Results

Demographic and outcome distribution

Total cohort
The study population comprised 3217 (82%) of a total of 3904 patients aged 18 years or above having undergone cardiac surgery from April 2013 to April 2014 (Figure 1). Median age was 68 (range 23 – 95), with almost 70% being 60-79 years, 76% were male and 2085 (65%) had a spouse or partner. Most patients (2340 (73%)) had an income of 50-150% of the median for the total population. Isolated CABG were performed on 1548 (48%) patients and 707 (22%) had non-isolated procedures of cardiac surgery (Table 1). Of the total population, 110 (3.4%) patients died within 90 days, 286 (9%) patients experienced prolonged ICU stay, 1653 (51%) patients prolonged hospital admission and 1249 (39%) patients were readmitted to hospital within 90 days after cardiac surgery. A total of 542 (14%) patients were excluded due to lack of follow-up data in the national registers. Furthermore, we had to exclude 145 (3%) patients for whom we did not have information on EuroSCORE (Figure 1, flow-chart).
DenHeart cohort
In the nested DenHeart sub-sample, including a total of 982 patients, 771 (79%) were male, the median age was 67, 320 (33%) lived alone and 705 (72%) had an income between 50-150% of the median (Table 1). Only 1 patient in the nested subsample died, 63 (6%) patients experienced prolonged LOS-ICU, 443 (45%) patients experienced prolonged LOS-HOSP and 348 (35%) patients were readmitted to hospital within 90 days after cardiac surgery. Since only one patient died in the DenHeart nested sub-sample, death was not included as an outcome in these analyses. In total 1576 (49%) did not participate in the DenHeart cross-sectional survey and were therefore excluded from the nested sub-sample (see flow-chart, Figure 1).

Individual candidate predictor associations in logistic models

Total cohort
Logistic analyses adjusted for EuroSCORE revealed that patients who were living alone (OR, 1.19; 95%CI, 1.02-1.38) and had a lower educational level (OR, 1.27; 95% CI, 1.04-1.54) were more likely to experience prolonged LOS-HOSP, whilst patients in the highest quartile for income were less likely to experience prolonged LOS-HOSP (OR, 0.83; 95% CI, 0.69-0.98) (Table 2).
Regarding prolonged LOS-ICU, patients who lived alone had an increased OR when adjusting for sex and age (OR, 1.33; 95% CI, 1.03-1.70), but the association was not present when adjusting for EuroSCORE (Table 2). The outcomes of readmission and death did not show any statistically significant associations.
DenHeart cohort
Logistic regression analyses in the nested DenHeart population revealed that a lower score on the physical component scale of the SF-12 was associated with LOS-HOSP when adjusting for age and sex, but not when adjusting for EuroSCORE. However, the mental component scale was associated with LOS-HOSP in both models (OR, 1.43; 95% CI, 1.02-2.01) and with LOS-ICU when adjusting for age and sex, but not when adjusted for EuroSCORE. Furthermore, a global score lower than the median on the HeartQoL questionnaire and a high score on the B-IPQ were associated with LOS-HOSP in both models (OR 1.35; 95% CI, 1.04-1.75 and OR, 1.58; 95% CI, 1.09-2.29, respectively). By contrast a medium score on the B-IPQ was found to be associated with LOS-ICU. Finally, being alone though preferring to be with others was associated with an increased OR for readmission within 90 days (OR, 1.41; 95% CI, 1.03-1.91) and not having someone to talk to was associated with an increased OR for LOS-HOSP (OR, 1.95; 95%CI, 1.25-3.04) when adjusting for EuroSCORE (Table 3).

Prediction equation of emotional, social, functional and behavioural factors

Total cohort
The multiple regression models based on comprehensive data revealed low educational level and living alone as predictors of prolonged LOS-HOSP (Table 4). The discriminative value was acceptable based on AUC of 0.625, however, the Brier score of 0.238 indicates a poor informative model (Table 4). Furthermore, living alone was found to be a predictor of LOS-ICU with an acceptable Brier score of 0.078 and AUC of 0.676 (Table 4), and death with a Brier score of 0.032 and AUC of 0.710 (Table 4). None of the candidate variables were predictive of readmission including EuroSCORE, which did not predict readmissions either (AUC, 0.53; Brier score, 0.24) (Table 4 and Supplementary figure 1).
DenHeart cohort
The multiple regression model for prolonged LOS-HOSP included a low global HeartQoL score and not having someone to talk to as predictors (Table 5). The discrimination was acceptable (AUC 0.62), however, the informative value of the model was poor (Brier score of 0.24) (Table 5). None of the candidate predictors in the DenHeart study were found to be predictive of prolonged LOS-ICU (Table 5), however, being alone though preferring to be with others predicted readmission (Table 5). Model fit was found to be poor (Brier score of 0.23), and discriminating ability was low (AUC of 0.56) (Table 5 and Supplementary figure 2).