Download presentation
Presentation is loading. Please wait.
Published byMabel Turner Modified over 6 years ago
1
Using SEER*Stat MP-SIR to investigate cancer incidence among occupational cohorts or “Are you catching cancer at work?” Presented by: Chris Johnson, MPH Epidemiologist, Cancer Data Registry of Idaho 2014 NAACCR ANNUAL CONFERENCE Wednesday, June 25, 2014 Ottawa, Ontario
2
Many Thanks To Steve Scoppa Senior Systems Analyst
Information Management Services Inc. Angela Mariotto Chief, Data Modeling Branch National Cancer Institute The Centers for Disease Control and Prevention is acknowledged for support of CDRI under cooperative agreement U58DP The findings and conclusions in this presentation are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or the National Cancer Institute.
4
What is a disease cluster?
A disease cluster is the occurrence of more than the expected number of people diagnosed with a certain disease within a specific group, a geographic area or a period of time.
5
Cancer Cluster Investigation Themes
Occupational and medical clusters have aided our understanding of cancer etiology. Studies of geographic clusters rarely identify environmental risk factors. Science can be overtaken by public concern. Geography might be symptom rather than cause.
6
CDRI has a protocol for investigating non-infectious disease clusters
The Cancer Data Registry of Idaho – CDRI - has a protocol for investigating non-infectious disease clusters which we updated in April The protocol covers cancer and non-cancer reports and is generally based on the 2013 cancer cluster Guidelines from CDC and CSTE. This document is designed to address geographic clusters.
7
CDRI Protocol Worksite Cancer Clusters ??? 2014
CDRI is developing a protocol for investigating worksite/occupational cancer clusters CDRI Protocol Worksite Cancer Clusters ??? 2014
8
Two Recent Worksite/Occupational Cancer Cluster Investigations
Study #1 - Worksite Study #2 – Occupational group (licensed)
9
Study #2 – Background An oncologist noticed what she believed to be an unusually large number of cases in a particular occupational group (licensed professional) among her patients. October 2013: this concern was brought to the state of Idaho Board that governs this group. January 2014: the Board voted to share data with CDRI to investigate the concern.
10
Study #2 – Background February 2014: CDRI received dataset on the current and historical licensed professionals No SSN Month and year of birth, not day February-March 2014: deduplication, linkages, Accurint & SSDI lookups April 2014: statistical analysis May 2014: report writing ??? 2014: meet with Board to share results At least 8 months…
11
Mock-up of Case and Person-time Contributions by Licensed Professional
For each person in the roster, we counted the time in months from the later of their start date or January 1971 (when CDRI became population-based) until the end of the study period, which was the end of July 2013. The count of person-time is the denominator for a rate. For the numerator, we counted the cases that occurred during their employment, or after their employment, between January 1971 and July 2013. * (asterisk) means cancer case. Vertical line means person let their license expire. End of colored bar before July 2013 means person died before end of study period. We did not count person time that occurred prior to January 1971 because CDRI was not population based until 1971, so we could not assess cases that may have occurred prior to this date.
12
This is a screen shot of SEER Prep with the special data dictionary file for a cohort study.
There is no population file associated with the cohort file, but the cohort contains both people who were diagnosed with cancer, and people who were not. If they were not diagnosed with cancer, they have one record in the data file. If they were diagnosed with cancer, they will have a record for every primary cancer.
13
Study #2 – Statistical Methods
We compared the observed numbers of cancer cases among occupational cohort to expected numbers based on referent rates from Idaho and the US. Standardized incidence ratio (SIR) = obs/exp Referent rates: National Cancer Institute’s Surveillance, Epidemiology, and End Results Program (SEER ) for whites Idaho overall Results were tabulated using CDRI annual report primary site categories, which include malignant cases and benign brain and other nervous system cases.
14
SEER*Stat MP-SIR Session
MP-SIR (Multiple Primary -- Standardized Incidence Ratios) sessions can be used to compare incidence of cancer in a defined cohort of persons previously diagnosed with cancer to the incidence of cancer in the general population. Can also be used to compare incidence of cancer in other cohorts to the incidence of cancer in the general population.
15
This is a screen shot from the SEER. Stat MP-SIR session
This is a screen shot from the SEER*Stat MP-SIR session. On the Rates tab, you choose a population-based referent rate file to use in your analysis. The box at the top of the window lists the available referent rate files. The Rate File Details box contains information about the highlighted rate file, such as the database on which it is based and the variables it contains. The box at the bottom of the window indicates how SEER*Stat will handle discrepancies between variables in your selected database and the corresponding variables in the rate file.
16
You use the Parameters Tab in an MP-SIR session to specify your desired settings for the dates and length of your study and the cut-points for the time-dependent variables in the analysis. A subject begins accruing person time at risk, and SEER*Stat begins counting their events, at the latest of these three dates: the Exposure Date plus the Latency Exclusion Period the Start Date the study's Cutoff Start Date
17
You use the table tab to calculate statistics stratified by the variables of your choice.
18
Study #2 – Results Analytic datasets contained 881 records.
If a person in the occupational cohort did not link to CDRI, they have 1 record for person-time calculations. For persons who did link to CDRI, they have one record for every primary cancer case. 86 cancer cases among cohort, spanning 59 deceased persons in cohort. As with Study #1, some cancer cases occurred before the person joined the occupational cohort.
19
Study #2 – Results – Table 1
20
Study #2 – Results – Table 2
21
Study #2 – Results – Table 3
22
Lessons Learned from the 2 Studies
CDRI not in control of employer/licensure data Availability Content Completeness Timeliness May take a long time from initial request to final report What you have to do in terms of data cleanup, extra linkages, depends on employer data Difficulty of linkages with cancer registry depends on fields available from employer data and data quality
23
Lessons Learned from the 2 Studies
Interpretability of results may be contingent on structure of human resources database Compared to geographic cancer clusters, seemingly more buy-in from concerned persons if negative results SEER*Stat provides a user-friendly interface and consistent approach to conducting worksite/occupational cancer cluster investigations
24
SEER*Stat MP-SIR It is hoped that this approach to the analysis of cancer incidence among occupational cohorts could be easily adopted by other central cancer registries and will be a useful model for addressing worksite cancer cluster concerns. “SEER*Stat does that!”
25
Thank You Chris Johnson, CDRI Epidemiologist 208-489-1380
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.