35%)	65.16 (20.04)	71.53 (19.09)	0.65 (0.28)	0.70 (0.31)
Diagnosis	1487
Stroke (including SAH)	1127 (76%)	66.04 (19.96)	67.65 (20.91)	0.63 (0.29)	0.65 (0.31)
Traumatic brain injury	214 (14%)	68.92 (20.51)	71.69 (19.84)	0.66 (0.31)	0.67 (0.31)
Other kind of brain injury	146 (10%)	66.52 (19.84)	67.97 (20.88)	0.64 (0.29)	0.68 (0.32)
Length of hospital stay	1485
Short (0–24)	379 (25%)	72.97 (17.07)	75.40 (18.04)	0.75 (0.22)	0.76 (0.23)
Moderate (25–68)	738 (50%)	66.55 (19.34)	68.72 (20.13)	0.65 (0.28)	0.67 (0.30)
Long (69–365)	368 (25%)	59.85 (22.02)	59.76 (21.86)	0.51 (0.33)	0.51 (0.36)
Aphasia/dysphasia (at discharge)	957
No	766 (80%)	65.88 (20.67)	67.63 (21.01)	0.63 (0.30)	0.64 (0.33)
Yes	191 (20%)	63.34 (19.98)	65.86 (21.80)	0.63 (0.27)	0.62 (0.32)
FIM (at discharge) Motor	1218
Total assistance 1–2.99	41 (3%)	49.66 (17.88)	46.63 (24.11)	0.07 (0.32)	0.09 (0.38)
Some help needed 3–5.99	266 (23%)	57.70 (20.75)	57.96 (22.69)	0.40 (0.29)	0.44 (0.35)
No help needed 6–7	911 (74%)	68.88 (19.26)	71.18 (18.86)	0.72 (0.23)	0.73 (0.25)
Cognition	1218
Total assistance 1–2.99	20 (2%)	59.00 (19.32)	60.14 (30.71)	0.28 (0.43)	0.40 (0.41)
Some help needed 3–5.99	394 (32%)	62.61 (21.37)	62.35 (22.68)	0.53 (0.35)	0.52 (0.38)
No help needed 6–7	804 (66%)	67.52 (19.57)	70.25 (19.33)	0.68 (0.25)	0.71 (0.27)
GOSE (at discharge)	1030
Very severe disability 1	–	–	–	–	–
Very severe disability 2	–	–	–	–	–
Severe disability 3	48 (5%)	52.73 (24.03)	52.74 (24.04)	0.22 (0.37)	0.26 (0.40)
Severe disability 4	119 (11%)	59.57 (22.53)	60.40 (21.90)	0.47 (0.34)	0.48 (0.36)
Moderate disability 5	171 (17%)	63.82 (21.97)	66.17 (20.93)	0.63 (0.30)	0.64 (0.32)
Moderate disability 6	467 (45%)	68.05 (18.87)	69.98 (19.71)	0.69 (0.24)	0.71 (0.26)
Good recovery 7	209 (20%)	70.80 (18.73)	74.74 (19.53)	0.75 (0.24)	0.76 (0.24)
Good recovery 8	16 (2%)	73.19 (22.12)	78.21 (9.53)	0.75 (0.19)	0.77 (0.17)
Change marital status	1394
Single–Single	398 (28%)	65.54 (21.00)	67.28 (21.61)	0.64 (0.27)	0.67 (0.29)
Single–Married/Partner	60 (5%)	70.00 (18.84)	72.63 (20.07)	0.68 (0.28)	0.69 (0.30)
Married/Partner–Single	59 (4%)	69.89 (18.08)	67.28 (23.75)	0.66 (0.29)	0.64 (0.33)
Married/Partner–Married/Partner	877 (63%)	66.28 (19.61)	68.53 (20.08)	0.64 (0.30)	0.65 (0.31)
Change children living in household	1394
No children–No children	898 (64%)	65.73 (20.09)	67.40 (21.19)	0.63 (0.29)	0.65 (0.31)
No children–Children	69 (5%)	66.01 (21.73)	68.01 (22.21)	0.66 (0.30)	0.68 (0.32)
Children–No children	70 (5%)	65.92 (20.42)	67.04 (21.04)	0.66 (0.30)	0.68 (0.28)
Children–Children	357 (26%)	68.17 (19.04)	70.91 (18.85)	0.65 (0.27)	0.68 (0.31)
Change personal support	179
No support–No Support	160 (11%)	70.93 (18.02)	73.13 18.24	0.72 (0.24)	0.74 (0.23)
No support–Support	9 (1%)	43.13 (18.70)	53.33 21.79	0.42 (0.29)	0.57 (0.23)
Support–No support	8 (1%)	74.25 (6.48)	70.50 20.68	0.67 (0.23)	0.62 (0.34)
Support–Support	2 (0.1%)	57.00 (0.00)	52.50 10.61	0.49 (0.28)	0.49 (0.28)
Change driving licence (DL)	1276
No DL–No DL	585 (46%)	62.99 (20.93)	62.82 (21.91)	0.57 (0.32)	0.56 (0.35)
No DL–DL	590 (46%)	70.91 (17.76)	74.88 (17.09)	0.72 (0.24)	0.77 (0.23)
DL–No DL	19 (1%)	63.63 (23.83)	50.26 (24.63)	0.68 (0.24)	0.48 (0.39)
DL–DL	82 (7%)	69.63 (18.83)	72.06 (20.61)	0.73 (0.24)	0.74 (0.23)

