Скачать презентацию Improving the Accuracy of American Indian Data through Скачать презентацию Improving the Accuracy of American Indian Data through

bf06dd55eeec1839e474dc0b52610eb6.ppt

  • Количество слайдов: 47

Improving the Accuracy of American Indian Data through Linkage to Tribal Rolls NAPHSIS 2008 Improving the Accuracy of American Indian Data through Linkage to Tribal Rolls NAPHSIS 2008

Need for Accurate American Indian Health Data Public Health Priority Issue Known Health Disparities Need for Accurate American Indian Health Data Public Health Priority Issue Known Health Disparities Strong Tribal Interest in Data Recognized Lack of Reliable Data

Federally Recognized Tribes Michigan Keweenaw Bay Chippewa Lac Vieux Desert Band of Chippewa Saginaw Federally Recognized Tribes Michigan Keweenaw Bay Chippewa Lac Vieux Desert Band of Chippewa Saginaw Chippewa The Sault Tribe of Chippewa Little Traverse Bay Band of Odawa Pokegon Band of Potawatomi Gun Lake Pottawattomi Little River band of Ottawa Grand Traverse Band of Ottawa and Chippewa Hannahville Potawatomi Huron Potawatomi Bay Mills Chippewa

Concern for Accuracy of American Indian Data History of Underreporting Extremely Low Disease Rates Concern for Accuracy of American Indian Data History of Underreporting Extremely Low Disease Rates Special Interest Population Tribal Health Planners lack Reliable Data 8 th largest American Indian Population 73, 295 Michigan American Indians in 2004

Summary of Previous of AI/AN Linkages Study Incidence (# cases) before match Incidence (# Summary of Previous of AI/AN Linkages Study Incidence (# cases) before match Incidence (# cases) after match Change Washington (1992 -93) 153. 5 267. 5 74% Minnesota (1988 -93) 245. 2 390. 25 59% California (1988 -92) 89. 6 211. 0 135. 5% Puget Sound (1974 -89) (137) (233) 70%

Cancer Incidence in American Indians appears to be MUCH LOWER than Total U. S. Cancer Incidence in American Indians appears to be MUCH LOWER than Total U. S. Population Swan, J and Edwards BK. Cancer Rates among American Indians and Alaska Natives: Is there a National Perspective? Cancer. 2003; 98(6): 1262 -1272.

CDC and IHS Initiative Link IHS registry to health data Target cancer and death CDC and IHS Initiative Link IHS registry to health data Target cancer and death registries Promote use of improved data Key contacts are n David Espey, CDC/IHS w David. [email protected] gov n Melissa Jim, CDC/IHS w Melissa. [email protected] gov

Assessing and improving the quality of cancer surveillance data for American Indians • Link Assessing and improving the quality of cancer surveillance data for American Indians • Link records from state cancer registries/death records with IHS patient registration database • Identify AI/AN cancer cases/deaths misclassified as non -Native • Use “improved” data to report cancer burden of American Indians

Approvals IHS National IRB CDC National Center for Chronic Disease Prevention and Health Promotion Approvals IHS National IRB CDC National Center for Chronic Disease Prevention and Health Promotion State IRB approvals (i. e. Department of Health & Senior Services, Office of Public Health) A data quality improvement activity in support of routine public health surveillance – public health practice not research

Data Management MOA or state Data Use Agreement Data sent to ABQ or travel Data Management MOA or state Data Use Agreement Data sent to ABQ or travel to state Linkage done using Link. Plus Send back Match File 3 n Contains both IHS & state information that was used to determine if the match was a link Create de-identified analysis files

Linkages between IHS and state cancer registries WA NH ME ND MT OR MN Linkages between IHS and state cancer registries WA NH ME ND MT OR MN ID WY NV NV SD WI PA IA NE UT IL CO CA AZ NY MI KS OK NM MO OH KY WV VA NC TN AR MS AK TX IN AL SC GA LA FL HI VT MA RI CT NJ DE MD

