Medical Coding and Clinical Documentation Accuracy Assessment for Health Information and Revenue Cycle TeamsThe note said one side. The claim said both. How far would that have travelled before anybody caught it?

Methodology: Thirty-eight original exercises across six formats: ten clinical-note sweeps, each a short note with six coding lines of which three are supported by the note and three are not, delivered as select-every-line-the-note-supports; ten single-choice exercises about what a record supports; seven true-or-false claims about documentation rules; four match-the-following exercises; four ordering exercises on the query, escalation, principal-diagnosis and support-check procedures; and three numeric estimates on a slider. One response instruction is declared for the whole instrument and it is a KNOWLEDGE instruction: which coding the documentation supports, never what the respondent would do or feel. Construct statement: this measures whether a person can read a clinical note and tell which coding the documentation actually supports, in two directions that are never netted, coding what the note does not support and missing what the note does support, together with knowing when a documentation query is raised instead of a guess; it does not measure knowledge of any code set, any code number, any national programme's rules, any payer's policy, clinical judgement, typing speed, or how a person codes under a production target, and it is not a coding credential. Scoring is two-sided coverage against a declared selection prior (C19). On every sweep the keyed set is the SUPPORTED class and the complement is what the note leaves unsupported; content.option_classes states the class of every line beside the key so the two can be checked against each other across the bank. Every line carries option_base_rate, the declared share of working coders expected to mark it. Two figures are computed from the sweeps and never combined: not coded beyond the note, the share of unsupported lines left unmarked, and nothing supported left out, the share of supported lines marked. Each is corrected against a coder marking every line at its declared prior, so zero on either figure is the ordinary coder, and each carries its own n, its own assumed omega printed unrounded and its own standard-error band. A figure is printed only when at least eight sweeps were answered and omega is at or above .70; otherwise a three-way placement is printed in the figure's place. The weakest sweep is printed beside the mean. The report is an escape ladder (D23): five downstream controls in a coding department, the coder's own second read, the internal coding audit sample, the claim edit before submission, the payer query or denial, and the external documentation audit, each carrying a declared multiplier of 1, 3, 8, 20 and 40 printed above the ladder. Every coding line, option, pair, ordering and estimate in the seed carries the rung of the first control whose job is to catch an error of that kind, and the respondent's errors are pinned to those rungs. The escape index is one minus the sum of the multipliers at the rungs the respondent's errors reached, divided by the sum over every answered exercise of its worst reachable rung; it is normalised against the form and never against the respondent's own error count, is 1.00 for a clean sitting, and is printed beside what a coder answering at the declared priors would score on the same form. The multipliers are a value judgement and not a measurement, and are printed so they can be disagreed with. The ladder prices where an error would have been detected, not whether anything went wrong, and the panel says so in its first sentence. The single-choice, true-or-false, matching, ordering and estimation exercises are corrected against the declared prior on every option, pair, position or value and feed four competency areas together with the sweeps: documented support, specificity and laterality, sequencing and the principal diagnosis, and query and escalation. An area with fewer than eight answered exercises or an omega under .70 carries a three-way placement and no number. No true-or-false prior in the file is .50/.50. Refusal rules: an unanswered exercise leaves the numerator, the denominator and the chance term together; an empty sitting scores exactly zero on every figure; a sitting with fewer than twenty-three of the thirty-eight exercises answered is not reported, and the refusal is printed where the ladder and the figures would have been. A careless-responding count is computed for the operator and never shown to the respondent as a judgement. Legal: no code number from any code set appears anywhere in this instrument, no code list is printed, and every exercise describes a coding decision in words. This instrument is not affiliated with, endorsed by or derived from the World Health Organization, any national coding authority, any professional association, any certifying body, any payer or any software vendor, and it names none of them except the one attribution in the sources below. It is jurisdiction-neutral: every exercise states the rule it applies as the guideline your organisation follows, and no exercise depends on one country's programme rules. It is not a coding credential, does not qualify anybody to code for submission, and audits no real record; every note is invented, and no real patient, clinician, hospital or payer is described. Sources drawn on: the World Health Organization's instruction manual to the International Statistical Classification of Diseases and Related Health Problems, for the conventions and the definition of the main condition (the classification and its name belong to the World Health Organization; no affiliation or endorsement is implied); the public definition of the principal diagnosis as the condition established after study to be chiefly responsible for the admission, as used in hospital statistical reporting since the Uniform Hospital Discharge Data Set (1985); the public compliant-query practice of listing clinical indicators and offering an open set of options, as described in professional guidance on documentation queries (2008 onward, revised 2013, 2019 and 2022); Cheng, Gilchrist, Robinson and Paul, The risk and consequences of clinical miscoding due to inadequate medical documentation (2009); Campbell, Campbell, Grimshaw and Walker, A systematic review of discharge coding accuracy (2001); Burns, Rigby, Mamidanna and others, Systematic review of discharge coding accuracy (2012), for the size and direction of coding error; O'Malley, Cook, Price and others, Measuring diagnoses (2005), for the sources of error along the documentation-to-code path; Dixon, Sanderson, Elliott and others, Assessment of the reproducibility of clinical coding in routinely collected hospital activity data (1998); Nouraei, Hudovsky, Frampton and others, A multidisciplinary audit of clinical coding accuracy (2015), for the internal audit as a control; Reason, Human Error (1990), and Hollnagel, Barriers and Accident Prevention (2004), for the layered-defence model behind the escape ladder; Macmillan and Creelman, Detection Theory: A User's Guide (2005), for the two-error framing; Gollwitzer and Sheeran, Implementation intentions and goal achievement, a meta-analysis (2006), for the if-then action; and Haladyna, Downing and Rodriguez (2002) for the item-writing rules. All exercises are original works written for this instrument. No commercial instrument's items or name are used or implied, and no code set's numbers are reproduced.