Outcome Measures and PROMs

What a PROM is. A patient-reported outcome measure (PROM) is a standardised, validated questionnaire that the patient completes without clinician interpretation. It captures the patient's own perspective on health status, symptoms, function and quality of life.

Why PROMs matter. Surgeon assessment may not match the patient's experience, and pain, function and satisfaction cannot be objectively measured. A PROM quantifies these subjective outcomes and captures what matters to patients: pain, daily activities and quality of life. PROMs are essential in clinical trials to demonstrate treatment efficacy, and they are also used for registry benchmarking and value-based care (see Clinical Application and Relevance).

PROMs and clinician measures. Clinician measures such as range of motion and strength are important, but they may not correlate with patient satisfaction. Best practice is to use both PROMs and objective measures.

Generic against specific. Generic PROMs assess overall health status in any condition and allow comparison between diseases, populations and population norms, but they are less sensitive to specific joint pathology. Joint-specific PROMs are highly sensitive to pathology in a single joint but cannot be compared across different joints or with the general population. Use both when possible, to capture the joint and overall health together.

View larger

Generic PROMs

Region-, Joint- and Disease-Specific PROMs

Choosing Between PROMs

The most common exam error is treating all PROMs as interchangeable. The table contrasts the major instrument types so that a choice can be justified under viva pressure.

Three properties decide whether a PROM can be trusted: validity, reliability and responsiveness. A measure with poor properties produces unreliable conclusions.

Validity asks whether the measure assesses what it claims to assess.

Reliability asks whether the measure gives consistent results when the condition is stable.

Responsiveness asks whether the measure can detect clinically meaningful change over time. A floor effect is present when a high proportion of patients, over 15%, score at the minimum, the worst possible score, and the measure cannot detect worsening in them. A ceiling effect is the same at the maximum, the best possible score, and the measure cannot detect improvement.

Quantifying responsiveness. Responsiveness is expressed as the standardised response mean (SRM) or effect size:

• SRM greater than 0.8 - large responsiveness (good)
• SRM 0.5 to 0.8 - moderate responsiveness
• SRM less than 0.5 - small responsiveness (may miss change)

Definition. The MCID is the smallest change in PROM score that patients perceive as beneficial and that would mandate a change in management. Its purpose is to distinguish a statistically significant change from a clinically meaningful one: a change with p less than 0.05 may not reach the MCID, and then it may not matter to patients.

View larger

Deriving it. There are two methods:

1. Anchor-based - compare the PROM change with an external anchor, a patient global assessment such as "Compared to before surgery, how would you rate your improvement: much better, better, same, worse?" The MCID is the mean change in the "better" group.
2. Distribution-based - use a statistical threshold, either 0.5 × the standard deviation or the standard error of measurement. This is less clinically intuitive than the anchor-based method.

Applying it. If the mean improvement is 8 points and the MCID is 10, the improvement is statistically significant but not clinically meaningful. If the 95% CI runs from 12 to 18 points against an MCID of 10, the entire interval exceeds the MCID and the improvement is clinically meaningful. An interval that crosses the MCID leaves the clinical significance uncertain, as Scenario 2 below works through. Always compare treatment effects with the MCID, not just with p-values.

Real before important. A change score must clear two separate hurdles. The minimal detectable change (MDC) is the smallest change that exceeds measurement error, so that it is real rather than noise; the MCID is the smallest change patients value. They answer different questions and must not be conflated.

Classical and modern instruments. Legacy PROMs use a fixed list of items, each scored equally (classical test theory). The modern direction is item-response theory (IRT) instruments, of which PROMIS (Patient-Reported Outcomes Measurement Information System) is the leading example, increasingly used by US registries.

What PROMIS is. A set of NIH-funded item banks, such as Physical Function and Pain Interference, calibrated by item-response theory so that each question has a known difficulty. Scores are reported on a T-score metric, mean 50 and standard deviation 10, anchored to the US general population, so a score of 40 is one SD worse than average. This common metric lets different item sets from the same bank be compared directly.

Computer-adaptive testing (CAT). Because every item is calibrated, a computer-adaptive test can choose the next question from the previous answer, homing in on the patient's level in typically 4-8 items instead of 30. CAT reduces respondent burden and floor and ceiling effects while maintaining precision. Where adaptive delivery is not feasible, PROMIS can also be given as fixed short forms.

The hierarchy. Outcome measurement rests on what you measure, how you measure it, and how you interpret it. The WHO ICF framework (body structure and function, activity, participation) is a useful map: PROMs predominantly capture the activity and participation levels, while clinician measures (range of motion, strength, radiographs) capture body structure and function.

