Enhancing Incidence Data with Passive End-Results Jill MacKinnon, Sarah Manson, and Mayra Alvarez Florida Cancer Data System
Overview Current data enhancements Pilot project Results of pilot project –Data enhancements Dates of last contact Co-morbidities
Florida Cancer Data System Incidence registry (1981) –NPCR required data + 3 state specific fields Current enhancement –FL Vital Statistics Mortality Passive end results and retrospective QC case finding –FL discharge records (cancer dx only) Retrospective QC case finding
Hospital Discharge Data nteragency agreement allowed for: –Retrospective casefinding –Augmenting incident data Demographic
Hospital Discharge Data - Pilot Interagency agreement amended to allow: –Augmenting treatment data (longitudinally) –Calculating co-morbid conditions –Updating date of last contact
Data Enhancement Pilot Project “Team Science” Tobacco related cancers Funding was provided by a Team Science Award from James and Esther King Biomedical Research Program to the University of Miami Miller School of Medicine
Discharge Data (AHCA) Agency for Health Care Administration Current –Cancer diagnosis only Pilot project –All cause
Tobacco Related Cancers lung, oral, esophageal, pancreas, larynx, cervical, kidney, bladder, acute myeloid leukemia, stomach and breast cancer
Pilot Project Address two main questions –Impact on patient date of last contact –Ability to address co-morbidities
Pilot Results
Discharge Database Linkage Probabilistic –No patient names Hospital facility –Inpatient and outpatient Ambulatory surgical facilities Radiation therapy facilities Ambi
FCDS (Dx 97-04) Total FCDS tobacco related = 389,965 –Alive = 211,196 –Expired = 178,769
First Question Enhance date of last contact? To what extent?
Source of Date of Last Contract Living Patients Only No Match4.8 All Cause Hosp13.1 All Cause Ambi13.9 Ca only Hosp25.2 Ca only Ambi43.0
Dx 1997 Cases “Latest” Date of Last Contact (All cause discharges) Years Passive F/U
Percent Distribution by Age Group
Stage Distribution x DLC Source AHCA Data ( discharges)
Second Question Co-morbid conditions –Calculate –Implications
Co-morbid Conditions Combine all discharge records (n~1.3 million) Assign co-morbid conditions based on primary and/or secondary diagnosis –Each patient can have one or more co-morbid conditions
Comorbidity Indices Elixhauser (30 conditions) Charlson (19 conditions) Outcome variable –Series of 0/1 or T/F for each condition
Top Co-morbid Conditions (Excluding Primary Cancer)
Co-morbid Conditions All Patients (living and dead) Overall 28% of records were not matched –Site specific no match ranged 0.1% % Head and Neck smallest % missing Bladder largest % missing –Kidney 16.0% –Female breast 11.9% –Lung 6.8% –Liver 6.0%
At the end of the day
We are not an end-results registry but We can make our passive Registries more active
Thank you