The trial was conducted on a subpopulation from the open randomised parallel group controlled trial, LC-REHAB [20]. Patients were randomly allocated to the intervention arm (LC strategies in addition to standard CR) or to the control arm (standard CR) in a 1:1 ratio stratified for hospital unit, gender and diagnosis (IHD or HF) in blocks of two to four using a web-based system [20]. The allocation procedure was generated independently by the research team. Additional eligibility criteria in the present trial were applied after randomisation to exclude patients assessed with permanent work disability at inclusion. Information on inclusion dates was retrieved from the LC-REHAB trial [20]. Follow-up was defined as the week in which the date equivalent to 12 months after inclusion appeared. The trial was conducted and reported according to the CONSORT standards – extension for randomised trials of non-pharmacological treatment [22].

Patients were recruited between 30th November 2010 and 20th December 2012. Trial information was sent by postal mail to eligible patients referred to CR. Information of the trial was provided by telephone by the last author of this paper. Written informed consent, enrolment and randomisation were performed by health professionals at the CR units [20]. A total of 827 patients hospitalised for either IHD or HF were included in the LC-REHAB trial (Fig. 1). Two patients were excluded due to error in randomisation procedure. Of the remaining 825, 413 were randomised to the LC arm and 412 to the control arm.

Patients were enrolled at the CR unit and were eligible for the LC-REHAB trial if they were aged above 18, referred to, and motivated for CR after hospitalisation for IHD or HF. If the patients were diagnosed with both IHD and HF, they were classified as having HF. For specific ICD-10 codes included see the initial protocol for the LC-REHAB trial and previous study publication [20, 21]. The exclusion criteria were acute coronary syndrome within the last 5 days before inclusion; active peri-, myo-, or endocarditis; symptomatic and/or untreated heart valve disease; severe hypertension with blood pressure > 200/110 mmHg; other severe cardiac or extra cardiac disease; planned revascularisation; senile dementia, assessed as having poor compliance for participation in and completion of the trial; or previously participation in the trial.

Patients in both arms received a phase II CR program lasting 8 weeks based on the national Danish guidelines on CR starting the first workday after inclusion [20, 23]. The programme was delivered in a hospital setting and consisted of group based sessions all lasting 1.5 h with a weekly education session and three exercise sessions per week. The content of the education sessions in both arms were split in eight topics, which were chosen in collaboration with experienced health professionals in CR and experienced patients who previously had undergone CR [20]. The topics were: Function and symptoms of the heart, lifestyle effects on the development of IHD and HF, emotional reactions, medication, tiredness, the importance of relatives or other networks, importance and types of exercise, and future life with a chronic disease. The education sessions were primarily managed by a nurse. Exercise sessions consisted of aerobe exercise and muscle strength training managed by a physical therapist.

Both arms received CR by the same pair of a nurse and a physiotherapist throughout the CR programme; the pairs were designated to either the intervention or control arm throughout the trial. Due to the nature of the intervention, blinding of health professionals or patients was not possible. Sessions in the two different arms were performed at different times of the day [20].

In addition to the described CR intervention, LC strategies took a situated, reflexive, and inductive approach to education and exercise. The rationale behind the pedagogical approach was based on theories behind LC strategies and was described in the initial protocol for the LC-REHAB [20]. The rationale was applied through practical implications consisting of: Two individual clarifying interviews (before and after CR), experienced patients as co-educators, material developed for each topic including background literature and questions to facilitate discussions. The approach was ensured by health professionals completing an 8 days competence-education in LC strategies with experienced patients participating the last 4 days, and 1 h evaluation meetings between the pair of health professionals and experienced patient once a week [20].

The CR programme in the control arm was the formerly used in the hospital units (standard CR) [23]. The rationale was not described and education and exercise consisted of structured deductive teaching. Identical pre-written, educational slide-shows were used as material for the education sessions.

The primary outcome was RTW status at 1 year follow up. After trial completion, information on the RTW outcome was retrieved from DREAM, which is administered by The Danish Ministry of Employment [24]. The register includes all Danish citizens, who at some point since 1991 have received public benefits. Patients were identified in DREAM by their social security number. Each person is registered once a week with a code representing the type of transfer payment received that particular week [25]. For patients to be categorised as having a RTW status (yes) at 1 year follow up, the four consecutive weeks prior to the week of one-year follow up were either codeless (self-support) or had codes representing State Education Fund grants or flexi jobs. For the secondary outcome, each patient was categorised as having an event of RTW during one-year follow up (yes/no). The first event of four consecutive weeks of either self support or codes representing State Education Fund grants or flexi jobs was categorised as RTW (yes). The patients who experienced the event of RTW during follow up but were not registered RTW at 1 year follow up were identified. They were referred to as “relapsed patients”.

