Preoperative determinants of quality of life a year after coronary artery bypass grafting: a historical cohort study

Design

This was a single centre retrospective cohort study. Since patients were not subjected to investigational actions and were treated according to standard guidelines the need for informed consent was waived by the local review board of the ethical committee (Medical research Ethics Committee United, number W15.069). The study was conducted in accordance with the principles of the Declaration of Helsinki.

Population

All patients older than 18 years who underwent elective isolated CABG between the first of July 2011 and the first of May 2014 were eligible for inclusion. Inclusion criteria were a completed medical outcomes study Short Form (SF) 12 version 2 questionnaire at baseline and at 12 months after CABG. Surgical procedures were performed by multiple cardiothoracic surgery consultants and their supervised trainees in a tertiary referral hospital (St. Antonius Hospital, Nieuwegein, The Netherlands). Perioperative care was carried out according to standard clinical practice and based on international guidelines for all patients [10].

Clinical characteristics and data collection

As potential determinants for postoperative HRQL patient characteristics, medical history and comorbidities, preoperative laboratory tests and preoperative HRQL were considered.

Data on age, gender, body mass index and preoperative laboratory tests were collected from the routine preoperative anaesthesia visit. Medical history, additive euroSCORE and postoperative complications were collected from the hospital’s electronic database for cardiac surgery patients. Registration of postoperative complications was conducted in the context of a national registry of cardiac interventions in The Netherlands (Supervisory Committee for Cardiac Interventions (Begeleidingscommissie Hartinterventies Nederland) [11]. Intraoperative data was collected from the computerized medical record (MetaVision 5.46.44, iMDsoft®, Düsseldorf, Germany). A manual review of the data was performed to check for accuracy and missing values by a member of the research team (LV). Missing data on laboratory values, medication use and surgical characteristics were fully completed by manually checking electronic patient records.

Statistical analysis

Data are presented as frequencies and percentages of total for categorical data, as mean ± standard deviation (SD) for normally distributed continuous data and as median and interquartile range (IQR) for non-normally distributed continuous data. Normality was tested using visual inspection of histograms and Kolmogorov-Smirnov test. Scores for preoperative HRQL were compared to the Dutch population mean using a one sample t-test. A paired sample t-test was used to compare preoperative HRQL to scores at 1 year after surgery. A delta score was calculated for change in HRQL by subtracting preoperative SF-12 scores from postoperative SF-12 scores. A positive delta score represents an improvement in HRQL.

To study the clinical relevance of the difference in HRQL a threshold value was used to compare groups. As standard threshold values are lacking in CABG surgery for the SF-12 questionnaire, we based our threshold value on two articles using the SF-36 in elderly patients undergoing CABG or aortic valve surgery. Welke et al. defined a clinically relevant change as a difference of ≥5.42 points in physical or ≥ 6.33 in mental HRQL [15]. Jansen-Klomp et al. used a cut-off point of ≥2.5 points [16]. As with the SF-12 questionnaire the HRQL could increase or decrease as much as four points by changing the results of a single question, we set the threshold value at 5 or more points decrease or increase to ensure a clinically relevant change. Subsequently, the study cohort was stratified in three groups according to change in HRQL: worse, no change or better HRQL. Differences between the three groups were tested using Chi square test for dichotomous or categorical variables and one way ANOVA or Kruskal Wallis test for continuous data depending on normality.

To analyse the independent effects of all risk factors a multivariable model was built using multinomial logistic regression analysis. This analysis allows for a dependent variable with more than two categories. Patients with no clinically relevant change in HRQL were used as reference category to which the outcomes ‘worse HRQL’ and ‘better HRQL’ were compared. No variable selection took place and all variables were added simultaneously. To prevent multicollinearity, correlations between all variables were tested using Pearson’s correlation coefficient. Of variables with a correlation > 0.8 one variable was excluded from the model. Non-linearity of the continuous variables regarding preoperative physical and mental HRQL was investigated by adding various transformations and assessing model fit in terms of log likelihood. If model fit improved, the transformations were retained in the multivariable model. Results are presented as odds ratios with their accompanying confidence intervals. For statistical analysis IBM SPSS version 22 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.) was used. A p-value < 0.05 was considered as statistically significant for all analyses.

