To investigate, in elderly individuals registered at a secondary outpatient clinic, the prevalence of frailty and pre-frailty and to identify the discriminatory power of anthropometric measurements and nutritional risk in identifying these conditions.

Cross-sectional study with data extracted from medical records.

Elderly patients (60+ years) from a geriatric outpatient clinic, located in the southeast area of São Paulo, Brazil.

Frailty was assessed using five criteria proposed by Fried et al (2001), with some modifications. Nutritional risk was identified using Mini Nutritional Assessment (MNA). Body weight and body height were measured and used to calculate the body mass index (BMI). The discriminatory power of these parameters for the identification of frailty was determined by Receiver Operating Characteristics (ROC) curves

The final sample was composed of 254 patients, from which 31.1% were identified as frail and 53.5% as prefrail. The MNA indicated that 3.1% were malnourished and 35.4% were at risk of malnutrition. The BMI values 39.4% as overweight/obese and 19.9% as undernourished. As just the MNA revealed differences for frailty classification, only this parameter was investigated by ROC curve. The discriminatory power of the MNA for frailty presented a best cut-off point of ≤23.0 and the AUC was 0.812 (sensitivity=55.7; specificity=94.9), with a youden index of 0.5057 (95%CI= 0.3146-0.5946). MNA did not present sufficient discriminatory power to detect pre-frailty.

The MNA was capable of indicating frailty, but not pre-frailty in this sample. BMI did not display significant predictive power for frailty or prefrailty.