Health Care Data
Data quality and the numbers you must know
Data quality characteristics include accuracy, accessibility, comprehensiveness, consistency, currency, definition, granularity, precision, relevancy, and validity. If the same element is recorded differently in two systems, the issue is consistency; if it is simply wrong, it is accuracy.
Formulas to memorize:
- Average daily census (ADC) = total inpatient days ÷ days in period
- Bed occupancy = inpatient days ÷ (bed count × days in period) × 100
- Length of stay = discharge date − admission date (same-day admit and discharge = 1 day)
- Gross death rate = deaths ÷ discharges (including deaths) × 100; the net death rate excludes deaths within 48 hours of admission from both numerator and denominator
Data sets: UHDDS (inpatient; source of the principal diagnosis definition and uniform data elements), MDS for long-term care, OASIS for home health, HEDIS for managed care quality measures.
Terminologies: ICD-10-CM/PCS and CPT classify and drive payment; SNOMED CT captures clinical concepts; LOINC covers laboratory and observations; RxNorm and NDC cover medications; UMLS is the metathesaurus linking them. Interoperability standards include HL7 v2 messaging, CDA documents, and FHIR APIs. Data governance assigns ownership, stewardship, and quality monitoring.
Secondary uses of data include quality improvement, research, public health reporting, population health, and reimbursement. Case-mix index (CMI) is the average DRG relative weight for a population; a rising CMI increases expected payment and warrants documentation review.
Worked example: a 40-bed unit has 900 inpatient days in a 30-day month. ADC = 900 ÷ 30 = 30. Occupancy = 900 ÷ (40 × 30) = 75%. A patient admitted and discharged the same day contributes 1 day, not 0.
Traps: confusing ADC with occupancy; using discharges as the ADC denominator; treating HEDIS as a coding classification; assuming a higher CMI proves better documentation.