Determinants of non-recovery in physical health-related quality of life one year after cardiac surgery: a prospective single Centre observational study

Traditionally, objective outcome measures, including survival, complication rates and need for re-interventions have been used to benchmark cardiac surgery. During the last decades additional assessment of health-related quality of life (HRQoL) scores have been used in the evaluation of cardiac interventions. Typically, HRQoL scores serve as secondary outcome parameters in the comparison of strategies related to coronary artery bypass grafting (CABG), e.g. on-pump versus off-pump techniques or CABG versus percutaneous coronary interventions (PCI) [1, 2]. Alternatively, studies focus on specific baseline characteristics, such as age, sex or severity of angina in relation to HRQoL-related outcome measures [3‐7]. However, despite a well-established general improvement in HRQoL in various settings of cardiac surgery, a recent study underlined the observation that 20% of patients experienced worsening of HRQoL 1 year after CABG as compared to the preoperative status [8]. Although this is generally considered acceptable, such assessment may differ substantially from the perception of individual patients, especially in the setting of limited reduction in HRQoL prior to surgery [6]. The purpose of this study was to identify risk factors that interfere with improvement of HRQoL 1 year after cardiac surgery.

In the present study 21.9% of all cardiac surgery patients did not show an improvement in the physical domain score of the SF-36 HRQoL between the preoperative period and 1 year after surgery. Baseline SF-36 physical domain score, diabetes, female sex, any postoperative infection and the need for PCI in the first year were identified as independent risk factors for non-recovery. Selection of isolated CABG patients did not change this picture. An important finding in our study is the lack of predictive power of the currently used EuroSCORE II risk model with respect to physical recovery. Specifically designed and calibrated for the prediction of short-term mortality, the predictive power of the EuroSCORE II risk model is known to diminish with respect to 1 and 5 year mortality [12, 13]. Notwithstanding the fact that other risk models claim better performance for long-term mortality, the relationship between such risk scores and HRQoL-related endpoints remains largely unexplored [14, 15].

At first glance our findings seem to contradict the existing literature. After all, cardiac surgery including CABG, is associated with an improvement in even very long-term HRQoL [7, 16, 17]. A hallmark of these papers is a comparison in HRQoL over time for a substantial cohort of patients. However, such general improvement does not rule out the existence of subgroups of patients that do not benefit from surgical intervention. Recently, a reduction in the physical component scale of the Veterans Rand (VR)-36 questionnaire was observed in 13.7% of CABG patients 1 year after surgery, despite a general improvement in the total population [8]. In addition, in 39.7% of the patients physical performance remained unchanged over time. Similarly, a general improvement in HRQoL up till 18 months was observed in a cohort of CABG patients. However, the authors highlighted the fact that they also observed an age-dependent decline in HRQoL in patients > 75 years of age between 6 and 18 months after surgery [18]. In a more selected group of high-risk patients with an indication for either surgical aortic valve replacement or transcatheter aortic valve implantation, an absence in increased physical performance at 6-months was as high as 58.7% [19]. Our data do not confirm age as independent risk factor, as previously reported in the setting of elective CABG and aortic valve replacement [20, 21]. This may be due to differences in inclusion criteria (our study also reports on non-elective surgery patients) or the fact that age was included in our multivariate model as a continuous, rather than a categorical variable.

In general, comparison of our findings with the existing literature on HRQoL after cardiac surgery is hampered by a the fact that many papers have a retrospective design, lack baseline values and use different HRQoL grading system [4]. Our data confirm a previous publication in which diabetes was also identified as an independent risk factor for deterioration in VR-36 derived physical performance prior to and 1 year after surgery [8]. However, others observed a similar improvement in overall HRQoL 5 years after CABG in diabetic and non-diabetic patients, despite lower baseline values [22]. Among multiple factors, both diabetes and female sex appeared to be independent risk factors for non-improvement in HRQoL in a cohort of CABG patients [23]. The observation of female sex as a specific determinant for worsening HRQoL after 6 months and 10 years respectively has been confirmed by others [5, 17, 22]. To the analogy of our observations, baseline HRQoL as an independent marker of improvement in HRQoL was additionally identified in large cohort of CABG patients with a 10 years follow up. Risk factors for non-recovery beyond baseline characteristics are scarcely studied. In a mixed cardiac surgery group prolonged length of stay (> 2 days) in the Intensive Care Unit was associated with reduced recovery in SF-36 scores as compared to propensity matched controls [16]. Although we were unable to replicate these data, it is clear that the observed influence of any infection during hospital stay is likely to reflect morbidity from a different perspective.

We believe our data may have profound implications for the way we inform patients prior to surgery. The inability of commonly applied risk models to predict physical non-recovery adequately urges us to extend our knowledge over the full chain of events in the preoperative, perioperative and postoperative setting. The fact that preoperative physical performance is strongly related to the patient’s ability to improve (i.e. the odds for non-recovery increased roughly 5% for each increase in baseline SF-36 physical domain score by one point) seems to make sense from an analytical perspective. However, from the patient’s perspective this may be very relevant. If the aim of the surgical procedure is to improve the physical performance of the patient, an unsatisfactory result becomes more likely in case baseline physical performance is only mildly affected. In addition, women seem less likely to improve in comparison to men. If so, it is not only fair to incorporate such wisdom in the way we obtain informed consent, but it also necessitates further research to unravel sex-related mechanisms beyond the acute coronary syndrome [24]. Efforts to reduce infections in the postoperative phase seem worthwhile and such knowledge may help to motivate staff for better compliance with infection prevention bundles. Finally, the possibility of recurrence of symptoms that fuelled the process towards a surgical procedure in the first place (i.e. angina), represented by the need for PCI within 1 year after surgery, may deserve more attention in our communication with the patient. In addition, further research to unravel underlying mechanisms, including the interplay between recurrence of angina with perioperative infections, is needed [25].

The current study has several limitations. The single centre design does not allow to extend our conclusions beyond the specific setting. Despite the prospective design, including assessment of baseline HRQoL, the amount of missing data is substantial. Furthermore, we deliberately chose to restrict our analysis to the physical performance domain of the SF-36, were others suggest to combine different domains of HRQoL in order to provide additional HRQoL information. However, multivariate models with multiple endpoints become very difficult to interpret and suffer from statistical power, even in a substantial cohort with over 800 patients. Instead we chose to present data on the (non)recovery of mental HRQoL, and its overlap with physical HRQoL in the ESM. In general, our study fully complies with 8 out of 10 recommended quality criteria for HRQoL research in the specific field of cardiac surgery [26]. Furthermore we were unable to incorporate frailty in our model, which may have yielded additional information [27, 28]. Overlap with items such as ‘poor mobility’ that characterizes vitality to some extent remain to be established. Although the response rate of 45% is generally considered acceptable in this field of research, we have to acknowledge the fact that the substantial part of the population was not included. From the ESM Table 5 it becomes clear that non-responders were in general significantly more at risk. However how this translates to HRQoL remains to be established. Finally, it must be noted that pseudo R-squares of our multivariate models (0.23 for all patients and 0.26 for CABG) indicate that yet unnoticed factors may be of influence outside the scope of our analysis. This should fuel further research that incorporates other factors yet unaccounted for, such as social status, feeding and rehabilitation programs.

One year after cardiac surgery 22 % of all cardiac surgery patients experienced a decrease in physical HRQoL. Baseline physical domain score, diabetes, female sex, any postoperative infection and the need for PCI in the first year were identified as independent risk factors for non-recovery. Further research is needed to tailor the patient selection procedure prior to surgery and potentially modify risk factors in the perioperative process.