400 New ICD code-based case definition correctly identifies ICI-IA from administrative health data
Bibliographic record
Abstract
Background Immune checkpoint inhibitors (ICIs) treat numerous malignancies by blocking the intrinsic inhibitors of immunity, and may have off-target effects known as immune-related adverse events (irAEs), including ICI-inflammatory arthritis (ICI-IA) is one of the most common rheumatic irAEs (Rh-irAEs), but is often underreported in clinical trials where nonspecific symptoms, such as joint pain, are reported without identifying the underlying causes. To date, there are no validated International Classification of Diseases (ICD) code-based case definitions for ICI-IA, limiting the ability to conduct reliable epidemiological research using administrative health data.We aimed to bridge this gap, in order to facilitate population-based studies of ICI-related irAEs.Methods We conducted a validation study using the Alberta CanRIO database of patients with rheumatologist-confirmed Rh-irAEs diagnoses, linked to provincial administrative health data. ICI-IA confirmation (our gold standard) required the new onset of at least one swollen joint on exam or synovitis on imaging following ICI exposure, without another cause or pre-existing IA. We compared ICD-9 (714.x and 696.0) and ICD-10 (M05.x, M06.x, and M07.x) diagnostic codes used for IA, against our gold standard. Seven core algorithms of different combinations of outpatient and hospitalization codes were tested for sensitivity and specificity. Results were stratified by sex. Two sensitivity analyses included: (1) exclusion of late-onset ICI-IA cases, defined as those developing more than 365 days after ICI discontinuation; and (2) incorporating emergency department (ED) visit ICD codes.Results We included 228 patients in the final analysis: 100 with ICI-IA and 128 without. Across all algorithms tested, sensitivity ranged from 4.0% (95%CI 0.2, 7.8) to 88.0% (95%CI 81.6, 94.4), while specificity ranged from 87.5% (95%CI 81.8, 93.2) to 99.2% (95%CI 97.7, 100.0) ( table 1). The algorithm requiring ≥1 outpatient IA ICD-9 code achieved the best balance of sensitivity (88, 95%CI: 81.6, 94.4) and specificity (87.5, 95%CI: 81.8, 93.2). Case definition performance was similar between sexes. Excluding late-onset ICI-IA cases resulted in a small decrease in sensitivity. Adding ED visit codes did not improve performance.Conclusions The best algorithm to identify ICI-IA from administrative health data included ≥1 outpatient ICD diagnostic codes. External validation is required to assess the performance of ICD-based ICI-IA case definitions in other locales and patient populations. This new tool could pave the way for better population-based studies of irAEs.Ethics Approval The study was approved by the Health Research Ethics Board for the University of Alberta (Pro00118986), which granted waiver of individual consent.Abstract 400 Table 1Validation of International Classification of Diseases (ICD)-based case definitions within confirmed* (N=228) immune checkpoint inhibitor associated inflammatory arthritis cases (95% confidence intervals, CI)*Amongst patients seen by rheumatologists for rheumatic immune-related adverse events, PPV: positive predictive value, NPV: negative predictive value, P: outpatient physician billing claims codes, H: hospitalization code or inpatient physician claims code
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 | 0.002 | 0.028 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.003 |
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".