NCI Comorbidity Index for SEER-Medicare
A cancer-focused claims comorbidity score derived from Medicare claims in SEER-Medicare studies, commonly built with NCI SAS macros using diagnosis and procedure evidence before cancer diagnosis.
In plain language
The NCI Comorbidity Index is a practical cancer-research comorbidity score used with SEER-Medicare claims. It helps researchers account for how sick cancer patients were before diagnosis or treatment, using Medicare claim evidence and NCI-supported macros.
The NCI Comorbidity Index is a SEER-Medicare-oriented implementation of claims-based comorbidity measurement for cancer outcomes research. It adapts the Charlson/Deyo/Klabunde lineage to cancer registry-linked Medicare claims and provides practical SAS macro support for calculating comorbidity variables from Medicare claims before a cancer diagnosis or study index date.
The core operational idea is familiar: use baseline claims to identify comorbid conditions, exclude codes that are likely part of the cancer diagnosis or treatment process, apply source-specific rules, and summarize burden for risk adjustment. What makes the NCI implementation distinctive is its SEER-Medicare cancer context, including cancer-site concerns, timing relative to diagnosis, and macro-based reproducibility.
Analysts should avoid treating the NCI Comorbidity Index as a generic all-disease score. It was designed for cancer registry-linked Medicare data. It is strongest when the research question, population age, claim types, and baseline window match that context; outside it, Charlson/Quan, Elixhauser, frailty, or disease-specific covariates may be more appropriate.
Pros, cons, and trade-offs
The NCI implementation is useful because it is anchored in the SEER-Medicare cancer research workflow and has reproducible macro support. It handles cancer timing and baseline claims context more directly than a generic comorbidity score. The trade-off is portability. The score inherits assumptions about Medicare FFS observability, older cancer populations, registry linkage, diagnosis timing, and macro version. It may not transport to Medicare Advantage, commercial oncology data, younger patients, or EHR-only cohorts without revalidation.
When to use
Use the NCI Comorbidity Index when the study is SEER-Medicare or closely comparable registry-linked Medicare cancer research, baseline comorbidity is measured before diagnosis or treatment, and the analysis can preserve macro version, claim files, lookback window, and component flags. It is especially appropriate when reviewers expect SEER-Medicare conventions.
When NOT to use - and when it is actively misleading
Do not use it as a universal cancer severity score, as a post-diagnosis disease-burden measure, or as a generic comorbidity score outside Medicare cancer cohorts. It is actively misleading to count cancer diagnosis or treatment-process codes as baseline comorbidity, or to ignore incomplete FFS observability caused by Medicare Advantage enrollment.
Index definitions
Source-backed definitions and variants for the index or checklist family.
| name | definition | source | use | notes |
|---|---|---|---|---|
| NCI Comorbidity Index | Claims-based cancer comorbidity score used in SEER-Medicare research, implemented through NCI calculation guidance and SAS macros. | NCI Healthcare Delivery Research Program | Baseline risk adjustment in cancer registry-linked Medicare studies. | Use the macro version and documentation matching the study period and claim files. |
| NCI Comorbidity SAS macro | NCI-distributed SAS implementation for calculating comorbidity variables from SEER-Medicare claims. | NCI SEER-Medicare macro documentation | Reproducible score construction in Medicare cancer cohorts. | Store macro version, input files, and parameter settings in the study archive. |
| Charlson/Deyo/Klabunde cancer adaptation | The methodological lineage of claims-based comorbidity construction adapted for cancer outcomes analyses. | SEER-Medicare comorbidity methods literature | Interpretation of what the NCI score represents and how it differs from generic CCI. | Cancer diagnosis and treatment timing can contaminate baseline comorbidity if not handled explicitly. |
Worked example
Scenario
A SEER-Medicare lung cancer study defines baseline comorbidity during the 12 months before diagnosis, excluding the diagnosis month. The patient has claims for COPD, congestive heart failure, and diabetes before diagnosis; metastatic cancer codes after diagnosis are not counted as baseline comorbidity.
Dataset
Simplified baseline comorbidity evidence.
| claim_month | claim_source | condition_signal | counted_for_nci_index |
|---|---|---|---|
| -11 | MedPAR | COPD | True |
| -7 | Carrier claim | congestive heart failure | yes if rule-out criteria met |
| -3 | outpatient claim | diabetes | yes if repeated or supported by macro rule |
| 2 | oncology claim | metastatic cancer | no; post-diagnosis cancer process |
Steps
Require complete Medicare FFS observability for the baseline window.
Apply the NCI macro using claim files and diagnosis timing specified in the analysis plan.
Exclude the diagnosis month and cancer-related conditions as instructed.
Carry the score, component flags, and macro version into the analytic archive.
Result
COPD, CHF, and diabetes contribute to baseline illness burden; post-diagnosis metastatic cancer does not become a baseline comorbidity flag.