(Abbreviations: EQ-5D, EuroQol five-dimension questionnaire; EQ VAS, EQ-5D visual analog scale.)

The univariate linear regression analyses were performed on all variables, and the results are given in Table 2. As hypothesised, patients returning to work (n = 570) showed a greater improvement in QoL (EQ VAS and EQ-5D index) from discharge to follow-up compared to those who did not return to work (n = 917). Similarly, the analyses showed that an increased GOSE score from discharge to follow-up was related to increased QoL scores (EQ-5D index). Moreover, the analyses showed that patients with a university degree showed a greater improvement in QoL (EQ VAS) than patients who had only completed compulsory school (9 years).

Table 2. Mean and SD in EQ VAS and EQ-5D index change (change between discharge and follow-up). Univariate regression coefficients B;β in QoL change (change between discharge and follow-up) in EQ VAS and EQ-5D index.

	EQ VAS change (discharge to follow-up) (n = 1360)		EQ-5D index change (discharge to follow-up) (n = 1486)
Variable	Mean (SD)	B; β	Mean (SD)	B; β
Returned to work
No	−0.67 (22.58)	REF	−0.003 (0.31)	REF
Yes	4.93 (20.30)	5.60; 0.13***	0.05 (0.26)	0.05; 0.09***
Age (at admission)
18–30	2.37 (18.48)	REF	−0.03 (0.25)	REF
31–40	3.61 (22.94)	1.23; 0.02	0.03 (0.26)	0.07; 0.07
41–50	2.48 (22.78)	0.10; 0.002	0.03 (0.29)	0.06; 0.09
51–60	1.01 (21.67)	–1.37; −0.03	0.02 (0.29)	0.05; 0.09
61–66	0.45 (21.76)	–1.93; 0.04	0.003 (0.30)	0.04; 0.05
Gender
Men	1.25 (21.49)	REF	0.01 (0.29)	REF
Women	1.94 (22.53)	0.70; 0.02	0.02 (0.29)	0,01; 0.02
Country of birth
Sweden	1.82 (19.87)	REF	0.03 (0.28)	REF
Nordic country	-3.46 (18.76)	−5.28; −0.04	−0.01 (0.20)	−0.03; −0.02
European country	2.79 (23.88)	0.96; 0.01	0.004 (0.30)	–0.02; −0.02
Non-European country	6.75 (25.85)	4.93; 0.05	0.001 (0.31)	–0.03; −0.02
Education
Compulsory school (9)	−0.69 (18.99)	REF	0.01 (0.32)	REF
Upper secondary school (≥12)	−0.42 (20.00)	0.27; 0.01	0.01 (0.28)	0.002; 0.004
University	6.37 (20.22)	7.06; 0.17**	0.05 (0.27)	0.04; 0.07
Diagnosis
Stroke (including SAH)	1.21 (21.81)	REF	0.02 (0.28)	REF
Traumatic brain injury	2.63 (23.18)	1.41; 0.02	0.01 (0.33)	-0.01; −0.01
Other kind of brain injury	2.15 (20.52)	0.94; 0.01	0.04 (0.28)	0.02; 0.02
Length of hospital stay
Short (0–24)	2.19 (19.35)	REF	0.01 (0.26)	REF
Moderate (25–68)	1.74 (22.31)	−0.45; −0.01	0.02 (0.30)	0.01; 0.02
Long (68–357)	0.27 (23.52)	–1.92; −0.04	0.004 (0.30)	–0.01; −0.02
Aphasia/dysphasia (at discharge)
No	1.53 (22.08)	REF	0.01 (0.29)	REF
Yes	1.27 (26.15)	1.22; −0.05	−0.01 (0.32)	0.01; 0.03
FIM (at discharge) Motor (13–91)
Total assistance (1.00–2.99)	−5.00 (22.08)	REF	0.02 (0.43)	REF
Some help needed (3.00–5.99)	0.02 (25.50)	5.02; 0.09	0.04 (0.35)	0.02; 0.03
No help needed (6.00–7.00)	2.19 (21.27)	7.19; 0.14	0.01 (0.27)	–0.01; −0.02
Cognition (5–35)
Total assistance (1.00–2.99)	2.64 (29.17)	REF	0.12 (0.41)	REF
Some help needed (3.00–5.99)	−1.14 (23.64)	−3.78; −0.08	−0.01 (0.32)	−0.13; −0.21
No help needed (6.00–7.00)	2.74 (21.46)	0.11; 0.002	0.03 (0.28)	–0.09; −0.14
Change in GOSE
No Change in GOSE	1.41 (23.60)	REF	0.005 (0.29)	REF
Decreased GOSE	−1.48 (24.22)	−2.89; −0.04	−0.15 (0.31)	−0.16; −0.18***
Increased GOSE	5.09 (22.49)	3.68; 0.08	0.07 (0.28)	0.07; 0.11*
Change in marital status
Single–Single	1.77 (22.96)	REF	0.03 (0.29)	REF
Single–Married/Partner	2.81 (23.89)	1.04; 0.01	0.01 (0.35)	−0.02; −0.02
Married/Partner–Single	−2.84 (25.14)	–4.61; −0.04	−0.02 (0.29)	–0.05; −0.04
Married/Partner–Married/Partner	1.88 (21.07)	0.11; 0.002	0.02 (0.29)	–0.01; −0.02
Change in children living in household
No children–No children	1.29 (22.17)	REF	0.02 (0.29)	REF
No children–Children	2.08 (21.24)	0.78; 0.01	0.02 (0.25)	0.01; 0.01
Children–No children	1.55 (27.57)	0.25; 0.003	0.02 (0.32)	0.01; 0.01
Children–Children	2.62 (20.21)	1.33; 0.03	0.03 (0.28)	0.01; 0.02
Change in personal support
No support–No support	1.92 (16.77)	REF	0.02 (0.24)	REF
No support–Support	9.38 (17.82)	7.45; 0.09	−0.15 (0.27)	0.13; 0.12
Support–No support	−3.75 (23.21)	–5.67; −0.071	−0.04 (0.33)	–0.06; −0.05
Support–Support	3.00 (0.00)	0.08; 0.01	0.00 (0.00)	–0.02; −0.01
Change in driving licence (DL)
No DL–No DL	−0.39 (23.80)	REF	−0.01 (0.31)	REF
No DL–DL	3.56 (20.01)	3.95; 0.09**	0.05 (0.26)	0.06; 0.11***
DL–No DL	−13.37 (19.37)	−12.98; −0.08*	−0.20 (0.31)	−0.19; −0.08**
DL–DL	1.85 (17.64)	2.24; 0.03	0.02 (0.23)	0.03; 0.02