Types of outcome. Outcome measures fall into four types:

• Patient-reported outcome measures (PROMs) - the patient's own rating of symptoms, function and quality of life, with no clinician interpretation
• Clinician-reported outcomes (ClinROs) - examiner-derived, such as range of motion, Constant strength or neurological grade
• Performance outcomes (PerfOs) - observed task performance, such as the Timed Up-and-Go or the six-minute walk
• Composite scores - blend domains, as the Constant-Murley combines patient pain with examiner-measured strength and range; this improves breadth but can obscure which domain drives a change

Anchoring interpretation. Interpretation rests on four anchoring concepts. The MCID, the smallest change a patient perceives as worthwhile, has its own section below, and floor and ceiling effects, which distort responsiveness when too many patients cluster at the extremes, are covered under Measurement Properties. Two further anchors are defined here:

• PASS (Patient Acceptable Symptom State) - the post-treatment score above which a patient considers their state satisfactory. It is increasingly preferred to MCID because it reports an attainable end state rather than a change.
• SCB (Substantial Clinical Benefit) - a higher threshold than MCID, denoting a large, clearly meaningful improvement.

What a good study does. It pre-specifies a single primary PROM, justifies it on measurement properties, and reports both mean change against the MCID and the proportion of patients reaching MCID or PASS.

Registries. The major arthroplasty registries (NJR, AJRR, AOANJRR, SHAR) collect PROMs at a pre-operative baseline and at post-operative follow-up, at 1 year and 5 years. That allows benchmarking of performance and quality improvement.

Value-based care. Payers increasingly link reimbursement to PROMs. Demonstrating patient-reported improvement justifies procedures, and PROMs are essential for value-based contracts. Linking them to payment has risks of its own, set out under Controversies below.

PROM collection has shifted from research tool to routine quality infrastructure worldwide, but the chosen instruments, mandate and uptake vary by region.

Global epidemiology and uptake. Joint arthroplasty is among the most-studied PROM settings: large registry programmes consistently show that the majority of hip and knee replacement patients achieve improvements exceeding the MCID at one year, while a meaningful minority (commonly cited around one in five for knees) report being unsatisfied - a gap PROMs make visible that complication rates alone miss. Uptake is high in publicly funded, registry-linked systems and far patchier where collection is voluntary or unfunded.

Guidance and Registry Programmes Side by Side

Methodological standards. The COSMIN initiative (Mokkink and colleagues) provides the most widely cited international consensus framework for selecting and appraising PROMs, complementing earlier quality criteria. ICHOM standard sets push toward globally comparable outcome reporting.

High- versus limited-resource practice variation. In well-resourced, registry-linked systems, electronic PROM (ePROM) capture is increasingly embedded in routine care, enabling case-mix-adjusted benchmarking. In limited-resource settings, barriers include literacy and language (validated translations are not universal), lack of electronic infrastructure, staffing for follow-up, and the cost of licensed instruments - so brief, free, culturally validated tools (and pragmatic VAS/EQ-5D use) are favoured. The principle that statistical significance does not equal clinical significance, and that MCID/PASS should anchor interpretation, applies in every setting regardless of resource level.

PROM science is evolving, and several issues remain genuinely unsettled. They make useful "areas of debate" answers in a viva.

MCID is not a single number. The same PROM yields different MCIDs depending on whether an anchor-based or distribution-based method is used, the anchor question, baseline severity and follow-up timing. Quoting "the MCID" as if it were fixed is a recognised pitfall: always state the method and population.

MCID versus PASS. A patient can exceed the MCID yet remain symptomatic and dissatisfied, and many groups now favour PASS, or the proportion reaching a "good outcome", as more patient-relevant than mean change.

Ceiling effects in legacy scores. Widely used scores (Constant, Harris Hip, some Oxford items) show marked ceiling effects in well-functioning patients. That masks further improvement and biases comparisons of already good results.

Response shift and missing data. Patients recalibrate their internal standard for "good" over time (response shift), which complicates before-and-after comparisons. Differential loss to follow-up, with sicker patients dropping out, inflates apparent improvement, and complete-case analysis is a common source of bias.

Linking PROMs to payment. Using PROMs for reimbursement or surgeon-level ranking risks gaming, risk-aversion (avoiding complex patients) and inadequate case-mix adjustment. These are the reasons several systems publish PROMs for benchmarking rather than for direct pay-for-performance.