Registry linkages ²Linkage status captured in “IHS Link” variable in NAACCR layout ²Surveillance monograph Registry linkages ²Linkage status captured in “IHS Link” variable in NAACCR layout ²Surveillance monograph in journal Cancer ²Annual Report to the Nation with special section on cancer in AI/AN

Record Linkage between NPCR Program registries and IHS administrative records Graph with smaller scale Record Linkage between NPCR Program registries and IHS administrative records Graph with smaller scale for lower 20 NPCR Program registries

CHSDA Non-CHSDA CHSDA Non-CHSDA

Michigan NPCR-IHS Linkage Results, 19952002 CHSDA n n n Before: 250 individuals After: 612 Michigan NPCR-IHS Linkage Results, 19952002 CHSDA n n n Before: 250 individuals After: 612 individuals Difference: 362 individuals that were previously coded as race other than Al/AN in registry Non-CHSDA n n n Before: 248 After: 350 Difference: 102 individuals that were previously coded as race other than AI/AN in registry

Linkages between IHS and state death records Similar problems of misclassification Much larger databases Linkages between IHS and state death records Similar problems of misclassification Much larger databases Less standardization than cancer registries

IHS/state death records data linkage WA NH ME ND MT OR MN ID WY IHS/state death records data linkage WA NH ME ND MT OR MN ID WY NV NV SD WI PA IA NE UT IL CO CA AZ NY MI KS OK NM MO KY WV VA NC TN AR MS AK TX OH IN AL SC GA LA FL HI Linkage completed Data rec’d – linkage in progress Interest expressed – data not rec’d Not participating VT MA RI CT NJ DE MD

Michigan Division of Vital Records and Health Statistics-IHS Linkage Results, 1990 -2002 Before: 5, Michigan Division of Vital Records and Health Statistics-IHS Linkage Results, 1990 -2002 Before: 5, 585 individuals After: 5, 924 individuals Difference: 339 individuals that were previously coded as race other than AI/AN on death certificate

Linkage of death records from State VS with administrative records from the IHS (1990 Linkage of death records from State VS with administrative records from the IHS (1990 -2003)

Collecting ancestry in 1989 noticeably improved the reporting of American Indian Collecting ancestry in 1989 noticeably improved the reporting of American Indian

Multiple Responses now Collected Up to 5 choices for race n Collected on Deaths Multiple Responses now Collected Up to 5 choices for race n Collected on Deaths since 2004 w 450 Native American 1 st choice w 840 “any mention” in 2005 data n Collected on Births since 2006

Considering Both Parents for Newborns Increases Cases Significantly During 2005, n 736 Native American Considering Both Parents for Newborns Increases Cases Significantly During 2005, n 736 Native American Mothers 734 Native American Fathers n 1, 239 either parent Native American n

Additional Change in “Process” Collecting Principle Tribe Additional Change in “Process” Collecting Principle Tribe

Additional Change in “Process” American Indian Additional Change in “Process” American Indian

Beginning to Collect Principle Tribe Little Traverse Bay Ottawa Beginning to Collect Principle Tribe Little Traverse Bay Ottawa

Issues What proportion in Michigan are IHS? IHS recipients more or less likely to Issues What proportion in Michigan are IHS? IHS recipients more or less likely to be reported as American Indian? Census population issues compound problem? Who should be considered American Indian?

Solution Appears to be Tribal Links Applies tribal definition of membership Enables analysis of Solution Appears to be Tribal Links Applies tribal definition of membership Enables analysis of Tribe as a cohort Empowers tribal health planners Resolves numerator/denominator concerns

Proposal for Link to Tribal Rolls Interested Tribe would Provide Tribal Roll Link Tribal Proposal for Link to Tribal Rolls Interested Tribe would Provide Tribal Roll Link Tribal Roll to Dataset of Interest: n n n Births Deaths Cancer Registry Release Results on Tribe to Participating Tribe n DCH would provide a de-identified file for analysis Augment state level data n Special reports on American Indian Health Validate accuracy of case reports Evaluate principle tribe reporting

