Patient-Reported Outcomes

What a patient-reported outcome is

A patient-reported outcome (PRO) is "any report of the status of a patient's health condition that comes directly from the patient without interpretation of the patient's response by a clinician or anyone else," according to the FDA definition quoted in the Cochrane Handbook [1]. The tool used to collect it is a patient-reported outcome measure (PROM), most often a self-report questionnaire [1]. Reports from a caregiver, parent or clinician are not PROMs; they are observer-reported or clinician-reported outcomes [1].

PROs are one of several ways to assess clinical outcomes in trials, alongside biomarkers, measures of illness such as stroke or heart attack, hospitalization and survival [1]. This page is education and not advice about any treatment.

Why they matter

Some things can only be reported by the person who feels them. The Handbook notes that PROs are essential when externally observable patient-important outcomes are rare or unavailable, and gives pain syndromes, fatigue, irritable bowel syndrome, sexual dysfunction, emotional function and adverse effects such as nausea and anxiety as examples where physiological measurements are limited [1].

PROs also differ from lab or device measures. Self-reported outcomes often correlate poorly with physiological or performance measures: in one example from the Handbook, differences in exercise capacity explained only 3% of the variability in breathing problems reported on a questionnaire in people with asthma [1]. A blood or device marker can improve without the person feeling better, and the reverse. See Biomarker Changes vs Meaningful Health Outcomes.

What makes a good questionnaire

Three properties are examined before a PROM is trusted in research [1]:

• Reliability: it gives consistent results and can distinguish between people despite measurement error.
• Validity: it measures what it is intended to measure, including content validity, judged with patients and clinicians.
• Responsiveness: it can tell apart people who stay the same, improve or get worse. A tool that detects change poorly can produce a false-negative result, in which a treatment helps how people feel but the instrument fails to detect it.

Scores also need a yardstick. The smallest change patients notice is called the minimal important difference (MID); for one respiratory questionnaire scored 1 to 7, 0.5 is a well-established MID [2]. A statistically significant change smaller than the MID may not matter to patients; see Effect Sizes and Confidence Intervals in Plain Language.

An example of a standardized system

PROMIS, developed with funding from several U.S. National Institutes of Health institutes starting in 2004, includes more than 300 measures of physical, mental and social health for adults, children and parents reporting on children [3]. Its measures include short forms (4 to 10 questions on one topic) and computer adaptive tests that choose each next question based on prior answers [3]. Scores use T-scores with an average of 50 and a standard deviation of 10 in the general population [3]. A shared metric lets studies be compared and combined.

Where PROs can mislead

• Expectation and bias. In placebo trials, the Cochrane review concluded it is difficult to distinguish patient-reported effects of placebo from biased reporting [7]. Trials without proper blinding or concealment overstated effects, especially for subjective outcomes [8]. See Randomization, Blinding and Placebo.
• Missing data. Answers that are missing or lost to drop-out can distort a result, which is why CONSORT-PRO asks trials to state how missing data were handled [6].
• Poor instrument. A questionnaire with weak validity or responsiveness can hide or invent effects [1].

Questions to ask when you see a PRO result

1. What exactly was asked, and by which questionnaire?
2. Were participants blinded to which treatment they received?
3. How big was the change compared with the MID?
4. How many people dropped out or skipped questions?
5. Does the result match any objective measure?