The incidence rate of colorectal cancer in the United States is XX.X (95% CI: xx.x-xx.x), and is X times higher/lower among American Indian/Alaskan Natives from XXXX-XXXXIn Arizona, the incidence rate of colorectal cancer is XX.X (95% CI: xx.x-xx.x) and is x times higher/lower among American Indian/Alaskan Natives from XXXX-XXXX
The highest rates are seen among Black (6.2) and White males (7.8)
Arizona cancer registry NPCR stateSEER- Arizona Indians data provided to the New Mexico registry for calculation estimates2000-2009 eight participating hospitals in AZ
Annual Arizona Conference for Tribal BCCEDP Collaboration, Flagstaff, AZ
Inter Tribal Council of Arizona, Inc. Tribal Epidemiology Center TEC Director Jamie Ritchey, MPH, PhD August 7, 2012
Objectives• Overview of Inter Tribal Council of Arizona (ITCA), Inc. Tribal Epidemiology Center (TEC)• Epidemiology Basics• Practical Applications of Epidemiology
Overview of Inter TribalCouncil of Arizona, Inc. (ITCA)Tribal Epidemiology Center (TEC)
Tribal Epidemiology Center• Who we are• Where we are• What services are provided• How to request services and partnerships
Tribal Epidemiology CenterITCA, Inc. Regional Epidemiology Center:• Established in 1996• Mission: Empowering the American Indian Tribes in Arizona, Nevada, and Utah in the further development of health services and systems• Purpose: To support Tribally-driven Health Surveillance Systems that can assess both individual and community health status, facilitate planning, and manage existing health services
Tribal Epidemiology Center• Services – Community health profile (CHP) assistance – Epidemiology and other public health trainings – Study and survey design – Data collection and analysis – Technical report creation and review – Educational materials for health-related topics – Coordination of services during outbreaks or disease cluster investigations – CHP and Community Health Accreditation (CHA) tool kits for Tribes coming soon!
Tribal Epidemiology Center• How can I request ITCA, Inc. TEC services? – E-mail ITCA, Inc. TEC for assistance directly at: TECinfo@itcaonline.com – TEC staff will respond within 48 hours and provide you with a fillable form to complete – TEC staff will meet with you by phone or e-mail to discuss a project work plan – TEC staff will decide with you the format and delivery method of the final product1 1Please allow at least 2 weeks for project completion, possibly longer depending on the scope of the project
Epidemiology Basics• Epidemiology – Definition – How public health professionals use it• Types of Epidemiology• Descriptive Epidemiology – Person, place, time – Measures – Relationship between measures
Epidemiology Basics Epidemiology is defined as: “…the study of the distribution and determinants of health-related states or events in specified populations and the application of this study to control of health problems.”11 Gordis, L. Modern Epidemiology 2nd Edition. 2000. W.B. Saunders Company, Philadelphia.ISBN 0-7216-8338-X
Epidemiology Basics• It enables public health professionals to: – Understand the local disease patterns – Identify populations at risk for disease – Establish associations with risk factors and disease – Determine causes of disease – Develop new prevention programs and policies – Set health-based standards
Epidemiology Basics• Descriptive epidemiology1 – Person, place, time – Measures: counts, proportions, rates – Explains or quantifies a particular disease or problem (ex. Cancer rates)• Analytic epidemiology – Tests a hypothesis – Measures: relative risk, odds ratios – Describes associations between a risk factor and a disease (ex. Smoking and lung cancer relationship)1Focus for today’s talk
Epidemiology Basics• Person – Ex. Diagnosed Colorectal cancer cases• Place – Arizona Community Health Analysis Areas• Time – 1995-2000 – 2001-2004
Epidemiology Basics• Uses statistical measures to describe: – New cases of disease and death – People living with disease – Identify possible risk factors for the disease
Epidemiology Basics• Counts / FrequencyThe number of events (“cases”) that occurin a population of interest – Example: There were 87 cancer cases in Tribe A • Is this story complete? • What else would you like to know?
