The Best Outcome Measure Is the One You Take Twice

Most therapy charts contain a graveyard of outcome measures: a thorough evaluation with four standardized scores, none of which is ever administered again. The instrument was chosen for the evaluation — comprehensive, impressive, diagnostic — and it is exactly those qualities that guarantee nobody has forty minutes to repeat it on a Tuesday between sessions. A measure taken once is a description of a person on a day. It cannot show change, and showing change is the entire job. So the selection question for a working SLP, OT, or PT is not "which outcome measure is best for this condition?" It is "which measure am I certain I will take again in a few weeks, under the same conditions, in the time a real caseload allows?" This article gives you three criteria for answering that question, a comparison of the main classes of measures against them, and a cadence that makes the repeat administration automatic instead of aspirational.

Payers do not reimburse therapy because the patient is impaired; they reimburse it because skilled treatment is producing change the record can demonstrate. Medicare’s documentation manual makes the mechanism explicit: the chart needs objective measurement, and the progress report — due at least once every 10 treatment days under Part B — must present objective evidence of progress toward the goals. A baseline score with no follow-up administration cannot do that work. Neither can a second administration of a different instrument, because two different rulers cannot measure one distance. The evidence payers and referrers can actually use is the same measure, administered the same way, at two points in time.

This is also why the evaluation battery is usually the wrong place to look for your outcome measure. Norm-referenced diagnostic assessments answer a different question — how does this person compare to peers? — and answering it well makes them long, expensive to administer, and often explicitly not designed for frequent re-administration. The outcome measure is a different tool with a different job: small, repeatable, and pointed at the function you are treating. Choose it separately, and choose it at the evaluation, because the baseline you fail to take on day one cannot be reconstructed later.

Score every candidate measure on three questions, in this order. First, administration time — not the time it takes once, but the cost you pay at every repeat. A measure you administer at baseline, at every progress report, and at discharge gets administered five or six times in a typical episode; a ten-minute difference per administration is an hour of clinical time over the episode, multiplied across the caseload. Second, sensitivity to change: can the score realistically move within your treatment window, for your population, in a way that exceeds the instrument’s own noise? A tool with a ceiling your patient will hit in three weeks, or one so coarse that a season of progress moves it one point, will produce a flat line that argues against your own treatment. Third, payer recognition: when the number appears in a progress report, can a reviewer who has never met the patient read it as functional progress? Measures anchored to observable daily function travel better in a chart than raw scores that need a manual to interpret.

Every instrument has measurement error, and researchers publish two thresholds that tell you how to read a score change against it. The minimal detectable change (MDC) is the smallest change larger than the instrument’s random error — movement below it may be noise, not progress. The minimal clinically important difference (MCID) is the smallest change patients experience as meaningful improvement. These are established per instrument and per population in the peer-reviewed literature, and they are the difference between "the score went up two points" and "the score improved beyond measurement error." Before adopting a measure, look up whether an MDC has been published for a population like yours; when you report change, report it against that threshold rather than celebrating movement the instrument itself cannot distinguish from noise. Where no threshold is published for your population, say what the change means in observed function instead of implying statistical certainty the literature does not offer.

Nearly every instrument you will consider falls into one of five classes, and the classes behave consistently on the three criteria even though individual instruments vary. Choose the class first, then the instrument inside it that has published measurement properties for your population.

The pattern in that table is the article’s argument in miniature: the classes that are cheap to repeat are the ones that show change, and the expensive one is the one designed for a different question. A defensible chart usually pairs two cheap classes — one standardized instrument for comparability, one individualized or performance-based measure for the functions this patient actually cares about — and leaves the diagnostic battery at the evaluation where it belongs.

Physical therapists have the most developed infrastructure for this decision. APTA maintains a searchable Tests and Measures library with summaries of individual instruments, and APTA Acute Care has published a clinical practice guideline recommending a core set of measures for assessing physical function in hospitalized adults — a useful model of the "small set, repeated often" philosophy even for outpatient settings. When a clinical practice guideline for your patient population names specific instruments, start there: guideline-named measures come with published measurement properties and instant recognizability.

Occupational therapists should anchor selection to the outcomes framework in AOTA’s Occupational Therapy Practice Framework: outcomes in OT are results in occupational performance and participation, not component skills alone. That argues for pairing one standardized instrument with an individualized, patient-specific measure in which the client names and rates the occupations that matter to them — the class of measure that is cheapest to repeat and speaks most directly to why the person came. A grip-strength number can support the story; the re-rated occupational performance score is the story.

Speech-language pathologists have the hardest version of the problem, because so much of the SLP toolbox is norm-referenced diagnostic assessment that resists frequent re-administration. Two workable paths: ASHA’s National Outcomes Measurement System (NOMS) rates function on seven-point Functional Communication Measures from evaluation to discharge, which gives episode-level change a nationally recognized vocabulary; and structured probe data — counted accuracy on defined targets under defined conditions — gives visit-level sensitivity that a seven-point scale cannot. The pairing covers both time horizons: probes move week to week, the FCM moves across the episode.

Selection fails in the calendar, not the catalog. The fix is to make re-measurement an appointment the schedule already contains rather than a task someone must remember. Under Medicare Part B the progress report is due at least every 10 treatment days and must contain objective evidence of progress — so the progress-report schedule is a re-measurement schedule the payer has already built for you. Other payers set their own intervals; the mechanic works the same way once you know the interval.

1. 01

   Take the baseline at the evaluation, and record the conditions

   Administer the chosen measure on day one and note how: position, equipment, instructions, assistance level, time of day if it matters. A repeat under different conditions measures the conditions, not the patient.

2. 02

   Name the re-measurement trigger in the plan

   Write the sentence at the evaluation: "Re-administer [measure] at each progress report and at discharge." Tie it to the reporting interval your payer already enforces so the deadline arrives with the paperwork that needs the number.

3. 03

   Repeat under the same conditions, inside a normal session

   If the repeat cannot fit inside a treatment session, the measure failed criterion one — swap it for a shorter one now, at the first repeat, rather than quietly abandoning measurement for the rest of the episode.

4. 04

   Interpret against the published threshold

   Compare the change to the instrument’s published MDC for a comparable population when one exists. Change beyond it is progress you can assert; change inside it is a reason to keep treating and re-measure at the next interval, stated honestly.

5. 05

   Put the comparison next to the goal, in one sentence

   The progress report line a reviewer can use reads: baseline score, current score, threshold, and what the change means in daily function. Two administrations of one instrument, one sentence of interpretation.

6. 06

   At discharge, close the loop with the same instrument

   The discharge administration turns the episode into a complete data story — baseline, trajectory, endpoint — that supports the discharge decision and gives the referrer a result they can read in ten seconds.

Payer recognition does not mean there is a national list of approved instruments — for most payers there is not, and Medicare’s manual asks for objective measurement without prescribing a catalog. Some Medicare contractors and commercial plans do name preferred instruments in their local coverage policies and medical-necessity guidelines, and those documents, not habit, are the authority for a specific payer. Recognition in practice is humbler: the measure is named, the score has a direction a reader can follow, the same measure appears at least twice, and the change is translated into function. A chart can fail all four with a famous instrument and pass all four with structured probe data, so spend your care on the repetition and the translation rather than on hunting for a magic instrument.