Study population

In total, 2100 patients underwent elective isolated CABG. Of these patients, 1225 completed a baseline SF-12. Thirty-two patients died within 1 year. 658 (31.3% of total, 53.7% of 1225) patients returned the 12 month follow up questionnaire and were included in the final cohort (Fig. 1). Overall, non-responders had more comorbidities compared to responders. This was expressed by significant differences in age, incidence of male gender, diabetes mellitus, prior myocardial infarction, LVEF and euroSCORE. Additional file 2: Table S2 shows other baseline variables for responders and non-responders.

Patient characteristics and outcome

Determinants of a clinically relevant change in physical HRQL

Physical HRQL 1 year after surgery improved in 150 (22.8%) patients, did not change in 403 (61.2%) patients and got worse in 105 (16.0%) patients. Preoperative median EuroSCORE was 3 (IQR 1–4), 3 (IQR 1–4), and 3 (IQR 1–5) for worse, no change or better physical HRQL respectively (p = 0.802). Median preoperative physical HRQL was 45.3 (IQR 41.5–48.5) for patients with worse physical HRQL; 39.7 (IQR 37.1–42.1) for patients with no change in physical HRQL and 34.8 (IQR 31.3–37.2) for patients with better HRQL (p < 0.001). Median preoperative mental HRQL was 47.2 (IQR 35.9–54.1); 50.5 (IQR 40.8–59.2) and 52.6 (IQR 43.1–60.8) for patients with worse, no change or better HRQL respectively (p = 0.001). Postoperative complications occurred in 6.7% of patients with worse physical HRQL, in 5.5% of patients with no change in physical HRQL and in 8.0% of patients with better physical HRQL (p = 0.536).

Determinants of a clinically relevant change in mental HRQL

Mental HRQL improved in 328 (49.8%) patients at 1 year after surgery, remained unchanged in 227 (34.5%) patients and worsened in 103 (15.7%) patients. Preoperative median euroSCORE was 3 (IQR 1–5) for patients with worse mental HRQL, 3 (IQR 1–4) for patients with no change in mental HRQL, and 3 (IQR 1–4) for patients with better mental HRQL (p = 0.887). Prior to surgery, physical HRQL was 37.9 (IQR 35.0–41.7); 37.9 (IQR 35.1–41.4) and 40.5 (IQR 37.0–43.9) for patients with worse, no change or better HRQL respectively (p < 0.001). Preoperative mental HRQL was 55.2 (IQR 46.9–63.7); 55.6 (IQR 46.6–63.6) and 45.1 (IQR 34.7–53.4) for patients with worse, no change or better HRQL respectively (p < 0.001). Postoperative complications were seen in 8.7% of patients with worse mental HRQL, in 5.3% of patients with no change in mental HRQL and in 6.1% of patients with better mental HRQL (p = 0.481).

Change in HRQL

In the majority of patients physical HRQL hardly changed, as was reflected by a mean increase of 0.7 points. Mental HRQL increased in half of patients, with a mean increase of 5.5 points.

Contrasting to our findings, other studies using the SF-36, showed a mean increase in physical HRQL ranging from 4.8 to 5.3 points and a mean increase in mental HRQL of 1.2 to 1.9 points [17, 18]. In a cohort of 1744 patients, Rumsfeld et al. assessed SF-36 before and 6 months after CABG surgery. Health related quality of life increased 5.3 and 1.9 points for physical and mental HRQL respectively. Comparable to our results, preoperative physical HRQL was identified as the most important risk factor for a change in HRQL after taking other preoperative cardiac and non-cardiac risk factors into account [17].

