Article Text

Download PDFPDF

Original research article
Trajectories and characteristics of work disability before and after acute myocardial infarction
  1. Mo Wang1,
  2. Marjan Vaez1,
  3. Thomas Ernst Dorner2,
  4. Jari Tiihonen3,
  5. Margaretha Voss1,4,
  6. Torbjörn Ivert5,6,
  7. Ellenor Mittendorfer-Rutz1
  1. 1 Division of Insurance Medicine, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
  2. 2 Department for Social and Preventive Medicine, Centre for Public Health, Medical University of Vienna, Vienna, Austria
  3. 3 Department of Clinical Neuroscience, Centre for Psychiatric Research, Karolinska Institutet, Stockholm, Sweden
  4. 4 Department for Analysis and Forecast, Swedish Social Insurance Agency, Stockholm, Sweden
  5. 5 Department of Cardiothoracic Surgery and Anesthesiology, Karolinska University Hospital, Stockholm, Sweden
  6. 6 Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Dr Mo Wang, Department of Clinical Neuroscience, Division of Insurance Medicine, Karolinska Institutet Stockholm SE-171 77, Sweden;{at}


Objective Scientific knowledge on work disability in terms of sickness absence and disability pension (SA/DP) among patients with acute myocardial infarction (AMI) is scarce. The study aimed to investigate trajectories of SA/DP among individuals with or without AMI and examined the associations between sociodemographic, morbidity and coronary revascularisation characteristics with such trajectories among patients with AMI.

Methods This is a population-based cohort study of 10 255 individuals aged 30–60 years who had a first AMI during 2008–2010 and were alive 30 days after AMI. Each case was matched by sociodemographics to one control without AMI. Trajectories of annual SA/DP months over a 6-year period for cases and controls were analysed by group-based trajectory modelling. Associations of characteristics with trajectory groups were estimated by Χ2 test and multinomial logistic regression.

Results The majority of patients (59%) had slightly increased annual levels of SA/DP (1 month) at the time of AMI, which returned to no SA/DP 2 years after the event. One group (4%) had increasing SA/DP months preceding and after AMI. Three groups showed constant SA/DP months on low (14%), medium (9%) and high (15%) levels. Sex, unemployment, education, musculoskeletal disorders and prescription of antidepressants were the strongest factors discriminating the SA/DP trajectories (R2 difference=0.02–0.03, p<0.01).

Conclusions The majority of patients with AMI have a good outcome in terms of low levels of work disability within 2 years after AMI. Patterns of SA/DP before AMI, sex, socioeconomic status as well as comorbid musculoskeletal and mental disorders provide crucial clinical information on work disability after AMI.

  • Acute myocardial infarction
  • Sick leave
  • Disability pension
  • Insurance Medicine
  • Trajectory

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Acute myocardial infarction (AMI) has become an economic and public health problem since it is the most common cause of burden of disease and premature mortality worldwide.1 During 2010 about 20 000 patients were treated for an AMI in Sweden, of whom about 30% were younger than 65 years,2 which in most societies represent the working-age population. To date, temporal work disability, that is, sickness absence (SA), is a necessary rehabilitation measure in healthcare for patients to recover after an AMI.3 Current Swedish clinical guidelines recommend 4–8 weeks of SA in case of AMI without complications.3 These guidelines note that the work ability can be permanently impaired after a major AMI with complications, requiring granting of disability pension (DP). National cohort studies revealed that a quarter of patients had DP already at the time of coronary revascularisation, and up to 32% were granted DP during a follow-up of 5 years.4 5 Still, there is sparse scientific knowledge on trajectories of SA/DP before and after AMI.

A higher risk of prolonged SA and DP following AMI is reported, especially among women, and is associated with socioeconomic position, comorbid mental and musculoskeletal disorders, and work disability in terms of SA/DP during the year before AMI.5 6 However, the course of work disability after AMI is likely to be heterogeneous. Work disability in some individuals might improve, while others will remain at stable levels over time after AMI. Understanding the heterogeneity of work disability is of crucial clinical importance in order to design individualised treatment and rehabilitation measures. Here, comparison of work disability trajectories in individuals without AMI can provide reference information of the general population.

