Heart failure (HF) is a complex clinical syndrome characterized by dyspnea, fatigue, and fluid retention. Globally, there is an estimated 20 million people suffering from this syndrome, with a projected increase of 25% in prevalence in 2030 [1
]. Despite advances in management, HF remains a major cause of mortality and morbidity worldwide, imposing a huge burden on the health care system [2
]. Therefore, improving primary prevention of HF is of great importance for public health.
Previous studies demonstrated that consumption of chocolate products may confer salutary cardiovascular effects [3
]. It is found that both acute and chronic chocolate intake reduce blood pressure [4
], which is a major risk factor for HF. In addition, chocolate consumption has been shown to be associated with lower incidence of myocardial infarction and stroke [5
], lower cardiovascular mortality [6
], and improved vascular function in HF patients [7
]. However, there are limited studies focusing on the association between chocolate consumption and risk of HF, and their results remain inconsistent. We, therefore, carried out a dose–response meta-analysis to determine the role of chocolate intake in prevention of HF.
There were limited studies that have investigated the association between chocolate intake and risk of HF. Our meta-analysis showed that light-to-moderate, but not high, consumption of chocolate was associated with a reduced risk of HF. Besides, we observed a nonlinear relationship between chocolate consumption and risk of HF in dose–response meta-analysis.
Chocolate is an important dietary source of flavonoids, a subclass of polyphenols. Previous studies have demonstrated that flavonoids in chocolate may be responsible for the salutary effects of chocolate on blood pressure, possibly by acting as an angiotensin converting enzyme inhibitor [20
]. Flavonoids also offer improvements in other risk factors for HF, including increasing high-density lipoprotein cholesterol, reducing inflammation, and improving endothelial function [21
]. In addition, it is well known that flavonoids have great antioxidant properties [23
]. Experiments using human endothelial cells in culture have also indicated that flavonoids activates nitric oxide synthase, thus stimulating the generation of nitric oxide and contributing to the maintenance of normal cardiac function [24
]. Therefore, these factors in combination may explain the protective role of chocolate against HF.
In dose–response meta-analysis, we found a J-shaped relationship between chocolate consumption and risk of HF. Initially, compared to no chocolate consumption, a moderate consumption is associated with a risk reduction in HF incidence of 16%. A higher than moderate intake is not associated with a decreased risk. This result may be attributable to the high calorie content of commercially available chocolate. When consuming chocolate in high amounts, the high-energy content of chocolate may lead to increased weight gain [26
], a known risk factor for developing HF [27
]. In addition, it is plausible that higher consumption of chocolate causes less energy intake from other foods that may be salutary for the prevention of HF. Thus, the benefits of flavonoids appear to be countered by the adverse effects of high energy intake when consuming excessive chocolate products. In three individual studies subgroups of BMI < 25 kg/m2
] and those with lower energy intake [10
], a more linear pattern of risk reduction with more chocolate consumption was observed. However, in the overall results, all reported individual studies showed a J-shaped curve.
Our finding is in contrast to a recent meta-analysis [10
], which showed that chocolate consumption was not associated with incident HF (HR: 0.87, 95% CI: 0.71–1.06). In that study, the authors only pooled the risk estimates of HF for the highest versus lowest category of chocolate intake, and because of considerable heterogeneity, were very prudent with results, pointing towards a non-significant benefit of consuming chocolate. A previous study [18
] indicated that the association between chocolate consumption and HF was stronger in lean than in overweight/obese subjects. In the present study, we found no difference in risk of HF between subgroups with normal or elevated BMI. The reason for the lack of difference is unclear. However, this result should be treated with caution, given the limited studies available for subgroup analysis. Future studies concerning this issue are required.
Of course, some remaining points need to be addressed. In all of the included studies, the incidence of HF was reported as cumulative incidence in years, and we do not know the HF-free survival curves. Although our results suggest that chocolate consumption may play a role in the prevention against HF, we cannot derive from the data how many years of chocolate consumption are needed to reach this result, and cannot exclude the possibility that it may contribute only to delaying the occurrence of HF, even possibly even only by delaying the occurrence of other cardiac events that would trigger the development of HF. In addition, it is unclear whether the competing risk of death may affect our final results. Competing risk of death is an important consideration in geriatric studies [29
]; however, none of the original studies have addressed the effect of this competing risk on HF incidence. Due to a lack of access to individual participant data, performing additional competing risk analysis in our work is also infeasible.
There are several limitations in our work. First of all, the recall and selection bias cannot be eliminated because of the observational nature of original studies. However, all included studies had a prospective design, and we only selected the maximally adjusted HRs for pooling analyses. Second, due to the lack of relevant data, we cannot evaluate the risk of HF for consuming different types of chocolate or different amounts of energy intake as was noted to have an interaction with the effects of chocolate consumption in the study of Kwok et al. [10
]. For the latter interaction, we would argue that the HRs of the individual studies have been adjusted for energy intake, and are less likely to have influenced our results in this meta-analysis. Third, the lower HF incidence for low-to-mild chocolate consumption observed in our study may be associated with healthier dietary habits. Subjects consuming less chocolate may have healthier diets than those consuming higher amount of chocolate. Nevertheless, most dietary factors were not adjusted in the original studies. Fourth, all included studies were conducted in USA or Europe. Thus, generalization of our findings to other populations should be used with caution.