Epidemiology Basics• Proportions give a magnitude to events• Useful info might include: – Time • 87 cancer cases in 1 year (1999) – Total Number of Deaths • 87 cancer deaths/1,000 total deaths = 0.087 • Multiply by 100% 0.087 x 100% = 8.7% of deaths were cancer cases in 1999
Epidemiology BasicsTypes of Rates• Crude rates• Stratified or Specific Rates – Better detail – Uses specific population (age group, sex, ethnic group, etc.) – Ex. Cancer death rates in males & females• Adjusted rates – age
Epidemiology Basics• Proportions – Prevalence (NOT a rate) – Not directly comparable across groups – Used for public health planning purposes to determine the burden of disease• Rates – Incidence and Mortality rates – Allow health comparisons within and between groups
Epidemiology Basics • Incidence rate: Risk of disease development in the population (new cases) • Prevalence: Fraction of population with illness in population • Mortality rate: Risk of Death • Incidence and prevalence are related: I X P = Duration
Epidemiology Basics• How do epidemiologists know when rates are statistically different?There are measures that can determine if differences are statistically significant… – P-values of p<0.05 indicate that measures have a statistically significant difference – 95% Confidence intervals that do not overlap are considered a way to tell if measures show a statistically significant difference. These intervals can also tell us about the magnitude of the difference.
Epidemiology Basics Statistical measures are used to tell astory…but where do I get data to tell it?
Practical Applications of Epidemiology• Arizona Cancer Registry• Statistics for working documents – Scenarios
Arizona Cancer Registry• Began in 1981• Mandatory cancer reporting in 1988 Arizona Revised Statute §36-133• Rules for case reporting in 1992 Arizona Administrative Code Title 9, Chapter 4• Provide data to New Mexico Tumor Registry for American Indian registry for SEER statistics• Population-based NPCR registry: Arizona Cancer Registry Home page: – Cancer site http://www.azdhs.gov/phs/phstats/acr/ – Case demographics – Year of cancer diagnosis
Arizona Cancer Registry What statistical measures are available? • Cancer case counts • Cancer incidence rates1 • Cancer mortality rates2 • Population estimates – Denominator data1Age-adjusted incidence rates; 2Crude mortality rates per 100,000
Statistics for working documents Scenario 1. A Tribal community cancer researcher has heard rumors from concerned community members that there may be a high risk of colorectal cancer in her area. From her training, she remembers that the national trend of age-adjusted incidence rate of colorectal cancers from 1992-2009 were dropping in the US1. She wants to determine if her community has a high risk of colorectal cancer compared to others. • What are the person, place, and time components? • What measure is needed to determine risk and can be directly compared across geographical areas? • How can I get this information from the Arizona cancer registry?1 Age-adjusted colorectal cancer incidence rate trend slides from SEER are included in the presentation.
Statistics for working documents What data does the researcher need? • Person – Colorectal cancer cases among AZ residents and Tribal community members • Place – Arizona state in community health analysis areas (CHAAs)1 • Time – Not specified – Data lag, limited to what is available21Tribespecific data is not publicly available. Community Health Analysis Areas are used to estimateIncidence rates based on Census blocks in Tribal areas and may include non-Tribal members.2Cancer data takes at least 18 months to check for completeness from the central registry.http://azdhs.gov/phs/azchaa/CHAA_FAQ.pdf
Statistics for working documents What statistics do we use to determine the risk of colorectal cancer comparing areas? Use age-adjusted incidence rates: • Risk of getting disease • Comparisons of groups Use 95% confidence intervals: • Rate differences are statistically significant • Magnitude of the difference • Stability of the rates11Rates may be unreliable with small numbers of cases. The 95% CIs will often be a wide range.
Statistics for working documentsThree ways to get thestate data:• From the home page, go to the Cancer Data Query System link• Contact the Arizona Cancer Registry Data Section by e-mail or phone• Request services from http://www.azdhs.gov/phs/phstats/acr/ TEC
Statistics for working documentsThe Cancer DataQuery System,incidence rates can beaccess in two ways: – Age-adjusted cancer Incidence rates – ACR Community Health Analysis Area Maps11Tribespecific data is not publicly available. Community Health Analysis Areas are used to estimateIncidence rates based on Census blocks in Tribal areas and may include non-Tribal members.
Statistics for working documentsIncidence rates and 95%Confidence Intervals: – For AZ – By Year – All race/ethnicity – AI/ANs
Statistics for working documentsAZ colorectal cancer age-adjustedincidence rates show a decreasingtrend from 1995-2009 for both malesand females1• The age-adjusted incidence rate of colorectal cancer in 1995 for AZ men was 57.8 per 100,000 (95% CI: 54.2-61.4) and in 2009 was 40.9 (95% CI: 38.4-43.4) 2• The age-adjusted incidence rate of colorectal cancer in 1995 for AZ women was 38.1 per 100,000 (95% CI: 35.2-40.9) and in 2009 was 31.8 (95% CI: 29.6-33.9) 21 Results from a trend test would tell us if this downward trend is statistically significant2 Statistically significant difference.