In order to describe groups with different health development after AMI, scrutiny of a number of sociodemographic and morbidity characteristics is necessary. Low socioeconomic positions, female sex and older age may increase the risk of work disability.7 8 From a clinical perspective, prior studies have demonstrated a number of comorbidities in patients with AMI, including diabetes mellitus, stroke, hypertension, cancer, musculoskeletal disorders and chronic kidney disease.9 In addition, comorbid common mental disorders (CMDs) have been reported to be associated with work disability subsequent to AMI.10 Moreover, the type of coronary revascularisation might determine the outcome after AMI.11 The most frequent coronary revascularisation technique in case of AMI is percutaneous coronary intervention (PCI), followed by coronary artery bypass grafting (CABG).11 While some scientific knowledge exists on return to work after CABG,12 studies comparing the outcome after both CABG and PCI with regard to SA and DP are missing.


The aim of this study was to investigate trajectories of SA/DP among individuals with or without AMI. Furthermore, we examined the potential associations between different characteristics (sociodemographic factors, morbidity, ie, length of inpatient care due to AMI, comorbidity and infarction type, and type of coronary revascularisation) with such trajectories among patients with AMI.


Study population

The study base consisted of all individuals (n=11 579) who had a first AMI between 30 and 60 years of age during 2008–2010 in Sweden. AMI was defined based on the International Classification of Diseases (ICD)-10 codes: I21 (main diagnosis). Individuals with previous specialised healthcare due to AMI as main or side diagnoses from 1987 until the date of AMI diagnosis were excluded (n=990). Moreover, individuals who died within 30 days after the AMI (n=252)5 and individuals with missing information on sociodemographic variables (n=112) were excluded. The final study population comprised 10 225 individuals with a first AMI treated in inpatient care during 2008–2010. A comparison group with all individuals between 30 and 60 years of age, resident in Sweden during 2008–2010 and who did not have an AMI treated in inpatient care during 1987–2013 was introduced (n=10 225). For each AMI case, one control was randomly selected with an exact matching without any replacement on sex, age, education, type of living area, country of birth and family situation on December 31 of the year preceding baseline (ie, 2008–2010). National registers were linked by using the unique personal identity number assigned to all Swedish citizens.

Sickness absence and disability pension

In Sweden, sickness benefits cover every individual older than 16 years with an income from work or unemployment benefits who has a reduced work capacity due to disease or injury.13 In case of permanently impaired work capacity, all individuals can be granted DP. In the study, information on SA/DP was obtained from the Micro Data for Analysis of the Social Insurance (MiDAS) register from the Social Insurance Agency. The SA/DP months were combined, and the mean annual number of SA/DP months during 3 years before and after the AMI diagnosis was calculated.

Sociodemographic, morbidity and coronary revascularisation characteristics

Characteristics of individuals with AMI included sociodemographic characteristics (ie, sex, age, education, country of birth, type of living area, family situation and days with unemployment), morbidity characteristics (ie, length of inpatient care due to AMI, type of infarction, specific comorbid disorders measured by a history of diagnosis-specific healthcare from inpatient and outpatient care, prescribed psychiatric medication), as well as the type of coronary revascularisation. Sociodemographic data were obtained from Statistics Sweden (LISA register) and recorded on December 31 of the year preceding the AMI diagnosis and categorised as shown in table 1.

Table 1

Descriptive statistics of sociodemographic characteristics of individuals with a diagnosis of acute myocardial infarction (AMI) from inpatient care in 2008–2010 in Sweden (n=10 225)

Information on morbidity and type of coronary revascularisation was obtained from the National Patient Register and the National Prescribed Drug Register. Length of inpatient care due to AMI was categorised based on the median length of inpatient care. Type of infarction was classified as ST-elevation myocardial infarction (ICD-codes: I21.0-I21.3), non-ST-elevation myocardial infarction (ICD-codes: I21.4) or unspecified (ICD-codes: I21.9). These ICD-10 codes were based on detailed analyses of elevated troponin levels in combination with typical symptoms and/or ECG findings.14 A history of specialised healthcare was measured from 3 years before the AMI diagnosis. Somatic comorbidity was measured as any main or side diagnosis from inpatient or specialised outpatient care due to musculoskeletal diagnoses (ICD-codes: M00-99), renal insufficiency (ICD-codes: N17-N19), stroke (ICD-codes: I60, I61, I63, I64), hypertension (ICD-codes: I10), cancer (ICD-codes: C00-D48) and other somatic disorders (ICD-codes: A00-B99, D50-E07, E15-E90, G00-H95, J00-L99, N00-N16, N20-O75, O85-O99, Q00-T98). Individuals with any main or side diagnosis in inpatient or specialised outpatient care due to diabetes mellitus or having any prescribed antidiabetic medication during 3 years before the AMI diagnosis were coded according to ICD-codes: E10-E14 and the Anatomic Therapeutic Chemical classification system (ATC) code: A10.

