Prognostic association supports indexing size measures in echocardiography by body surface area

Abstract

Body surface area (BSA) is the most commonly used metric for body size indexation of echocardiographic measures, but its use in patients who are underweight or obese is questioned (body mass index (BMI) < 18.5 kg/m² or ≥ 30 kg/m², respectively). We aim to use survival analysis to identify an optimal body size indexation metric for echocardiographic measures that would be a better predictor of survival than BSA regardless of BMI. Adult patients with no prior valve replacement were selected from the National Echocardiography Database Australia. Survival analysis was performed for echocardiographic measures both unindexed and indexed to different body size metrics, with 5-year cardiovascular mortality as the primary endpoint. Indexation of echocardiographic measures (left ventricular end-diastolic diameter [n = 230,109] and mass [n = 224,244], left atrial volume [n = 150,540], aortic sinus diameter [n = 90,805], right atrial area [n = 59,516]) by BSA had better prognostic performance vs unindexed measures (underweight: C-statistic 0.655 vs 0.647; normal weight/overweight: average C-statistic 0.666 vs 0.625; obese: C-statistic 0.627 vs 0.613). Indexation by other body size metrics (lean body mass, height, and/or weight raised to different powers) did not improve prognostic performance versus BSA by a clinically relevant magnitude (average C-statistic increase ≤ 0.02), with smaller differences in other BMI subgroups. Indexing measures of cardiac and aortic size by BSA improves prognostic performance regardless of BMI, and no other body size metric has a clinically meaningful better performance.

Figure 1

Flowchart describing patient inclusion. Exact numbers for the respective cardiac measure specific populations are given in Supplementary Table 1.

Figure 2

Average prognostic strength (C-statistic) for predicting 5-year cardiovascular mortality when indexing for body size for right atrial area, left atrial volume, left ventricular diameter, left ventricular mass, and aortic sinus diameter. The axes represent the height and weight exponents of a body size indexation metric of the format height^x⋅weight^y. The color scale shows the average C-statistic from 0.60 to 0.70, each color increment representing a 1%-point improvement. Existing body size metrics were plotted: h height, w weight, hw height⋅weight, BMI body mass index, BSA body surface area, M Mosteller, D DuBois. In the normal weight/overweight group, from unindexed to BSA by Mosteller there is a 4%-point improvement, but further improvement is limited to < 1%-point. Similar trends can be observed in the underweight and obese group. Echocardiographic measures with a limited number of observations were not included. See text for details.

Figure 3

The combined prognostic strength of indexing by body surface area across cardiac measures regardless of BMI for the prediction of 5-year cardiovascular mortality. Left atrial size and left ventricular mass had the highest prognostic strength, with a C-statistic 10%-points higher than left ventricular size and most aortic dimensions. Right atrial and right ventricular sizes had an intermediate prognostic strength. Aortic sinus diameter had the strongest prognostic strength of all aortic measures, and aorta at sinotubular diameter had the weakest prognostic strength. RA right atrial, RV right ventricular, LA left atrial, LV left ventricular, IVS interventricular septum.

Figure 4

Flow-chart representing the logic, data, and reasoning for using body surface area for the indexation of all echocardiographic measures regardless of BMI.