In a population based study of 1585 patient treated with PCI we found that the patients’ self-rated health one month after the procedure was a significant better predictor of RTW compared to other variables including LVEF, both at short (12 weeks) and long (one year) term. Mental health was nearly as important as physical health. The results were robust to stratifications and restrictions.
This study was based on a large cohort of PCI patients with complete register based follow-up of working status. The clinical data from WDHR are complete with respect to individuals 
, and only few values are missing. Self-reported health integrates all the patient’s perceptions and beliefs and because the questionnaire was completed 4 weeks after the PCI, the patients already have some knowledge about the effect of the procedure and the doctor’s advice of sick leave. This could be a part of the causal pathway. The wording in SF-12 is retrospective “In the preceding 4 weeks…”, but for 345 patients the outcome of interest (RTW) occurred before answering the questionnaire. We made an additional analysis excluding these patients, but this changed the c-statistics only marginally.
Missing answers of single items of the SF12 scale resulted in 156 missing values of the mental component score (MCS) and the physical component score (PCS). Analyses based on multiple imputations using available SF-12 items, age, gender and working status did not change estimates or size of confidence intervals in analyses using the two component scores, and consequently the original data were used. A few missing values were present in the variables BMI, LVEF, smoking status and information about diabetes (). Since these variables originates from a clinical database where clinicians could forget to fill in some items, we consider these variables as missing completely at random (MCAR), since whether or not a variable was missing was not related to the outcome of interest and thus not likely to cause any bias. We had complete follow-up of weekly working status due to full coverage of registers of transfer income in the DREAM database, although there may be misclassifications as sick leave less than 2 weeks does not qualify for national benefit. The grouping of transfer-payment groups may cause misclassification, if a person on labour market–related benefit is not ready to work due to health problems, but fails to report this. However, the group was small and as citizens on labour market–related benefit are strongly encouraged to confirm their readiness to work every week, we consider this a minor problem. Misclassification of the outcome may occur in persons who are not working, but provided income by their spouse or live as rentiers. In Denmark this is rather uncommon as only 2% of the population between 40 and 67 years are without personal income, so we consider it a minor problem.
Measuring RTW as a time-to-event measure has the disadvantage that it ignores relapses if a new sick listing occurs later. A major strength of using the DREAM register in relation to RTW is the ability to establish a measure that captures the RTW dynamically, and not only time to first RTW. In this study 245 persons experienced one or more new sick listings. We made additional analysis by excluding these 245 patients with a dubious outcome, and this enhanced the associations found.
Non-responders had lower LVEF and were less likely to have worked just before the PCI. We repeated the analysis after placing non-responders in the lower groups of the component scores, and this did not change the estimates found for self-reported health.
The proportion of patients working before the PCI was similar to previous findings 
. RTW among patients working before the PCI compares well with previous studies 
. Age 
and gender 
have previously been found to predict RTW in mixed populations of heart patients. Another study found neither gender nor age related to RTW 
. However, Nielsen et al. found that gender modified the effect of low LVEF on RTW 
. Studies of mortality in heart disease have found that gender differences diminish after adjustment for age 
, but we did not reproduce this with the outcome RTW. Gender differences in sickness absence are well known in Western countries 
, and are suggested to come from both direct and indirect gender effects, such as differences in daily life and social position 
Both myocardial infarction prior to the PCI 
and LVEF 
has previously been identified as predictors. We found that both acute myocardial infarction and LVEF were predictors; however, the self-reported measures of health were even stronger predictors of RTW.
Self-reported health has previously been found to predict readmissions and mortality after cardiovascular disease 
. A large Dutch study of 1-year mortality after angiographic procedures found that problems with self-care and low self-rated health were the most powerful predictors among 22 clinical variables and reported that addition of self-reported health improved the model c-statistics from 0.78 to 0.81 
. In our study, an addition of the single-item general health from SF12 to the clinical information improves the model c-statistics from 0.66 to 0.73 for RTW at 12 weeks and from 0.66 to 0.75 at 1 year. Adding the complete SF12 score improved the c-statistics even more.
In both the analysis of risk factors and predictors we found that the mental health component score were nearly as important as the physical component scores. A recent review indicated that not only poor physical health, but also poor mental health was associated with adverse prognosis (mortality and rehospitalisation) in hearth disease 
. Anxiety and depression has previously been found associated with RTW after heart disease 
and recent work has suggested a close relationship between negative emotions/distressed personality and the risk of incident heart disease as well as poor prognosis 
This study covers nearly all incident PCI patients, under the age of 67 years, originating from a well-defined population in Denmark. The findings are supposed to have high external validity in relation to countries with similar rules and regulations regarding health-related benefits.
Patients’ subjective experiences may differ from clinicians’ views and objective measures such as LVEF. Patient-reported measures and objective measures should complement each other, and thus create a better basis for clinical advice to the patient and risk stratification in RTW.