Baseline variables concerning: age, gender, height, weight, diagnosis (IHD or HF), presence of diabetes, smoking, civil status and former participation in CR were reported to dedicated databases in the LC-REHAB trial by the nurses at the CR units [20].

At the first CR session, the health professionals handed out self-reported questionnaires for assessing: presence of depression, level of education, and the annual household income [20]. Depression (yes/no) was assed by Major Depression Inventory (MDI) [26]. Level of education was classified by low, medium, or high. Household income was classified by low, medium, or high. For elaborated classification on level of education and household income, see the initial protocol [20].

Since work status prior to CR was assumed associated with RTW, additional information on employment was retrieved from DREAM after allocation to the trial arms [14]. No self support prior to CR was dichotomised (yes/no) encompassing whether the patient within 6 months prior to inclusion were self-supported for at least 1 week.

Descriptive statistics were used to compare the baseline characteristics – Chi-squared for the binary and categorical outcomes, and Studen’s t-test for the continuous. RTW status at 1 year was compared between the two arms using a logistic regression model. The result was presented both as unadjusted and adjusted odds ratios with 95% confidence interval (CI). Adjustments were carried out for the stratification variables: gender, cardiac diagnosis and hospital unit. Additional adjustment for age was performed, as age was expected to be associated with RTW [11]. To address the secondary aim of the trial, frequencies and percentages described the number of patients who experienced the event of RTW during follow up across trial arms. A comparison of the relapsed patients across arms was performed by chi-square test. All analyses were performed based on the intention-to-treat principle. Analyses were performed using Stata 14 software [27]. Data management was performed blinded from allocation.

Suggesting a 14% difference in RTW proportions between arms estimated from Kruse et al. [11], the given sample size on 244 patients left this trial with a power on 89% testing on a 5% level of significance.

A total of 526 patients were excluded due to age criteria, 55 due to receiving disability pension (n = 47) or passive social assistance (n = 8) on the date of randomisation. Two-hundred-and-forty-four patients were included for the present trial on RTW; 119 and 125 in the LC arm and control arm, respectively. Mean age was 51.8 years and the majority of the patients were men (77.0%). A small fraction (16.8%) had no self support prior to CR (Table 1). Baseline variables balanced across arms with no statistically significant differences between the two arms on baseline variables except for gender (males accounting for 83.2% in the LC arm vs. 71.2% in the control arm (p = 0.03)) (Table 1). Non-responders balanced across arms in baseline variables containing missing values (results not shown).

Information bias was considered minimal; DREAM has been validated against workplace-registered job attendance and long-term sick listing and found to have high sensitivity and specificity [34]. Classification of RTW based on transfer payments from DREAM has elsewhere been defined based on various numbers of weeks (ranging from 1 week to 5 weeks) [12, 13, 16]. The chosen definition of four consecutive weeks in this trial might have affected the frequencies of RTW but was not expected to be differentiated between arms. Selection bias was furthermore not considered as there was complete follow up; therefore, threats to the internal validity of the trial were not assumed.

The trial was carried out in western Denmark where the population in general is lower educated than the total population of Denmark and the trial enrolled patients with HF [35]. Both level of education and living with HF are associated with increased risk of not returning to work and may have caused the overall lower RTW proportion (61-73%) in this trial compared to other studies [11, 12, 16].

According to the power calculation, a 14% difference in RTW was expected between the two trial arms; however the trial detected a 4 percentage point difference. It was expected that lacking practical implications in the LC strategies for improving RTW rather than a low sample size was the reason of no difference.

The estimate may have been affected towards the null hypothesis as mutual interaction between the arms was plausible due to lack of blinding of the health professionals. Moreover, patient education was delivered in both arms and the effect of the patient education in the control arm may have contributed to even out the effect of LC strategies.

Approximately, 50% of the population with IHD and HF asked to participate declined to participate in the trial [21]. However, no knowledge of the patients that declined was accessible and it was thus unknown if selection of the patients was present at enrolment. This caused limitation of the generalisability of the results and the trial was not able to provide answers about the effect on the total population of people with IHD or HF.

Temporal and contextual factors affect the ability to RTW and influence the external validity of the outcome measure. Also, this means that comparing results in an international context should be done carefully due to heterogeneity in RTW definitions and occupational systems.