How Outcome Scores Get Misread in a Medical Record Review

A defense summary reads: "Function improved from 62 percent disability to 28 percent over nine months."

That sentence sounds like it came from an instrument on a patient. It came from two questionnaires the patient filled out in a waiting room, eleven visits apart, with no record of who handed them over or whether anyone explained the questions.

Both facts can be true at once. The scores are real evidence, and the sentence overstates what they are. That gap is where outcome instruments cause trouble in a medical record review, and it is a gap that opens because the evidence arrives as a number.

What a Score Actually Is

A patient-reported outcome measure is a questionnaire the patient answers, converted to a score by a fixed rule. The rule is standardized. The input is the patient's account of their own function.

The ones that turn up most often in injury and disability files:

• Oswestry Disability Index. A self-report questionnaire on how back pain affects daily activities, reported as a disability percentage.
• Neck Disability Index. The same design adapted to the cervical spine.
• DASH and QuickDASH. Self-report on arm, shoulder and hand function, the short form using fewer items than the full version.
• Numeric pain rating. A single self-reported number, usually zero to ten, recorded at almost every visit.
• PHQ-9 and GAD-7. Screening instruments for depressive and anxiety symptoms. The PHQ-9 is a nine-item instrument asking about the frequency of symptoms over the preceding two weeks, per CDC documentation of its use in national health surveys.
• PROMIS measures. A family of instruments developed through the National Institutes of Health, reported on their own scale rather than as a raw total.

Nothing in that list is a measurement of the patient. Every one of them is a measurement of what the patient reported. That distinction sits on the same axis as separating patient reports from provider findings, with one difference that matters: these reports arrive pre-converted into numbers, and numbers survive summarization better than attribution does.

Why Serial Scores Are Worth Finding

In most injury files these are the only repeated, numerically comparable measures of function that exist.

Consider what else is available. Range of motion is measured inconsistently and often estimated. Strength grading is coarse. Imaging findings describe anatomy, not function, and they do not change on the schedule symptoms do. Narrative descriptions of function are rich and not comparable across providers.

An instrument administered at intake, at three months and at discharge produces three values on one scale, collected by the same rule. That is a trend line, and it is the closest thing a typical chart holds to one. This is why they belong in a narrative summary with their dates attached, and why a review that skips them has left the most comparable evidence in the file unused.

The value and the risk come from the same property. Because they are comparable, they get compared. Because they are numeric, they get compared carelessly.

Six Ways the Comparison Breaks

Each of these produces a number that is real and a comparison that is not.

• Different instruments, same body part. An Oswestry score at intake and a different back-specific instrument at follow-up cannot be placed side by side. Both describe back function. Neither converts into the other.
• Different versions of the same instrument. Short forms and full versions use different item counts. A QuickDASH and a full DASH are related instruments, not interchangeable values, and a summary that lists both as DASH scores has built a trend out of two different things.
• Recall window drift. Instruments ask about a defined period, whether that is today, the past week or the past two weeks. A score describing the last two weeks and a pain rating describing this moment are answering different questions.
• Administration mode. The same questionnaire can be completed alone on paper, filled in on a tablet, or read aloud by staff. National survey documentation for the PHQ-9 describes it being administered by trained interviewers rather than self-completed, which shows the same instrument moves between modes. Charts almost never record which mode was used, and the review should say so rather than assume.
• Incomplete administration. Most of these instruments have scoring rules for missing items, and those rules have limits. A questionnaire with three sections left blank may still produce a printed score, and that score is not equivalent to a complete one.
• Transcription and scoring error. Someone tallies the items, and someone else types the total into the note. A digit lost between those steps produces a clean number that no longer matches the completed form. Where the underlying questionnaire is in the production, the score can be checked against it. Where it is not, the score is a report of a report.

When Is a Change Meaningful?

That question has a real answer, and it is not one a record review gets to give.

Outcome measurement research works with the idea of a smallest change that matters to a patient, often discussed as a minimal clinically important difference. The principle is that a difference has to exceed some threshold before it represents genuine change rather than measurement noise or ordinary day-to-day variation.

Two things follow for a reviewer. Those thresholds are specific to the instrument, and they also vary with the population studied, the condition and the method used to derive them, which is why published values for a single instrument are not uniform. And selecting a threshold and applying it to a particular claimant is a clinical determination.