Proposed Approach Develop a Protocol and Data Use Agreement Establish as a Research Project Proposed Approach Develop a Protocol and Data Use Agreement Establish as a Research Project n Assure protection of the data Develop Approach that Assures Confidentiality n Options include: w Provide tribal roll to DCH for Link n Identifiable data used only for linkage then destroyed w Jointly conduct link on a portable devise n Role loaded/linked/destroyed DCH to retain which specific cases linked Tribe to receive an agreed de-identified data set

Issues Need interest of tribal health planners approval of tribal leaders to assure confidentiality Issues Need interest of tribal health planners approval of tribal leaders to assure confidentiality to maximize privacy of tribal roll

Approach Contact Michigan Intertribal Council n Build on history of collaboration Model after IHS Approach Contact Michigan Intertribal Council n Build on history of collaboration Model after IHS approach Develop tribal interest n Organize meeting through Intertribal Council Develop clear written agreement Obtain IRB review

Outcomes De-identified file of linked cases to tribe n n File is owned by Outcomes De-identified file of linked cases to tribe n n File is owned by tribe for use in analysis of cancer incidence and related issues Any results to be released only with approval of the tribe Id number field for linked cases to state n State can use to improve registry data on American Indian

Questions using Tribe-specific data: n n n What types of cancers are most/least common Questions using Tribe-specific data: n n n What types of cancers are most/least common in tribe members? How do tribal rates of cancer by type compare to other populations? What stage is cancer in tribal members at diagnosis? How long do tribal members with cancer live after being diagnosed? What is the geographic distribution of cancer (rural vs. urban, county-level characteristics)

In Attendance The Linkage was conducted on 7/20/2007 at the tribal Administration Building Tribal In Attendance The Linkage was conducted on 7/20/2007 at the tribal Administration Building Tribal Staff Present: n n Database Manager Tribal Registrar Tribal Attorney Environmental Manager State Staff: MI Cancer Registrar Study Staff: Study PI from University of Michigan

De-Identified Dataset Drop first, middle, last name, SSN, DOB Keep cancer characteristics (eg. primary De-Identified Dataset Drop first, middle, last name, SSN, DOB Keep cancer characteristics (eg. primary site, stage, year of diagnosis, age at diagnosis) Keep some demographics (sex, race and ethnicity as reported, survival time) Keep county-level characteristics (eg. urban/rural, % American Indian, % minorities) Tribe and State receive a copy of this de-identified dataset. Access to the state data set must be restricted under the terms of the data use agreement

Other Limitations Periods of open and closed enrollment–changes in the tribe’s population over time Other Limitations Periods of open and closed enrollment–changes in the tribe’s population over time are difficult to interpret Enrollment bias n n many people who identify as tribal members may not be in the database are sicker people more likely to enroll than healthy people? Detection bias n Cancer registry only picks up cancers in Michigan; if a tribal member was never treated for cancer in MI, they wouldn’t be recorded; tribal roster, on the other hand, is worldwide

Reported as American Indian to the State Death Registry Linked to Indian Health Services Reported as American Indian to the State Death Registry Linked to Indian Health Services patient files (IHS) 56 deaths 4, 975 deaths 200 deaths 154 deaths 128 cases Linked to Tribal Membership roster *IHS began linking with the MI registry in 1995, so this diagram is for 1995 – 2004.

Directions Develop interest in other tribes Assist with tribal cancer control plan Promote the Directions Develop interest in other tribes Assist with tribal cancer control plan Promote the linkage of births and deaths Work toward a periodic linkage routine n Establish baseline and monitor trends

Acknowledgements Intertribal Council of Michigan n Rick Haverkate, Health Director Noel Pingatore Elizabeth Kushman Acknowledgements Intertribal Council of Michigan n Rick Haverkate, Health Director Noel Pingatore Elizabeth Kushman Dan Tadgerson n Tribal Environmental Manager David Espey, CDC/IHS Melissa Jim, CDC/IHS Jennifer Johnson, UM Doctoral Candidate