Mental comorbidity included main or side diagnosis from inpatient or specialised outpatient care due to CMDs (ie, depressive (ICD-codes: F32-F33), anxiety (ICD-codes: F40-F42) and stress-related disorders (ICD-codes: F43), other mental disorders (ie, ICD-codes: F00-F31, F34-F39, F44-F99)) and prescribed psychiatric medication, which was measured in categories of defined daily doses (DDDs) during the year preceding the AMI diagnosis. Categories of prescribed psychiatric medication were grouped as follows: (1) no psychiatric medication, (2) small doses or occasional use (<0.5 DDD/day), (3) moderate doses (0.5–1.5 DDD/day) and (4) high doses (>1.5 DDD/day).15 Prescriptions of antidepressants, anxiolytics and sedatives were included following the ATC codes N06A, N05B and N05C.

Information on type of coronary revascularisation at the event was based on the Classification of Surgical Procedures and categorised as PCI (FNG00-FNG05), CABG (FNA-FNF, FNG30, FNW96) and others (ie, other surgeries or missing information) (table 2).

Table 2

Descriptive statistics of morbidity and coronary revascularisation characteristics of individuals with a diagnosis of AMI from inpatient care in 2008–2010 in Sweden (n=10 225)

Statistical analyses

We used Pearson’s Χ2 tests to estimate sex differences regarding sociodemographic characteristics, morbidity and type of coronary revascularisation among patients with AMI. Trajectories of annual SA/DP months among individuals with or without AMI were independently estimated during a 6-year window with group-based trajectory modelling.16 The group-based trajectory modelling is able to identify subgroups of individuals who follow distinct patterns of change over time and estimate the proportion of individuals in each group. The Bayesian information criterion was used to test the best fitted model in relation to the number of groups, and five groups were selected as the best fitted model for patients with AMI and for the comparison group. The estimated probabilities for each trajectory group belonging to the AMI group and the comparison group were 0.95 and 0.92, respectively, which indicated a good fit.17

Sociodemographic characteristics, morbidity and type of coronary revascularisation in each SA/DP trajectory group among all individuals with AMI (n=10 255) were explored by Pearson’s Χ2 test and multinomial logistic regression. We used the likelihood ratio tests to evaluate whether the characteristics were associated with all five trajectory groups in the full model including all covariates. Moreover, Nagelkerke pseudo-R2 were estimated to measure the strength of these associations.18 Each factor was consecutively excluded and reincluded from the full model in order to calculate the differences in R2 and to test the contribution of a given factor to the full model. Additionally, the data on SA/DP for an individual who died (information from the cause of death register) or emigrated (Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA) register) were considered as missing for the whole year of the event and onwards. A sensitivity analysis with those excluded individuals showed comparable results.

Data processing was performed using statistical software SAS for Windows V.9.4 (SAS-based procedure ‘Traj’) and SPSS for Windows V.22.0 (Χ2 test and multinomial logistic regression).

Ethical approval

The study was approved by the Regional Ethical Review Board of Stockholm, Sweden.


Of the 10 225 patients with first AMI during 2008–2010, 2341 (22.9%) were women (table 1). A larger proportion of the patient group was older (51–60 years, 69.6%), had a high school education (51.8%) and was not unemployed (90.3%). The majority of patients were treated emergently with PCI (70.7%) after the AMI, whereas CABG was uncommon (2.0%). There were more men than women (74.4% vs 57.9%) in the PCI group (tables 1 and 2).

Trajectory analyses of SA/DP

We identified five groups of trajectories of annual SA/DP months for the AMI and the comparison group. The groups of patients with AMI were labelled as the following: ‘Variable SA/DP’, ‘Increasing SA/DP’, ‘Low SA/DP’, ‘Middle SA/DP’ and ‘High SA/DP’. The groups for the comparison group were called ‘Low SA/DP’, ‘Increasing SA/DP’, ‘Decreasing SA/DP’, ‘Middle SA/DP’ and ‘High SA/DP’ (figure 1).

Figure 1

Trajectory groups of sickness absence and disability pension (SA/DP) months and percentages of individuals with acute myocardial infarction (AMI) from inpatient care in 2008–2010 and those without (comparison group) respectively (n=10 225) within each trajectory group. The dotted lines represent 95% CIs. The value for T0 was calculated as the mean value for T-1 and T1 and inserted here for pedagogic purposes.

