Right ventricular function and exercise capacity in adults with congenital heart disease: Methodological caveats that temper the clinical message

Dear Editor,

Vanreusel et al. deserve credit for combining a comprehensive two-dimensional multiplane echocardiographic assessment of the right ventricle with a randomised home based exercise programme in a heterogeneous adult congenital heart disease cohort [1]. The observation that right ventricular strain tracks with both exercise capacity and self reported health status is clinically intuitive and adds to a still thin evidence base. A few methodological features, however, deserve comment before the findings are generalised into practice.

The correlation analysis in Table 2 tested six echocardiographic parameters against two exercise capacity indices and eleven quality of life domains, yielding well over sixty pairwise comparisons, none corrected for multiplicity [1]. The authors acknowledge this in their limitations, but the practical consequence is understated: at an uncorrected alpha of 0.05, roughly three “significant” associations would be expected by chance alone even if right ventricular function had no true relationship with exercise capacity or quality of life. The reported correlations were themselves only small to moderate (r ≈ 0.3–0.5), precisely the range most vulnerable to inflation from multiplicity and from the modest feasibility of strain measurement in some walls (64–81%) [1,2]. A pre-specified primary comparison or at minimum a false-discovery-rate adjustment, would have made the hierarchy of findings easier to trust [3].

The headline efficacy result improved peak oxygen consumption, muscle strength and quality of life “in the patients who adhered to the training schedule” comes from a per-protocol analysis after 39% of the exercise arm was excluded for non-adherence. Per-protocol analysis breaks randomisation: patients who complete a home-based aerobic programme differ systematically from those who do not, in motivation, symptom burden and baseline function and this selection cannot be distinguished from a true treatment effect once non-adherers are removed [4]. The intention to treat analysis, relegated to the supplement, is the estimate that actually respects randomisation and it should have anchored the abstract and conclusion rather than the per-protocol figures.

The trial was inevitably unblinded and several of the outcomes that improved most convincingly energy/fatigue and the visual analogue scale were entirely self-reported. Unblinded trials with subjective, patient-reported endpoints are recognised to systematically overestimate intervention effects through expectation and observer bias [5]. This concern is sharpened, not lessened, by the fact that no right ventricular parameter improved with training: if the mechanistic hypothesis linking right ventricular performance to functional capacity and wellbeing is correct, the dissociation between unchanged cardiac structure/function and markedly improved self-reported energy and VAS scores merits more circumspect interpretation than the discussion currently offers.

None of this diminishes the value of a well conducted exploratory study in a difficult to recruit population. But given multiplicity uncorrected correlations, per-protocol efficacy estimates and unblinded subjective endpoints, the conclusion that “exercise training should be integrated into CHD treatment” is better framed as hypothesis generating than as a management directive, pending an adequately powered, intention to treat anchored confirmatory trial.

Funding

None.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.