Statistics for working documentsAIs in AZ1:• The age-adjusted incidence rate of colorectal cancer in 1995 for AI men was 50.6 per 100,000 (95% CI: 34.9-66.4) and in 2009 was 32.2 (95% CI: 21.8-42.5). In 2005, the rate was 41.0 (95% CI: 28.8-53.3).• The age-adjusted incidence rate of colorectal cancer in 1995 for AI women was 9.7 per 100,000 (95% CI: 3.1-16.3) and in 2009 was 18.4 (95% CI: 10.8-25.9). In 2006, the rate was 30.5 (20.4-40.5).21 Results from a trend test would tell us if this downward trend is statistically significant2 Statistically significant difference comparing 1995 and 2006 for women.
Statistics for working documentsThe Cancer Data QuerySystem, incidence ratescan be access in twoways: – Age-adjusted cancer Incidence rates – ACR Community Health Analysis Area Maps1 1Tribespecific data is not publicly available. Community Health Analysis Areas are used to estimate Incidence rates based on Census blocks in Tribal areas and may include non-Tribal members.
Statistics for working documents• What is an Arizona Community health analysis area (CHAA)?CHAA basics: – NOT Tribal specific data – Built on the 2000 Census Block groups – Contain a range of 5,000-190,000 residents – Cancer cases are assigned based on place of residence – PO boxes were assigned to the town of the zip code – About 2% of cancer cases did not get assigned to CHAA – Additional information on CHAA: http://www.azdhs.gov/phs/azchaa/CHAA_FAQ.pdf
Statistics for working documents• Choose the cancer site and years
Statistics for working documents• Select colorectal cancers and either 2001-2004 or 1995-2000
Statistics for working documents• Apply the filter for Indian community „yes/no‟
Statistics for working documents• 14 CHAA areas had fewer than 10 colorectal cases• Some CHAA incidence rates appear higher than others• Navajo Nation CHAA had the highest amount of cases in any CHAA (n=40) (But not the highest incidence rate!)
Statistics for working documentsCHAA age-adjusted incidence ratesfor colorectal cancer from 2001-20041:• The 95% CIs indicate that the rates for Yavapai-Prescott CHAA 282 per 100,000 (95% CI: 0 – 835), Cocopah CHAA 86.0 (95% CI: 0-254.8) and other CHAAs with a small number of cases are unstable2• The Navajo Nation CHAA has a stable rate of 13.7 (95% CI: 9.4- 18.0) 1 Limited to Indian Community in CHAAs.• The Fort Mohave [Mojave] CHAA 2 TECs are formalizing a small numbers protocol. Many rate of 39.4 (95% CI: 20.8-57.4) and Agencies do not report rates based on fewer than 20 cases. Salt River CHAA rate of 77.1 (95% CI: 40.9-113.3), these CHAAs are fairly stable2
Statistics for working documentsWhat is the researcher‟s story that describes colorectal cancer?• National and AZ age-adjusted incidence rates for colorectal cancer are decreasing, but getting screened for colorectal cancer on an individual level is still VERY important• Among AIs in AZ, age-adjusted incidence rates may have decreased for AI men and may have increased for AI women from 1995 to 2009, but data is limited• Navajo Nation CHAA had the highest case count of colorectal cancer (n=40) from 2001-2004• The age-adjusted incidence rate of 13.7 (95% CI: 9.4-18.0) for colorectal cancers in the Navajo Nation CHAA is lower than the state and national rates• The Navajo Nation CHAA age-adjusted incidence rate is lower than Fort Mohave [Mojave] CHAA 39.4 (95% CI: 20.8-57.4) and Salt River CHAA 77.1 (95% CI: 40.9-113.3) per 100,000. These differences are statistically significant.• Other CHAA areas had higher reported rates, but these rates are unstable due to small numbers of cases
Statistics for working documentsScenario 2. A multi-disciplinary team of Tribal cancer researchers want topropose a case control study with several community members in order toinvestigating the relationship between esophageal cancer and arsenic inthe drinking water among AI/AN populations in Arizona from 1995-2004.Writing the project proposal introduction, the researchers want to know:• What is the person, place, and time understudy?• What statistics can the team use to describe the problem of esophageal cancer in AZ?• How do we get the descriptive cancer data from the registry?• Does the registry contain the exposure information (e.g., arsenic levels or other environmental exposures)?• Does registry information tell us about the association between cancer and arsenic in the drinking water?
