Sensitivity and specificity of International Classification of Diseases algorithms (ICD-9 and ICD-10) used to identify opioid-related overdose cases: A systematic review and an example of estimation using Bayesian latent class models in the absence of gold standards
Bibliographic record
Abstract
OBJECTIVES: This study aimed to summarize validity estimates of International Classification of Diseases (ICD) codes in identifying opioid overdose (OOD) among patient data from emergency rooms, emergency medical services, inpatient, outpatient, administrative, medical claims, and mortality, and estimate the sensitivity and specificity of the algorithms in the absence of a perfect reference standard. METHODS: We systematically reviewed studies published before December 8, 2023, and identified with Medline and Embase. Studies reporting sufficient details to recreate a 2 × 2 table comparing the ICD algorithms to a reference standard in diagnosing OOD-related events were included. We used Bayesian latent class models (BLCM) to estimate the posterior sensitivity and specificity distributions of five ICD-10 algorithms and of the imperfect coroner's report review (CRR) in detecting prescription opioid-related deaths (POD) using one included study. RESULTS: Of a total of 1990 studies reviewed, three were included. The reported sensitivity estimates of ICD algorithms for OOD were low (range from 25.0% to 56.8%) for ICD-9 in diagnosing non-fatal OOD-related events and moderate (72% to 89%) for ICD-10 in diagnosing POD. The last included study used ICD-9 for non-fatal and fatal and ICD-10 for fatal OOD-related events and showed high sensitivity (i.e. above 97%). The specificity estimates of ICD algorithms were good to excellent in the three included studies. The misclassification-adjusted ICD-10 algorithm sensitivity estimates for POD from BLCM were consistently higher than reported sensitivity estimates that assumed CRR was perfect. CONCLUSION: Evidence on the performance of ICD algorithms in detecting OOD events is scarce, and the absence of bias correction for imperfect tests leads to an underestimation of the sensitivity of ICD code estimates.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.170 | 0.433 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.007 | 0.011 |
| Bibliometrics | 0.027 | 0.016 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.005 | 0.005 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".