Geert Kleinnibbelink

Chapter 9 190 DISCUSSION Confirming our hypothesis, we provide the first evidence that CR in patients with PH is associated with significantly lower 1-year mortality compared to patients who did not receive CR. Specifically, PH patients who were prescribed CR demonstrated 40% lower odds of 1-year mortality. Previous RCTs showed improvement in exercise capacity, symptomatology, and quality of life after a relatively short-term follow-up (3-15 weeks). 4-10 However, whether these surrogate endpoints are associated with outcomes in PH has never been demonstrated. 11, 12 Our data suggests CR and exercise programmes may ultimately translate to clinically meaningful improvements in mortality. The poor long-term survival in PH, which also is demonstrated in the high 1-year mortality rate in this cohort of 21%, stresses the urgency of adding CR and exercise to existing (non)- pharmacological treatment to further improve survival. Facilitated through improved digital registration, this retrospective cohort study represents a useful and feasible alternative to the extremely challenging and demanding prospectively designed studies. Nonetheless, our unique observations highlight the need to confirm the potential of CR to reduce mortality, using RCTs. Some limitations warrant consideration. First, characterization of disease states and therapy were based on ICD-codes, which may vary between healthcare organisations. 13 Along these lines, distinction was made between primary PH (i.e., caused by disease of the pulmonary arteries) and secondary PH (i.e., secondary to other, non-vascular causes) rather than the current WHO-classification for PH into group 1 to 5. 3 Given the complexity of PH phenotypes, some PH patients labelled as ’secondary PH’may also have a variant of primary PH. As CR is beneficial for secondary causes of PH, such as cardiovascular diseases and risk factors, the exclusion of these individuals does not necessarily alter the strength of our findings. A second limitation is that we were unable to evaluate characteristics of CR and/or adherence to CR, which limits the ability to identify successful factors of CR to reduce mortality. Third, although we matched patients for important co-morbidities and demographic factors, residual confounding and the relatively modest sample size should be considered. Finally, we were unable to evaluate the potential effect of CR on morbidity, primarily because of too small sample sizes of subgroups that were available for individual co-morbidities (e.g. stroke, myocardial infarction, heart failure). The relatively