For patients with AMI, a majority of the individuals (58.5%) belonged to the ‘Variable SA/DP’ group with constantly no or few months of SA/DP up to the year of the AMI. Their SA/DP months increased during T0–T1 from approximately 1 month to nearly 2 months and then decreased to no/very low SA/DP levels. Such pattern was not observed in the comparison group. Of the individuals with AMI in the ‘Increasing SA/DP’ group, 3.6% started with a low level of SA/DP and their SA/DP levels rapidly increased from T1 to T2. The ‘Low SA/DP’ (14.3%), ‘Middle SA/DP’ (8.7%) and ‘High SA/DP’ (15.0%) groups had on average around 1, 7 and 12 months of SA/DP/year, respectively. The majority of the individuals in the comparison group belonged to the ‘Low SA/DP’ group (76.3%) (figure 1).

All the characteristics were associated with trajectory groups in the full model, except for ‘type of infarction’ and ‘hypertension’ (tables 3–5). Sex, education, unemployment, musculoskeletal disorders and antidepressants (difference in R2=0.02–0.03) had a more important impact than other factors in the full model (tables 3–5).

Table 3

Distributions and associations of sociodemographic characteristics in each trajectory group of sickness absence/disability pension (SA/DP) months/year in individuals with a diagnosis of an acute myocardial infarction (AMI) from inpatient care in 2008–2010 in Sweden (n=10 225)

Table 4

Distributions and associations of somatic and mental morbidity characteristics in each trajectory group of sickness absence/disability pension (SA/DP) months/year in individuals with a diagnosis of an acute myocardial infarction (AMI) from inpatient care in 2008–2010 in Sweden (n=10 225)

Table 5

Distributions and associations of morbidity and coronary revascularisation characteristics in each trajectory group of SA/DP months/year in individuals with a diagnosis of AMI from inpatient care in 2008–2010 in Sweden (n=10 225)

Compared with other groups, men (83.6%) and individuals with higher education (26.4%) were more common in the ‘Variable SA/DP’ group. In contrast, the ‘High SA/DP’ group included a higher proportion of women, older individuals, individuals with lower education, who were born in other Nordic countries, were more likely to be single living without children, but less likely to be unemployed than other groups. Being born in a non-European country and unemployment more than 180 days was common in the ‘Increasing SA/DP’ group.

Additionally, individuals in the ‘Variable SA/DP’ group showed less comorbidity and less use of psychiatric medication. On the contrary, the ‘Middle SA/DP’ and the ‘High SA/DP’ groups had more specialised healthcare, particularly due to musculoskeletal disorders, diabetes mellitus, renal insufficiency, other somatic disorders and mental disorders, as well as more prescribed psychiatric medication compared with other groups. More individuals in the ‘Variable SA/DP’ group were treated with PCI, whereas CABG was more common in the ‘Increasing SA/DP’ group.


With a sizeable population-based cohort of patients diagnosed followed from 3 years before to 3 years after a first AMI, we have identified five different trajectories of SA/DP as well as found evidence for a heterogeneity in patients with AMI. Three groups had SA/DP levels that were hardly affected by AMI, that is, their levels were constant over the entire observation period. The ‘High SA/DP’ group had around 12 months of SA/DP annually over time, and thus it is likely that this group consisted of a larger proportions of individuals with long-term SA or DP. Although there was a relatively comparable sociodemographic profile between the trajectory groups, we found that women, older patients with AMI and individuals with lower education were more likely to be included in the ‘High SA/DP’ group. This sociodemographic profile is in line with what the existing literature suggests regarding sociodemographic factors among individuals with long-term SA and DP.8 19 Moreover, the patterns of DP at the time of revascularisation showed comparable results according to a previous study.4

The largest group, the ‘Variable SA/DP’ group, showed an obvious and rapid increase of SA/DP months during the year of AMI diagnosis. In fact, AMI events occur suddenly.20 In our study, this group showed an increase of SA/DP months from a very low level to 1 month during the year of AMI diagnosis, and this trend continued until a level of around 2 months to 1 year after. Afterwards, SA/DP months returned back to a very low level. The comparison group in the study did not show such trend in SA/DP months around baseline. According to the national guidelines for sickness certification published in Sweden in 2007, patients with AMI are generally recommended to be sickness absent for 4–8 weeks.3 It is therefore important to identify risk factors for SA up to 1 year after the AMI. On the other hand, we found that the ‘Increasing SA/DP’ group had increasing work disability after AMI. This group included more individuals who were born in a non-European country and those who were previously long-term unemployed. Immigrants and unemployed individuals are more prone to psychological stress and unhealthy behaviour and thus their burden of disease is greater.21 Therefore, the drastically increasing SA/DP months after AMI might reflect deteriorated health conditions or poor rehabilitation in this patient group.

