Does the obesity paradox apply to congenital heart disease?

The linked paper by Brida et al 1 is an observational study on a prevalence cohort consisting of people with adult congenital heart disease (ACHD). Observational studies are appropriate for generating hypotheses and for examining safety. The hypotheses should then be examined by randomised controlled trials (RCTs), or using randomised Mendelian techniques.2 Incidence cohorts give information about the natural history of a disease, but may lack statistical power for looking at mortality as the number of events initially is small. In prevalence cohorts, there is not a delay in events occurring and hence often the number of subjects studied is relatively small.

This present study looks at 3086 people. The numbers become quite small when looking at body mass index (BMI) groups: only 6.2% were underweight. The follow-up length was short: 2.54 years (IQR: 0.65–5.31 years).

They report a U-shaped relationship between BMI and mortality. Mortality is highest in those with low and high BMI.

Critically the lowest mortality was found at a BMI of 34.1 kg/m², much higher than anticipated.

As the authors state, the U or J relationship between BMI and mortality has been found previously in the general population,3 and in diverse long-term conditions including diabetes,4 coronary heart disease5 and heart failure,6 and even survival following surgery for, for example, hip replacement7 and cardiac surgery.8 In all the unselected studies the optimum BMI for low mortality was higher than anticipated, although not as high as in the present study.

What are the potential explanations for the U-shaped curve and the particularly high value of BMI associated with the lowest mortality in people with ACHD? It seems tenable that the reasons for this relationship apply to all the different clinical areas studied, but particularly exaggerated in this clinical situation. Smokers die young, and weigh less than the general population. Those with ill health may be cachectic and die sooner than the healthy population. The numbers in the present study are limited, which makes the results unreliable from subdividing into groups such as smokers, not presently smoking and never smoking.

It is useful, therefore, to examine the findings in the general population,3 with a meta-analysis of some 30.3 million people. In this study, using all the participants, they found a U-shaped relationship with the nadir at a BMI of 25. In people who had never smoked, the relationship was J-shaped, with the lowest mortality at a BMI of 23–24. In people without a long-term condition and who had never smoked, the nadir was a BMI of 22–23.

In addition, there are many diseases where weight loss precedes the diagnosis, and this can only be allowed by excluding people with a short follow-up, that is, within a year of follow-up. When this exclusion was performed, there was no increased risk in those with a BMI of 20, although there remained a J-shaped curve at lower values.

In the present study of people with ACHD, these two confounders—smoking and ill health directly from the heart disease—are likely to explain much of the U-shaped relationship.

Importantly, however, the authors have also demonstrated that weight loss in those with complex disorders is associated with an increased mortality. Out of the patients with ACHD, they had data on the change of weight in 1757 patients over 2.87 years. Of these, 467 patients had complex ACHD disorders.

It was only in this group that weight loss was significantly associated with mortality. This may possibly be just because the mortality was particularly high in this group.

An important hypothesis therefore is supported: that weight loss is associated with both increased mortality and lower BMI. Another is suggested: that treating weight loss nutritionally at an early stage might alter mortality.

The authors rightly point out the significance of this weight loss to the clinician in assessing the patient's prognosis.

Are they right to urge caution about encouraging weight loss in the mildly obese with congenital heart disease? Weight reduction can be achieved by eating less or exercising more, or both. There are a small number of studies comparing increased activity with control groups in people with ACHD: none showed increased mortality in the active group.9 There are a larger number of studies in heart failure, where there again is no statistically significant reduction in mortality with increased exercise, but clear evidence of safety.10 The impact of calorie restriction in the obese person with ACHD has not been studied in a large-scale RCT.

Cachexia is seen in several clinical situations11; the current definition is of unintentional weight loss (>5% of body weight) and a low BMI (≤20 in those aged less than 65 years or ≤22 in those aged more than 65 years).

This work1 suggests that this BMI value should be modified at least in people with ACHD. Should cachexia in ACHD be actively managed? As with other causes marked nutritional deficiencies can occur, and a dietetic assessment is required. Trials of attempts to encourage weight gain by diet in ACHD with cachexia are absent.

In someone with ACHD, low body weight is an important risk factor for early death. Weight loss, particularly unintentional, is an ominous sign. There is little evidence of harm from suggesting increased exercise but no convincing evidence that these authors could find on the safety of weight-reducing diet in people with ACHD.

As the authors suggest therefore, clinical trial evidence is needed on the safety of weight reduction diets in the obese and overweight person with ACHD. Meanwhile, attention to exercise is more appropriate.