Outcome measures for adult critical care: a systematic review.

OBJECTIVES: 1. To identify generic and disease specific measures of impairment, functional status and health-related quality of life that have been used in adult critical care (intensive and high-dependency care) survivors. 2. To review the validity, reliability and responsiveness of the measures in adult critical care survivors. 3. To consider the implications for future policy and to make recommendations for further methodological research. 4. To review what is currently known of the outcome of adult critical care. METHODS: DATA SOURCES: Searches of electronic databases (MEDLINE, EMBASE, CINAHL, PsycLIT, The Cochrane Library and SIGLE) from 1970 to August 1998. Manual searches of five journals (1985-98) not indexed in electronic databases and relevant conference proceedings (1993-98). Reference lists of six existing reviews, plus snowballing from reference lists of all relevant articles identified. STUDY SELECTION: Randomised trials, non-randomised trials (cohort studies) and case series that included data on outcomes after discharge from adult (16 years and over) critical care. DATA EXTRACTION AND SYNTHESIS: If reported, the following data were extracted from each paper: patient characteristics (age, gender, severity of illness, diagnostic category) number of patients eligible for study, follow-up period, number of deaths before follow-up, number and proportion of survivors included in follow-up method of presentation of outcome data - proportion normal as defined by reference values, or aggregate value (e.g. mean or median), or aggregate values plus an indication of variance (e.g. standard deviation or inter-quartile range). Evidence for three measurement properties was sought for each outcome measure that had been used in at least two studies - their validity, reliability and responsiveness in adult critical care. If the authors did not report these aspects explicitly, an attempt was made to use the data provided to provide these measurement properties. For measures that were used in at least ten studies, information on actual reported outcomes were also extracted. RESULTS: MEASURES USED IN CRITICAL CARE: Measures of impairment were largely confined to the respiratory system so are almost certainly not appropriate for many critical care survivors. They can be categorised as respiratory volumes (e.g. vital capacity), gas flow within the respiratory system (e.g. forced expiratory volume in 1 second (FEV1)), pulmonary diffusing capacity (e.g. carbon monoxide diffusing capacity) and visualisation of the upper airway (e.g. bronchoscopy). Multiple tests are often performed. Eight measures of physical functional status were used, five generic and three disease-specific. The most frequently used generic measures were multi-item scales. Two single-item global measures attempted to capture a person's overall activity level or functional status. Five multi-item measures of mental functional status were used, four generic and one specific to trauma patients. The generic measures were either confined to assessing depressive symptoms or also encompassed a measure of anxiety. Measures of neuropsychological functioning relate to a person's cognition, attention, ability to process information and memory. Apart from one single-item measure, which focused on communication level, six multi-item measures were used with critical care survivors. Such measures are particularly appropriate for use with survivors of head injury or other neurological insult and, in that sense, they are disease-specific rather than generic measures. Single item measures of recovery were frequently used but researchers often invented their own, so there was little consistency in the wording. These measures had five principal foci - return to work, return to own home, degree of recovery, productivity and chronic health status. One multi-item scale was also used. (ABSTRACT TRUNCATED)