Our present study adds important knowledge on the diversity of SA/DP in patients with AMI, which is well in line with the established fact that patients with AMI comprise a heterogeneous group with regard to aetiology, healthcare-seeking behaviour and comorbid diseases.20 For the majority of patients, atherosclerosis is the main pathological process leading to AMI.20 The rate of progression of atherosclerosis is influenced by cardiovascular risk factors, including smoking, sedentary lifestyle, obesity, hypertension, diabetes mellitus, associated comorbidities like renal insufficiency, and stroke.20 Hence, a larger proportion of patients who belonged to the ‘Middle SA/DP’ group and the ‘High SA/DP’ had different comorbidities prior to AMI. Comorbidity among patients with AMI may act as a risk factor for a high SA/DP level before AMI, and may also increase the risk of work disability after AMI. In contrast, the comorbid diseases were not prevalent in the ‘Variable SA/DP’ group. In this group, a larger proportion of men were observed. A possible explanation is that women may have more comorbidities and are more prone to seek help in healthcare services than men.7

Our results further indicate that musculoskeletal disorders and CMDs (measured as antidepressant use) are strongly related to SA/DP after AMI than other comorbid diseases. Mental and musculoskeletal disorders, which are the most common diagnoses for SA/DP in the general population,13 are also common among patients with AMI.4 A recent study reported a synergistic effect between ischaemic heart disease and CMD with regard to a future risk of DP.10 The underlying mechanism may involve stressful life circumstances, adverse health behaviour and low help-seeking behaviour, which are associated with mental and musculoskeletal disorders as well as AMI.22

In the current study, we did not find a strong association of the type of coronary revascularisation with SA/DP trajectories; still, a larger proportion of individuals in the ‘Variable SA/DP’ group were treated with PCI, while more of those in the ‘Increasing SA/DP’ group with increasing levels of work disability after AMI were treated with CABG. This might indicate that patients with AMI might have a better outcome in terms of lower levels of work disability after being treated with PCI than CABG. Further studies are needed to elucidate the association between type of revascularisation and future work disability.

Methodological considerations

The strengths of this study include the use of a population-based cohort design, which offered satisfactory statistical power for the analyses. The use of nationwide register data of high quality also minimises the risk of recall bias regarding exposure and outcome.23 Moreover, we could include a wide range of characteristics. Still there might be other factors than those studied here that are associated with work disability trajectories, for example, cardiac rehabilitation programmes.

Regarding limitations, we did not have information on sick-leave spells <14 days among employed individuals. This means that for employed individuals, the number of SA days contributing to the combined number of SA/DP days might be underestimated. As the Swedish government launched stricter rules for SA/DP in 2008, the shorter duration of SA and fewer transitions to DP after 2008 were observed in recent studies.24 Thus, the SA/DP trajectories after T0 might have been influenced by such regulation changes. However, there was not an obvious decreasing trajectory of SA/DP, which was found directly after T0 for the two cohorts. It is therefore not specifically likely that the regulation changes had a major effect on the trajectories observed in this study.


Most of the patients had low levels of work disability 2 years after AMI. Patterns of SA/DP before AMI, sex, socioeconomic status, as well as comorbid musculoskeletal and mental disorders are clinical crucial factors to explain work disability after AMI.

Key messages

What is already known on this subject?

A higher risk of prolonged sickness absence (SA) and disability pension (DP) following acute myocardial infarction (AMI) was reported previously.

What might this study add?

We found that nearly two-thirds of patients had no or very low SA/DP levels before AMI, with an increase at the time of the AMI and a decrease 2 years thereafter, while 41% of the remaining groups had constantly low, medium, high and increasing SA/DP levels over time. The lowest proportions of individuals with female sex, low education and comorbidity with mental and musculoskeletal disorders were in the groups with low SA/DP levels

How might this impact on clinical practice?

Most of the patients had low levels of work disability 2 years after AMI. This information is important to convey to patients after an AMI. Knowledge on the identified characteristics is useful for clinicians to design person-based rehabilitation measures.



  • Contributors All authors contributed to design of the study and the statistical analysis plan. MW was responsible for data management and data analysis. MW and EM-R drafted the manuscript, and all authors were involved in interpretation of the data, critical revision and approval of the final content before submission.

  • Funding This study was funded by the Swedish Research Council dnr 2015-02292.

  • Competing interests None declared.

  • Ethics approval Regional Ethical Review Board of Stockholm, Sweden.

  • Provenance and peer review Not commissioned; externally peer reviewed.