Reliability of scored patient-generated subjective global assessment in determining nutritional status of HIV infected adults attending TASO Mulago Clinic

Abstract

Background: Early identification of risk for malnutrition in people living with HIV (PLHIV) is needed so as to determine nutritional deficits and appropriate interventions. Malnutrition in many PLHIV in Uganda is frequently not assessed and is left untreated due to lack of reliable tools. Thus, reliable assessment tools are needed to accurately assess PLHIV at risk of malnutrition. PG-SGA is a potential tool that may serve this purpose in Uganda. It is an accepted tool for assessing nutritional status in cancer patients in developed countries. However, no studies have determined its reliability for assessing nutritional status in PLHIV. Objective: Determine reliability of PG-SGA in determining nutritional status of HIV-infected adults (18 to 67 years) who are not on antiretroviral therapy (ART). Method: A cross sectional study of HIV-infected adults who are not on ART in TASO, Mulago.Reliability (specificity and sensitivity) of PG-SGA in assessing risk for undernutrition was determined by comparing PG-SGA scores to BMI (a commonly used nutrition tool) using ROC curve. Reliability of PG-SGA in predicting immunosuppression was tested against CD4 counts. Results: Compared to BMI, PG-SGA correctly identified 69.2% underweight (BMI <18.5) and 57.1% normal participants (BMI ≥18.5), which shows PG-SGA scores can identify the risk of malnutrition among PLHIV. PG-SGA scores were negatively associated with BMI (r=-0.311, P=0.000) which confirms the ability of the tool to identify malnourished individuals. There was also a weak but significant association between PG-SGA scores with CD4 count (r=-0.221, P=0.002) indicating PG-SGA scores can predict risk to immunosuppression. PG-SGA scores were weakly but positively associated with infections (r =0.343, P=0.000) further confirming ability of the tool to predict immunosuppression. All PG-SGA Global Assessment categories significantly contributed to PG-SGA score and thus should be considered in PG-SGA. Of all socio-demographic factors, only years of education were associated with nutritional status indicated by PG-SGA scores (r = - 0.138, P=0.043). Conclusion: Compared to BMI, PG-SGA did not prove a reliable tool for assessing nutritional status of HIV-infected adults. However, the observed lack of reliability of PG-SGA does not invalidate it as rapid nutrition screening tool given BMI only considers weight and height. PG-SGA is thus useful in identifying individuals at risk for underweight and predicting immunosuppression especially where there is no objective equipment to determine nutritional status and CD4.