Journal of Rehabilitation Medicine 51-6 | Page 5

Comprehensive assessment for hospital-acquired deconditioning in acute hospitals Study design A systematic evidence scan was undertaken, guided by relevant methods and reporting frameworks (16, 18, 19). Inclusion criteria Whilst the literature search targeted assessment instruments designed for, or tested on, older people (considered to be most at risk of HAD), instruments relevant to any acute hospital inpatient 18 years or older were also considered (2, 4, 12, 15). Literature was included if: • it reported on the development, psychometric testing and/ or clinical utility of assessment instruments (12, 15, 19, 20) that assessed the performance of any body system relevant to adult HAD; • the instrument was applicable to change in performance of any body system occurring during acute hospital admissions (1, 2, 8–10, 19); • the instrument used P-i-T or recent (last 2–3 days), assessment time-frames (9); and • the instrument could feasibly be administered at least twice during an acute hospital stay (5, 8, 9, 15, 19). Search strategy A Population (P) – Intervention (I) – Time (T) framework was applied: P = any acute hospital inpatient but in particular, older people; I = assessment instruments relevant to detecting any element of incipient HAD; and T = time-period of acute hospital admissions (i.e. instruments with items appropriate for repeated short-term application). The expanded search terms and syntax for each database are reported in Table SI 1 . Library databases (MEDLINE, CINAHL, PubMed, PsychInfo and Google Scho- lar) were searched from database inception until January 2018. Exclusion criteria Literature was excluded if it: • reported on condition- or disease-specific assessments of body systems performance (e.g. cancer, trauma, stroke and other neurological conditions, chronic respiratory conditions or cardiac conditions). These conditions have predictable management pathways, specific diagnostic codes, condition- specific assessment items and time-frames of measurement, and predictable functional changes; • described screening (one-off) measures not designed to detect change (19, 20); • did not describe the assessment instrument; • included assessment items with lengthy reflective time- periods (2, 4, 12, 15); • did not provide information on psychometric properties or utility; • was unavailable in full text or English language; • reported conference abstracts or posters, described protocols or findings of experimental studies, or presented expert opi- nion or position papers. Experimental studies were excluded because it was anticipated that defensible assessment/out- come measures applied in experimental studies would have been chosen on their published psychometric properties. These assessments would therefore be detected in our search. http://www.medicaljournals.se/jrm/content/?doi=10.2340/16501977-2546 1 399 Moreover, whilst experimental studies may have provided information on sensitivity and minimal clinical significant change, this information would be generalizable only to the study population. Literature identification process Potentially-relevant articles were considered collaboratively by author-pairs (using title and abstract). They then collaboratively assessed full texts of potentially-relevant papers against inclu- sion and exclusion criteria. If more than one paper described psychometric testing of an assessment instrument, the earliest paper was included, and the number of subsequent psychometric testing papers was recorded. Disagreements were arbitrated by a third author. Hierarchy Study design was assigned using the Australian National Health and Medical Research Council hierarchy (22). Critical evaluation Items with measurement time-periods longer than 2–3 days were excluded, because they were not feasible for repeated adminis- tration during an acute inpatient stay. To evaluate psychometric properties and clinical utility, the validated iCAHE Ready Reckoner was applied by author-pairs (17). The iCAHE Ready Reckoner seeks evidence from the developmental literature that an assessment instrument demonstrates quality elements, but does not record the statistics: • validity (face, content, comparison and construct validity, sensitivity testing and factor analysis); • reliability (inter- and intra-tester, test-retest and/or internal consistency assessment); and • clinical utility (reported as relevance, efficiency and appli- cability): • item wording and simplicity of instructions: a score of 1 indicated items relevant for minimum reading age (23); • fewer than 20 items (scored as 1): shorter instruments are more likely to attract accurate responses than longer instruments (minimize response burden) (24). The number of instrument items was recorded; • level of calculation difficulty: this scored 1 if item scores could be calculated manually (as opposed to complex calculations); • administration time: this scored 1 for assessments admi- nistered in less than 15 min (24). This information was either extracted from the literature, or established by the researchers. The mean administration time was recorded; • normative values or cut-off-scores: if available, these items each scored 1; • target relevance: scored as 1 if relevant to older Australi- ans in unplanned admissions to acute hospital beds); and • no cost; or formal registration requirements: these items each scored 1. The iCAHE Ready Reckoner does not apply a specific thres- hold to detect instrument quality. Higher total scores indicate better quality instruments. Item collation All items in the included assessment instruments were extracted, the measurement purpose was identified, items were cross- J Rehabil Med 51, 2019