Estimation of the maximal heart rate for exercise prescription in heart failure patients: are we there yet?

Abstract

Exercise prescription for heart failure (HF) is a cornerstone for a clinically effective and safe cardiac rehabilitation (CR) programme. 1,2 Likewise, the prescription of the intensity according to the ventilatory thresholds, identified by cardiopulmonary exercise testing (CPET), is considered an optimal method. 2-4 Although with lower accuracy, when CPET is not available, a percentage of the measured maximal heart rate (MHR) during an exercise stress test (EST) is considered the minimum standard in recent CR guidelines worldwide. 2,3 However, professionals must be prepared to safely assess and monitor patients when this assessment is unavailable 5 since the evidence shows that most centres are still not performing this evaluation prior to beginning CR. 6 Thus, in the absence of maximal tests (CPET or EST) or a submax-imal volitional test (field and submaximal tests), the rating of perceived exertion and the talk test are two commonly recommended techniques to estimate the intensity for appropriate exercise prescription since the determination according to a percentage of the predicted MHR can be highly divergent in HF patients, mainly due to negative chronotropic influence of the beta-blockers, and disease-related dysfunctions. 2 A recent study by Magrì et al. 7 developed a new equation for MHR prediction in HF patients with reduced left ventricular ejection fraction (LVEF) and treated with beta-blockers. According to this cohort including 3487 stable HF subjects, the equation utilizes four variables [e.g. age, rest heart rate (HR), LVEF, and anaemia], as following [MHR = (109-(0.5 × age) + (0.5 × rest HR) + (0.2 × LVEF)-(5 if haemoglobin <11 g/ dL)]. The derived equation improved historical and cohort-based, most routinely used formulas, which can be valuable to determine the real maximal effort during an EST when the CPET is unavailable. The authors also concluded that a percentage range of 75-80% of the MHR, deriving from the new equation, might help to identify the intensity domains during a CR programme. The authors are applauded for the massive amount of work that has been put into this study, aiming to provide a clinically relevant toll with the best accuracy possible to be utilized in daily practice; however, despite the unquestionable relevance of the research, we felt that its indiscriminate use for exercise prescription should be evaluated with caution. Although more acceptable, the coefficient of determination provided with the new equation remained low (r 2 = 0.24) and the standard error high (17.5 b.p.m.), indicating that independent selected variables (age, rest HR, LVEF, and anaemia) are not explaining much the variation on the MHR for the majority of the patients (76%). The increased dispersion may generate substantial imprecision in exercise prescription domain identification, hereby potentially precluding its large-scale clinical use, particularly on those with different phenotypes and co-morbidities than those assessed in the study. Hansen et al. 8 have already demonstrated that even the intensity domains prescribed according to a percentage of a measured MHR can be extremely incompatible with the current guideline-based recommendations due to individual HR scattering in response to exercise. Likewise, an exercise intensity determination according to an estimated MHR can generate greater inconsistencies , lowering CR prescription accuracy, which may also decrease safety and efficacy. Hence, considering the importance of assessing the reproducibility and generalizability of prediction models, we felt that external validation of the equation from Magrì et al. 7 could be relevant. To assess the new equation applicability, we obtained the MHR of a sample of 191 HF patients with reduced LVEF (age: 55.8 ± 13.7 years; 69.6% male; body mass index 27.0 ± 4.5 kg/m 2) retrospectively assessed on treadmill CPET exams performed in a private outpatient cardiologic clinic in Brazil. These data are part of a multicentric study, approved by the institutional review board under number CAAE: 35706720.4.0000.8093. All patients continued beta-blocker treatment and achieved maximal effort (respiratory exchange ratio of 1.05 or higher) during the stress test and were in sinus rhythm. Heart failure was secondary to coronary artery disease in 51.3% of the cohort, with an average LVEF of 38.3 ± 6.7%. Conclusively, the median and interquartile range of the measured MHR was 135 (114-155) b.p.m., while the estimated MHR by Keteyian et al. 9 was 119 (114-126) b.p.m. and by the new equation 7 121 (116-128) b.p.m. These groups were compared by the post hoc for multiple comparisons as appropriate. Interestingly, the estimated MHR from both formulas (Keteyian et al. 9 and the new formula 7) were not statistically different from each other (P = 0.11), and the measured MHR was different from both estimation's formulas (P < 0.01 and P < 0.01). The association between measured and estimated MHR was assessed by the Spearman correlation coefficient, which revealed a moderate correlation between the measured and estimated values (rho = 0.576 for Keteyian and 0.592 for Magrì). Finally, the agreement between the measured and estimated MHR, assessed by