Prediction of all-cause and cardiovascular mortality with weight loss in patients with chronic heart failure: a meta-analysis

Abstract

Weight loss is a well-established strategy to prevent heart failure in overweight and obese individuals. However, weight loss for patients with established heart failure is a more complex topic due to an ‘obesity survival paradox’. Overweight and mild to moderate obesity in patients with heart failure is associated with an improved survival outcome compared with patients with normal weight. Patients with chronic heart failure (CHF) tend to experience progressive unintentional weight loss. The prevalence of unintentional weight loss, defined as a loss of 6% or more of body weight, ranges from 11.2% to 17% in patients with CHF. The prognostic values of weight loss for predicting mortality are controversial in patients with CHF. Therefore, we conducted this metaanalysis to evaluate the association of weight loss with cardiovascular and all-cause mortality risk in these patients. Two authors independently searched PubMed and Embase databases for studies from their inceptions to 5 July 2019 by using the following search terms in combination: ‘weigh loss’ AND ‘chronic heart failure’ AND ‘mortality’ OR ‘death’ AND ‘follow-up’. The reference lists of included articles were manually scanned for additional potentially relevant studies. Observational studies or post hoc analyses of randomised controlled trials evaluating the prediction of all-cause and cardiovascular mortality with weight loss in patients with CHF were included. Weight loss is defined by the loss of 5% of baseline body weight over 3 months or more. Studies including patients with acute heart failure were excluded. The methodological quality of the eligible studies was checked using the Newcastle–Ottawa scale (NOS) for cohort studies. Studies with a score of 7 or more were deemed to be of high quality. All data were analysed using STATA 12.0 (Stata, College Station, TX, USA). We pooled the hazard ratio (HR) with 95% confidence intervals (CIs) for the weight loss versus stable weight. Of 198 potentially relevant articles identified from the electronic databases, six articles (eight studies) enrolling 226,506 patients with CHF were ultimately included in this meta-analysis (see Supplementary Figure 1). Table 1 shows a summary of the study characteristics. These selected studies were published between 1997 and 2019. Most of the studies originated from Europe. Follow-up duration ranged from 9.0 months to 37.7 months. For methodological quality evaluation, the included studies were grouped as high quality (NOS score ranging from 7 to 8). All the included studies summarised the association of weight loss with all-cause mortality (see Supplementary Figure 2(a)). Comparison with weight loss and stable weight showed that the pooled HR of all-cause mortality was 1.74 (95% CI 1.35–2.24) in a random effect model. Evidence of significant heterogeneity was observed between studies (I1⁄4 84.2%; P< 0.001). Begg’s test (P1⁄4 0.266) and Egger’s test (P1⁄4 0.190) showed no evidence of publication bias. The sequential exclusion of individual studies minimally altered the original prognostic significance (HR range 1.54–1.86). Two articles (three studies) reported the data on cardiovascular mortality. As shown in Supplementary Figure 2(b), the pooled HR of cardiovascular mortality was 1.71 (95% CI 1.10–2.66) in a random effect model, with significant heterogeneity across studies (I1⁄4 90.2%; P< 0.001). Leave-one-out study sensitivity analyses further confirmed the reliability of the pooling risk estimate. This meta-analysis suggests that weight loss is independently associated with an increased risk of all-cause mortality in patients with CHF. CHF patients with weight loss had an approximately 74% and 71% higher risk of all-cause and cardiovascular mortality, respectively. These findings suggested that weight loss may be a strong and independent predictor of poor