Deutsch et al. performed a study in 106 octogenarians undergoing CABG, valve surgery or CABG combined with valve surgery. They assessed HRQL by SF-36 three and 12 months after surgery and compared it to preoperative scores. At 3 months physical HRQL significantly increased with 5.1 points and mental HRQL was comparable to preoperative levels. Cardiac and non-cardiac comorbidities and procedural data were not identified as relevant risk factors for change in HRQL. Unfortunately, baseline HRQL was not considered as possible risk factor for a change in HRQL in their study [18].

In contrast to other literature reports, physical HRQL at 1 year after surgery increased to a lesser extent in our study, while the increase in mental HRQL was more eminent [15, 17, 18].

These differences could be due to the moment of measuring HRQL. At 1 year after surgery, which was the moment of measurement in our study, HRQL could have been affected by other factors as well, resulting in lower scores.

Furthermore, CABG is mainly performed to relieve complaints of angina, which is likely to result in improved physical functioning. In our cohort less than 10% of patients suffered from unstable angina while this was present in up to 28–61% in patients in other articles [17, 19]. It is conceivable that patients in this study suffered from fewer complaints before surgery and therefore did not notice a relevant increase in physical HRQL. Also, questions regarding mental HRQL in the SF-12 include items as feeling full of energy, calm and peaceful or feeling downhearted. Although patients did not report an improvement in questions on physical functioning, CABG surgery may have improved mental status by relieving anxiety and enhancing feelings of security and self-esteem. The increase in mental HRQL in our study could reflect the overall benefit of the surgery.

Risk factors for change in HRQL after cardiac surgery

Preoperative risk stratification based on patient-centered outcomes, such as HRQL, could have great additional value in cardiac surgery but remains challenging as well designed risk models are lacking. Possible risk factors for change in HRQL that are readily available such as comorbidities, laboratory values or LVEF have been considered by others but resulted in conflicting results. Female gender [20, 21], older age [22], diabetes mellitus [15, 21, 23], body mass index > 35 [15], low LVEF [21], pulmonary disease [15], vascular disease [15], EuroSCORE > 3 [8], deprived socio-economic status [23] and smoking [23] have been associated with worse HRQL following cardiac surgery. Older age [15], high social support [23], and EuroSCORE > 6 [24] have been associated with better HRQL after cardiac surgery. However, several other studies, including ours, found no association between preoperative clinical factors and change in postoperative HRQL [17, 19]. Studies that included preoperative HRQL in their analysis concluded that HRQL prior to surgery was the most promising predictor for postoperative change in HRQL [15, 17, 19] and that routine preoperative assessment of HRQL should be incorporated in standard care to supplement current risk assessment [25, 26].

Some limitations should be addressed. First, the retrospective design limited the amount of available data. Comparison of responders versus non-responders showed that non-responders were older, more often male and showed a higher prevalence of diabetes mellitus, myocardial infarction, lower LVEF and higher euroSCORE. These factors can have a negative effect on postoperative HRQL [15, 20–23]. With inclusion of these patients likely greater differences in HRQL might have been present and possibly more preoperative predictors would have been identified. A possible reason for non-responding could be the method of approach during the follow up period, where questionnaires were sent by email without a reminder for unanswered questionnaires. Not all patients have email, which is more often the case for elderly. Second, obviously no SF-12 scores were available for deceased patients and these patients were excluded from the analysis. Mortality risk is highest for patients with more comorbidities. It is conceivable that this excluded group of patients had more comorbidities, leading to lower scores for preoperative HRQL and, subsequently different change scores. However, 1 year mortality was merely 2.6% and it seems unlikely this had a major influence on results. Third, only elective surgery patients were analysed limiting the generalisability and excluding patients with emergency CABG. However, the indication for emergency surgery is focussed on survival, while the main indication for elective CABG is to relieve angina. In elderly patients scheduled for elective surgery risk factors for postoperative HRQL are more essential for the decision making process than in patient presenting for emergency surgery.