So the review states the values and the dates. It does not characterize a change as significant, meaningful, or within normal variation. That belongs to the treating provider, the evaluating physician or a retained expert who can name the threshold they are using and defend it.

Ceiling, Floor and the Flat Line

An unchanged score is the most commonly overread value in the set.

A patient already answering at the most severe end of an instrument has nowhere further to go on it. Worsening function produces an identical score, because the instrument has run out of range. The same applies at the other end, where a patient near full function shows no improvement on the instrument while still reporting change.

A flat series can therefore mean stability, an instrument at its limit, a questionnaire completed without much attention, or a score copied forward from a prior visit into a template. The records sometimes distinguish these and often do not. Where they do not, a summary that reports no change has already interpreted.

Where the Instrument Came From

Not every questionnaire in a file is a validated instrument, and the record does not label the difference.

Alongside the recognized measures, injury files carry practice-designed intake forms, pain diagrams, symptom checklists and questionnaires supplied by an attorney or a vendor. Some produce numbers. Those numbers look like the others on the page.

The review's job is to name the source of each one. A published instrument administered on a date is a different record from a clinic's own form, and a form supplied by a party to the litigation is different again. None of the three is worthless. All three get read differently, and only the first supports comparison against published expectations at all.

Four Readers, One Requirement

Everyone downstream needs the same thing from these records, and each needs it for a different reason.

• IME and QME examiners. An examiner producing hands-on findings is often asked to reconcile them against the treating record's scores. Knowing those scores are self-report, and which instrument produced each one, is part of what medical record review before IME reports is for.
• Plaintiff firms. Serial scores can document persistence of symptoms in the patient's own terms across years, which narrative notes do less precisely. Presented as objective measurements, the same scores invite a cross-examination the file did not need.
• Defense counsel and carriers. An improving series is real evidence about reported function. Characterized as measured improvement, it overstates itself, and the overstatement is the part that gets challenged.
• Life care planners. A current functional score is a data point about reported status on a date. Projecting decades of care from a trend in self-report requires the clinician's judgment about what the trend represents.

Where the Review Stops

Three layers, kept apart.

What the records document. The instrument named, the date administered, the score recorded, and the completed form where it is in the production.

What a reviewer can identify. Instrument and version changes across the series, missing or partial administrations, undated scores, a value that does not match its form, a flat series, and the absence of any recorded administration mode.

What requires a qualified professional. Whether a change is clinically meaningful, whether a score is consistent with the clinical picture, whether an instrument was appropriate for this patient, and what any of it means for causation, impairment or future need.

Instrument selection and interpretation also depend on condition, specialty and the clinical context in which the questionnaire was given. Where that bears on a case, it is a question for the evaluating clinician rather than a line in a summary.

What to Request

Five things turn a scatter of numbers into a usable series.

• Every administration listed with the instrument named in full, including the version or short form.
• The completed questionnaires themselves, not only the scores transcribed into progress notes.
• Each score dated to its administration rather than to the visit it was discussed in.
• Instrument or version changes flagged wherever the series is not continuous.
• Partial administrations, undated scores and unrecorded administration mode identified as such.

Read the Number as a Sentence

Every one of these scores started as a person answering questions about their own life. The instrument turned those answers into a figure so they could be compared, which is useful, and the figure then stopped looking like an answer, which is the problem.

Put the sentence back. Name the instrument, date it, quote it, say who completed it, and leave the meaning to the people qualified to assign it. The series still does its work, and it stops claiming more than it holds.

LezDo TechMed supports examiners, planners and litigation teams through our medical record review services. We name, date, quote and flag. The clinical and legal conclusions stay with you and your experts.

Source Credit: The description of the PHQ-9 as a nine-item instrument asking about symptom frequency over the preceding two weeks, and its administration by trained interviewers in a national health survey, is from Centers for Disease Control and Prevention National Health and Nutrition Examination Survey documentation. PROMIS measures were developed through the National Institutes of Health. Instrument descriptions here are general; no scoring thresholds or clinically important difference values are asserted, because those are instrument-specific and population-dependent. The file described in this article is a hypothetical illustration, not a client matter. Company figures are LezDo TechMed's published figures. This article is general information for medico-legal and claims professionals, not legal or medical advice.