Assessment, Measurement and Outcome Evaluation
This domain tests whether you can choose a measure, administer it consistently, and interpret change against measurement error rather than against intuition. Expect psychometric calculations and questions about which outcome measure fits a given population.
Reliability, validity, responsiveness
- Reliability is consistency; intraclass correlation coefficients of 0.75–0.90 are good and above 0.90 excellent.
- Validity is whether the tool measures what it claims: content, criterion and construct validity are the main types.
- Responsiveness is the ability to detect real change over time; the minimal clinically important difference is the smallest change a patient would notice as worthwhile.
- Standard error of measurement: SEM = SD × square root of (1 − ICC).
- Minimal detectable change at 95% confidence: MDC95 = 1.96 × SEM × square root of 2.
Goniometry and manual muscle testing
Place the fulcrum over the joint axis, the stationary arm along the proximal segment and the moving arm along the distal segment; intrarater reliability is generally better than interrater. Oxford grading: 0 no contraction, 1 trace, 2 full range without gravity, 3 full range against gravity, 4 against gravity plus some resistance, 5 normal. Handheld dynamometry gives a more objective value than a grade.
Choosing an outcome measure
WOMAC for hip and knee osteoarthritis; DASH or QuickDASH for upper limb; LEFS for lower limb function; Berg Balance Scale for balance (0–56); Timed Up and Go, where over 13.5 seconds suggests falls risk; 10-metre walk test for gait speed; Barthel Index for activities of daily living; condition-specific measures such as the Neck Disability Index or Oswestry Disability Index.
Worked example
A scale has an ICC of 0.90 and the standard deviation of scores in the sample is 10 points. SEM = 10 × square root of 0.10 ≈ 3.16. MDC95 = 1.96 × 3.16 × 1.414 ≈ 8.8 points. A patient improves by 6 points: that is smaller than the measurement error, so you cannot claim a real change, even if the change looks encouraging in the notes.
Traps that cost marks
- Reporting a statistically significant p-value as if it proved a clinically meaningful change.
- Using 'reliable' and 'valid' as synonyms.
- Comparing scores across raters when only intrarater reliability was established.