Evaluating photographs as a replacement for the in-person physical examination of the scored Patient Generated Subjective Global Assessment in elderly hospital patients

Background: Undernourished patients discharged from the hospital require follow-up; however, attendance at return visits is low. Teleconsultations may allow remote follow-up of undernourished patients; however, no valid method to remotely perform physical examination, a critical component of assessing nutritional status, exists. Objective: This study aims to compare agreement between photographs taken by trained dietitians and in-person physical examinations conducted by trained dietitians to rate the overall physical examination section of the scored Patient Generated Subjective Global Assessment (PG-SGA). Design: Nested cross-sectional study. Participants/setting: Adults aged ≥60 years, admitted to the general medicine unit at Flinders Medical Centre between March 2015 and March 2016, were eligible. All components of the PG-SGA and photographs of muscle and fat sites were collected from 192 participants either in the hospital or at their place of residence after discharge. Main outcome measures: Validity of photograph-based physical examination was determined by collecting photographic and PG-SGA data from each participant at one encounter by trained dietitians. A dietitian blinded to data collection later assessed de-identified photographs on a computer. Statistical analyses performed: Percentage agreement, weighted kappa agreement, sensitivity, and specificity between the photographs and in-person physical examinations were calculated. All data collected were included in the analysis. Results: Overall, the photograph-based physical examination rating achieved a percentage agreement of 75.8% against the in-person assessment, with a weighted kappa agreement of 0.526 (95% CI: 0.416, 0.637; P<0.05) and a sensitivity-specificity pair of 66.9% (95% CI: 57.8%, 75.0%) and 92.4% (95% CI: 82.5%, 97.2%). Conclusions: Photograph-based physical examination by trained dietitians achieved a nearly acceptable percentage agreement, moderate weighted kappa, and fair sensitivity-specificity pair. Methodological refinement before field testing with other personnel may improve the agreement and accuracy of photograph-based physical examination.