* p < 0.05, ** p < 0.01, and *** p < 0.001. (Abbreviations: EQ-5D, EuroQol five-dimension questionnaire; EQ VAS, EQ-5D visual analog scale; N/A = not applicable.)

The analysis of the influence of change in possession of a driving licence on QoL showed that, compared to those who did not have licence at discharge nor at follow-up, patients whose licence was suspended at discharge and reinstated at follow-up reported improved QoL (EQ VAS and EQ-5D index), whereas those patients who had a driving licence at discharge but later had to surrender it reported decreased QoL (EQ VAS and EQ-5D index).

None of the demographic variables (age, gender, country of birth, marital status or having children in the household) or sickness- or disability-related variables (type of diagnosis, length of stay in hospital and aphasia) showed any significant influence on QoL change. Finally, neither FIM motor function or cognitive function showed any significant association with QoL change.

Multiple regression analysis

Finally, since there were many variables significantly related to QoL in addition to return to work, multiple regression analyses were performed to investigate the relationship between return to work and change in QoL when influences from the other variables were taken into account. However, because of missing data in some of the variables, only one independent variable at time could be adjusted for. The results from the multiple regression analyses showed no difference in regard to the relationship between return to work and change in QoL (EQ VAS and the EQ-5D index), as found in the univariate analysis.

Discussion

The aim of the present study was to investigate the change in QoL in relation to return to work among patients with ABI, using data from the WebRehab Sweden quality register. As hypothesised, the present study showed that returning to work was one of the most important factors related to QoL change in people with ABI, even when adjusting for the influence of other variables. In addition to return to work, this study showed that education level, GOSE score and possession of a driving licence were related to a significant change in patients’ QoL. As shown in Figure 2 (EQ VAS) and Figure 3 (EQ-5D index), patients’ QoL improved from admission to discharge, but only patients who returned to work showed a further improvement in QoL at the one-year follow-up.

