Estimation of body weight using anthropometric parameters in Sri Lankan hospitalized adult patients

Abstract

Body weight is an important clinical parameter for accurate dosing of drugs with a narrow therapeutic window, However, it is difficult to measure the body weight of a patient if they cannot stand on a scale. There are several anthropometrics-based equations to estimate the body weight, but most of these are derived from white Caucasian populations and are not validated for South Asians. This study aimed to validate existing anthropometrics-based weight estimation equations and develop a new equation for the same purpose for Sri Lankan adults. This prospective study was conducted at the National Hospital of Sri Lanka over a 6-month period, split into a development and a validation phase. During the development phase, estimated body weight of patients by doctors and nurses and patients themselves were noted and compared against their actual body weight. In addition, 13 anthropometric measurements were taken, which were used to validate 12 anthropometrics-based equations to estimate body weight described in literature previously. Two new gender specific regression models to estimate the body weight in the local population was also derived and validated. A total of 502 (males = 249) and 217 (males = 108) patients were recruited for the development and validation phases respectively. Both doctors and patients had comparable accuracy in predicting body weight (p>0.05). All anthropometric based equations were significantly correlated with actual body weight (correlation coefficients: 0.741-0.869), and the new equations derived from the local data performed similarly to the best performing equation identified from the literature during validation phase. However, even the best of these equations could not outperform patient/physician estimates. When the patient weight cannot be measured, an estimate by the patient or the doctor may be the best substitute.