The Nurses' Observation Scale for Geriatric Patients (NOSGER) is a rating scale for use in geriatric patients that can be applied by nurses or other caregivers. It deals with the daily behavior of elderly patients and measures impairment in six areas (dimensions): memory; instrumental activities of daily living (IADL); (basic) activities of daily living (ADL); mood; social behavior; and disturbing behavior. Objectivity, stability, construct validity, and acceptance of the scale have been established in previous studies using an earlier version of the NOSGER. The present validation study considered 50 healthy old subjects, 25 patients with mild dementia, 25 patients with advanced (mostly moderate according to DSM-III-R criteria) dementia, and 25 elderly patients with depression. The NOSGER was completed by relatives in the case of subjects living in their own homes and by nurses or other caregivers for institutionalized subjects. In addition to the NOSGER, selected tests of concentration, memory, and performance were applied as outside criteria.Interrater reliability (objectivity) was estimated by variance component analysis. Values between rtt = .68 and rtt = .89 (all p < .001) were found for the six NOSGER dimensions, the values being higher for the cognitive dimensions (memory, IADL, ADL) than for the noncognitive ones (mood, social behavior, disturbing behavior). Retest realibility (stability), which was calculated via rank order correlations, was somewhat higher for the cognitive NOSGER dimensions (memory rs = .91,IADLrs = .92, ADLrs = .88; p < .001) than for the noncognitive ones (mood rs = .85, social behavior rs = .87, disturbing behavior rs = .84; p < .001). All these values satisfy the level of rtt ≥ .80 required in accordance with psychometric standards. The concurrent validity of the NOSGER dimensions was assessed using correlations with external criteria with which similarity of content was expected. The NOSGER dimensions memory, IADL, ADL, and social behavior were found to correlate closely with external criteria of similar content, whereas no satisfactory concurrent validities were found for the dimensions mood or disturbing behavior. The NOSGER dimensions were also correlated with a number of unrelated external criteria so as to reveal any discordances. For the dimensions memory, IADL, ADL, and social behavior, no clear-cut discriminant validities were found. This suggests that these four dimensions may function as parameters not just of different areas of behavior, but also of a general factor that might be described as “cognitive intactness.” As a further aspect of construct validity, significant differences (all p < .001) between the four groups of subjects were found in five of the six NOSGER dimensions (memory, IADL, ADL, mood, social behavior): The healthy subjects differed significantly from all three patient groups in five of the six dimensions; the moderately demented group differed from the depressed group in four of the six dimensions and from the mildly demented group in two of the six dimensions; and the mildly demented group differed significantly from the depressed group in terms of mood (significance levels are after application of the Bonferroni correction). Significant group differences (p generally < .001) were also found for most of the objective performance tests used (data not presented).
The BETA® nursing measure: Calibrating construct validity with Rasch analyses*