Figure 2. Boxplot with median and quartiles for EQ-VAS at admission, discharge and follow-up for patients who had returned to work (RTW) and who had not returned to work (not-RTW) at follow-up. (Abbreviations: EQ VAS, EuroQol five-dimension questionnaire visual analog scale.).

Figure 3. Boxplot with median and quartiles for EQ-5D index at admission, discharge and follow-up for patients who had returned to work (RTW) and who had not returned to work (not-RTW) at follow-up. (Abbreviations: EQ-5D, EuroQol five-dimension questionnaire.).

Given the effect size, the variable that had the most impact on QoL was functional ability measured by GOSE, which is in keeping with findings from other studies (20). This result is consistent with a study that asked patients who had sustained an ABI within the past four years to rate their disability in Disability Rating Scale. Those patients who rated their disability as low found community integration easier and also rated their QoL higher than those who rated their disability as high (38).

Patients who had a university degree showed an improved QoL from discharge to follow-up. To the best of our knowledge, no other studies with patients with ABI have investigated the relationship between education level and QoL. In addition, a study of 11,175 individuals from South Korea examined a combination of education level and perceived social class on self-rated health and life satisfaction (39). The study found that a combination of education level and perceived social class was associated with life satisfaction and self-rated health. Thus, it can be hypothesised that also among patients with ABI, a higher educational level is related to better financial resources and lager social network which may impact QoL. Further research on the relationship between education level and QoL among patients with ABI is warranted.

Patients who had surrendered their driving licence at the time of discharge but had it reinstated at follow-up reported improved QoL. This is consistent with a US study showing that driving a car is associated with independent living and community integration among people who had suffered a traumatic brain injury (40). An interview study about relevant factors for a productive lifestyle after ABI found that driving was one of the main goals for regaining an independent life and also a contributor for empowerment (41). A driving licence is equally important for community integration and participation.

A number of variables hypothesised to be associated with QoL, such as age and gender, showed no statistical significance in the analyses. Previous studies in Sweden on the general population found that QoL, measured by SF-36, decreased with age, particularly for the physical functioning aspect of QoL (42). One reason for this discrepancy could be the use of different scales, SF-36 and EQ-5D. Perhaps more plausible, however, is that sustaining an ABI profoundly affects a person’s life and that any influence of age becomes minor in comparison. In another study on patients with transient ischemic attack, which has symptoms similar to stroke, women had lower QoL than men in all areas that were measured with the Nottingham Health Profile (work, housekeeping, family, sex, leisure time and vacation) except social life (43). Since we measured change in QoL but they measured QoL at a single time point, six to eight weeks after the injury, the results are not fully comparable.

Moreover, we found no association between QoL and being born outside Sweden, being married or in a partnership, or living with children. In particular, the lack of an association between being married or in a partnership and QoL was surprising, since people with traumatic brain injury who are married or living with a partner report greater QoL than those living alone (44,45). One reason for this difference between studies could be related to cultural variations, the other studies being made in Brazil and USA (44,45).

Finally, injury-related variables, such as type of diagnosis, length of stay in hospital, FIM and presence of aphasia, did not show any significant association with change in QoL. Having aphasia did not change QoL in the present study, which is not in line with previous research which found that stroke patients who had aphasia after one year reported poorer QoL (46).

The EQ VAS reflects a person’s subjective judgement of overall QoL and the EQ-5D index reflects the subjective judgement of capabilities compared to a preference-based health profile (27). Thus, these two measures in EQ-5D reflect somewhat different aspects of the patient’s QoL. In the present study, however, the results from the two measures were similar in magnitude and in the direction of the relationships, which adds to the validity of the results.

Study limitations

Although the WebRehab Sweden material is large and well suited to testing hypotheses such as those in the present study, it has its limitations. The proportion of missing data in some of the variables was problematic, although this is common in quality registers where registration of some variables is optional. As a consequence, in this study it was not possible to adjust for more than one confounding variable at a time in the multiple regression analysis and at the same time retain a reasonable sample size and reliable results. Thus, large sample size studies to explore what factors that increase QoL after ABI are warranted, and our method with linear regression first and then multiple regression could overcome this problem.

Conclusion

The present study showed that returning to work and improved functioning (on GOSE) were the most influential factors for increasing QoL in people living with ABI. In addition, education level, and getting a suspended driving licence reinstated also influenced QoL, but to a lesser degree. From a clinical perspective, rehabilitation interventions directed at improving QoL for patients with ABI should focus on the abilities that are necessary for the patient to be able to return to work. We suggest that future research on brain injury rehabilitation should focus on longitudinal studies and QoL changes, which could help us understand what can predict a sustainable QoL and working life.

Declaration of interest

The authors report no conflict of interest. This study was supported by grants from the University Health Care Research Centre, Region Örebro County, Sweden.

Acknowledgments

We wish to thank WebRehab Sweden for the provision of data.