Heslin · Medical care 2017 · Repeated cross-sectional time-trend analysis · n=?

Trends in Opioid-related Inpatient Stays Shifted After the US Transitioned to ICD-10-CM Diagnosis Coding in 2015.

Cited 99 times in the scientific literature.

Level 4 - case-series / case-control

Repeated cross-sectional surveillance study across a coding transition without an unexposed control group

PubMed 28930890 · doi:10.1097/MLR.0000000000000805 · record verified 2026-08-28

What was done

The authors evaluated how the October 2015 US transition from ICD-9-CM (~14,000 codes) to ICD-10-CM (68,000 codes) affected longitudinal trends in opioid-related hospitalizations. They analyzed quarterly inpatient discharge data from the Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases across 14 states (representing 26% of US acute care inpatient discharges) covering 2015 and the first three quarters of 2016, comparing changes before, during (fourth quarter of 2015), and after the transition.

What was found

Stays with any opioid-related diagnosis increased by 14.1% during the coding transition quarter. This contrasted with an average quarterly increase of 5.0% before the transition and 3.5% after the transition. In stratified analyses, stays involving adverse effects of opioids in therapeutic use increased by 63.2% during the transition, whereas stays involving opioid abuse diagnoses decreased by 21.1% and opioid poisoning diagnoses decreased by 12.4%.

Why it matters

The switch to ICD-10-CM introduced a major artificial discontinuity in opioid surveillance data rather than a true epidemiological shift. Researchers and public health analysts tracking longitudinal opioid trends across 2015 must account for coding-induced baseline shifts to avoid misinterpreting trends.

Limits

The dataset reflects only 14 states (26% of US acute care discharges) and may not fully generalize nationally. The abstract does not report absolute sample sizes (n of discharges), confidence intervals, or formal statistical interrupted time series modeling, and coding validity was not verified against clinical chart reviews.

Cited by