Prospective evaluation of the Sedation-Agitation Scale for adult critically ill patients

Abstract

Objective: Subjective scales to assess agitation and sedation in adult intensive care unit (ICU) patients have rarely been tested for validity or reliability. We revised and prospectively tested the Sedation-Agitation Scale (SAS) for interrater reliability and compared it with the Ramsay scale and the Harris scale to test construct validity.

Design: A convenience sample of ICU patients was simultaneously and independently examined by pairs of trained evaluators by using the revised SAS, Ramsay, and Harris Scales.

Setting: Multidisciplinary 34-bed ICU in a nonuniversity, academic medical center.

Patients: Forty-five ICU patients (surgical and medical) were examined a total of 69 times by evaluator pairs.

Measurements and main results: The mean patient age was 63.2 yrs, 36% were female, and 71% were intubated. When classified by using SAS, 45% were anxious or agitated (SAS 5 to 7), 26% were calm (SAS 4), and 29% were sedated (SAS 1 to 3). Interrater correlation was high for SAS (r2 = .83; p < .001) and the weighted kappa score for interrater agreement was 0.92 (p < .001). Of 41 assessments scored as Ramsay 1, 49% scored SAS 6, 41% were SAS 5, 5% were SAS 4, and 2% each were SAS 3 or 7. SAS was highly correlated with the Ramsay (r2 = .83; p < .001) and Harris (r2 = .86; p < .001) scales.

Conclusions: SAS is both reliable (high interrater agreement) and valid (high correlation with the Harris and Ramsay scales) in assessing agitation and sedation in adult ICU patients. SAS provides additional information by stratifying agitation into three categories (compared with one for the Ramsay scale) without sacrificing validity or reliability.