Statistics for working documents What data does the researcher need? • Person AI/AN esophageal cancer cases1 • Place Arizona state • Time 1995-2004 Data lag, limited to what is available21Tribespecific data is not publicly available. Community Health Analysis Areas are used to estimateIncidence rates based on Census blocks in Tribal areas and may include non-Tribal members.2Cancer data takes at least 18 months to check for completeness from the central registry.http://azdhs.gov/phs/azchaa/CHAA_FAQ.pdf
Statistics for working documentsWhat statistics do we use to determine therisk of esophageal cancers among AIs in AZ?Use age-adjusted incidence rates:• Risk of getting disease• Comparisons of groupsUse 95% confidence intervals:• Rate differences are statistically significant• Magnitude of the difference• Stability of the rates1 1Rates may be unreliable with small numbers of cases. The 95% CIs will be a wide range.
Statistics for working documentsThe Cancer Data QuerySystem, incidence ratescan be access in twoways: – Age-adjusted cancer Incidence rates – ACR CHAA Maps
Statistics for working documents Age-adjusted incidence rates of esophageal cancer per 100,000• All AZ residents 1995-2009: – 4.3 per 100,000 (95% CI: 4.0-4.9)• All AZ men: – 1995: 6.9 (95% CI: 5.5-8.3) – 2009: 8.0 (95% CI: 6.8-9.2)• All AZ women: – 1995: 1.6 (95% CI: 0-2.2) – 2009: 1.7 (95% CI: 1.1-2.3)
Statistics for working documents Age-adjusted incidence rates of esophageal cancer per 100,000 among AIs• All American Indians AZ 1995-2009: – 3.0 per 100,000 (95% CI: 2.1-3.8)• American Indian men in AZ: – 1995: 6.6 (95% CI: 0-15) – 2009: 5.0 (95% CI: 0-10.8)• American Indian women in AZ: – 1995: 3.5 (95% CI: 0-9.1) – 2009: 3.1 (95% CI: 0-7.3)
Statistics for working documentsWhat is the researchers story for the proposal?• In AZ, the age-adjusted incidence rate of esophageal cancer is 4.3 per 100,000 (95% CI: 4.0-4.9) from 1995-2009• Among AIs in AZ, the age-adjusted incidence rate of esophageal cancer of 3.0 per 100,000 (95% CI: 2.1-3.8) from 1995-2009, which is lower than the state rate; and, the difference is statistically significant.• The registry does not include environmental exposure information like arsenic• The registry information does not tell us about associations with exposure and disease risk
Summary• Overview of Inter Tribal Council of Arizona (ITCA), Inc. Tribal Epidemiology Center (TEC)• Epidemiology Basics• Practical Applications of Epidemiology – Data Scenarios 1 & 2
2214 North Central Avenue, Phoenix, Arizona 85004 p 602.258.4822, f 602.258.4825 www.itcaonline.com
More Statistical TestsP-values• Estimated probability of rejecting the null hypothesis (H0) of a study question.• Null hypothesis is usually a hypothesis of "no difference" – Ex: there is no difference between high perceived risk and low perceived risk groups• Alternative hypothesis is a hypothesis of “difference” – Ex: there is a difference between high perceived risk and low perceived risk groups
Statistical TestsP-values• To be statistically significant, the p-value will usually be set to less than 0.05 (p < 0.05)• If the p-value is less than 0.05, then the null hypothesis can be rejected and the alternative hypothesis can be accepted
Statistical Tests95% Confidence Intervals (95% CIs)• A CI consists of a range of values that act as good estimates of the unknown population parameter – Ex: A 95% CI is the interval that you are 95% certain contains the true population value as it might be estimated from a much larger study.• It is used to indicate if a measure is statistically significance of an estimate
Statistical Tests95% Confidence Intervals (95% CIs)• Can tell us about the magnitude of measure – Wide vs. narrow intervals – Wide intervals can indicate small sample or low power – Whether the measure is reliable
Public Cancer Data ResourcesPublic cancer data resources• National Program of Cancer Registries (NPCR) http://www.cdc.gov/cancer/npcr/• Surveillance Epidemiology and End Results (SEER) http://seer.cancer.gov/• National Cancer Data Base (NCDB) http://www.facs.org/cancer/ncdb/• Behavioral Risk Factor Surveillance Survey (BRFSS) http://www.cdc.gov/brfss/• Agency for Healthcare Research and Quality – Libraryhttp://www.ahrq.gov/clinic/ehclibrary/reslibcancer.htm• PubMed http://www.ncbi.nlm.nih.gov/pubmed/