Using and Interpreting Data Substance Abuse Epidemiology Unit Epidemiology and Response Division New Mexico Department of Health
Outline Measures of relative frequency Calculating rates Examples Age-adjustment Trends Small group exercises Sources of data Review of the New Mexico State Epi Profile Community examples
Ratios Ratio: a comparison of two groups Groups may be unrelated or subgroups of a larger category Ratio = A / B Examples: miles / gallon students / teacher males / females (sex ratio)
Proportions Proportion: a relationship of one part to the whole, e.g. percentage, fraction, decimal Percentage = __A__ x 100 A + B Numerator is always included in the denominator Examples: ______# of females in class______ # of females + # of males in class ___# of alcohol-related deaths in Sandoval county_____ Total # of alcohol-related deaths in all 33 NM counties
Calculating Proportions - Example Proportion of alcohol-related (A-R) deaths in Sandoval county Numerator = # A-R deaths in Sandoval county = 213 Denominator = # A-R deaths in New Mexico = 5,068 Time Period = Constant = 100 Proportion of statewide A-R = __213_ x 100 = 4.2% deaths in Sandoval county 5,068
Prevalence Prevalence = the number of existing cases or events in a certain population at a given point in time Prevalence is a proportion and can be expressed as a percentage Includes current cases/events Must indicate WHEN cases were enumerated Prevalence = # of existing cases = 15_ = 0.10 X 100 or 10% Total population 150
Prevalence- Example Youth Binge Drinking Chart 2: Binge Drinking by Sex and Grade Level, 2003 NM YRRS
Rates Frequency of a defined event in a specified population during a given time period Incorporates time into the measure Multiplied by a constant for ease of interpretation Rate = # events (deaths, cases, etc.) _______# people at risk_____ time Example: crude death rate
Calculating Rates - Example 2002 all cause crude death rate for NM Numerator = # of deaths in NM during 2002 = 14,344 Denominator = total population of NM in 2002 = 1,853,030 Time Period = 2002 Constant = 100,000 Crude death rate = __14,344__ X 100,000 = 774.1/100,000 1,853,030 Source: CDC Wonder
Age-adjustment of mortality rates Death rate – number of deaths occurring in a specified population during a given period of time Crude death rate – the death rate in the total population Adjusted death rate – recalculation of the death rate using a standard population reference Compensates for differences in the age distributions of populations being measured Allows comparisons between groups
Age-adjustment - Example Crude death rate ( ) Mexico: 469.6/100,000 US: 869.6/100,000 Adjusted death rate ( ) Mexico: 596.6/100,000 US: 519.1/100,000
Trends Looking at the same data over time Data collected the same way over periods of time Numbers large enough to calculate percentages or rates for each time period Same length of time in each period used as a data point
Source: Division of Government Research, University of New Mexico Alcohol-Related Motor Vehicle Crash Fatality Rates, New Mexico and US,
Drug-Related Death Rates* New Mexico and US, *Rates are age-adjusted to the 2000 US population Sources: Bureau of Vital Records and Health Statistics, NMDOH; CDC Wonder
Small Group Exercises
New Mexico Crude Death Rates By County, 2002 Source: CDC Wonder
New Mexico Age-Adjusted Death Rates By County, 2002 Source: CDC Wonder
Common Sources of Health Data Population and demographic data from the US Census Bureau and the Bureau of Business and Economic Research at UNM Birth and death data from the NM Bureau of Vital Records and Health Statistics Population-based survey data collected by the NM DOH YRRS: Youth Risk and Resiliency Survey BRFSS: Behavioral Risk Factor Surveillance System Hospitalization in-patient discharge data from the NM Health Policy Commission
New Mexico SPF-SIG State Epi Profile
Provides a systematic and comprehensive overview of ATODA-related consequences …. NM SPF-SIG State Epi Profile
… and ATODA-related consumption
How to use this report Outcome indicators: consequences Problem statement presents overview of the data and detailed statistics Outcome indicators – associated with two tables Number of deaths on the left side of the table Age-specific death rates per 100,000 population on the right side of the table* County bar chart showing age-adjusted rates for each NM county in descending order *Note: All-ages rate is per 100,000, age-adjusted to the 2000 US population
How to use this report Outcome indicators – Table 1 Table 1: deaths and death rates by sex, age group, and race/ethnicity Useful in determining the most important risk groups at a statewide level
How to use this report Outcome indicators – Table 1
How to use this report Outcome indicators – Table 2 Table 2: deaths and death rates for each NM county by race/ethnicity Useful in determining: Counties with the most severe substance abuse problems Groups with the highest risk within each county
How to use this report Outcome indicators – Table 2 *All rates are per 100,000, age-adjusted to the 2000 US population
How to use this report Outcome indicators – County bar charts Number of deaths and the percent of NM deaths occurring in each county are given next to the county name on the left side Highest death rates are at the top State rate shown with a darker bar for comparison
Example of county bar chart Alcohol-Related Chronic Liver Disease Death McKinley county AR-CLD death rate: 39.3/100,000 McKinley county deaths: 119/1,304 = 9.1% of AR-CLD statewide deaths New Mexico AR-CLD death rate: 14.4/100,000 *All rates are per 100,000, age-adjusted to the 2000 US population
How to use this report Rates and numbers Example: McKinley and Bernalillo counties Consider both prevalence and rate when designing interventions County with highest alcohol-related death rate deaths / 100,000 population County with highest proportion of alcohol-related deaths in the state 1,491/ 5,068 = 29.5%
Example of county bar chart Alcohol-Related Death Rates by County McKinley county alcohol-related death rate: 115.1/100,000 New Mexico alcohol-related death rate: 56.7/100,000 Bernalillo county 1494/5,068 = 29.5% of AR-CLD statewide deaths *All rates are per 100,000, age-adjusted to the 2000 US population
How to use this report ATODA consumption behaviors BRFSS data estimate number and percent of people in population engaging in behavior Table 1: Statewide estimates by age, sex, and race/ethnicity Table 2: County estimates by race/ethnicity County bar charts arranged in descending order
Example of county bar chart: BRFSS Adult Drinking and Driving Rates by County Luna county: 4.8% of adults reported drinking and driving at least once in past 30 days No estimates available for small counties Bernalillo county accounted for 47.8% of statewide drinking drivers *Estimate of percent of people in population group who drove after drinking at least once in previous 30 days
How to use this report ATODA consumption behaviors YRRS data estimate percent of public high school students engaging in behavior Table 1: Prevalence estimates by grade, sex, and race/ethnicity Chart 1: County bar charts with prevalence estimates arranged in descending order
Drinking and Driving Rates by County, Grades 9-12 Example of county bar chart: YRRS Union county: 35.7% of youth reported drinking and driving in the past 30 days New Mexico: 19.1% of youth
Survey Data BRFSS Telephone survey of adult health conditions and risk behaviors Random sample of adults 18 years of age or older in households with a land-line telephone Able to generate population-based estimates for adults YRRS School-based survey of health risk and resiliency behaviors among 9 th -12 th graders in NM School districts must agree to participate Estimates representative of public high school students Both surveys include self-reported data
Adult Drinking and Driving Rates by County Alcohol-Related MVC Death Rates by County
How to use this report Missing data Rates calculated using small numbers are unstable and difficult to interpret Exclusions of rates occurred if: Fewer than four deaths and population <20 in the rate denominator ( NM standard small numbers rule ) Fewer than two deaths per county per year Results in data gaps for certain groups
NM standard small numbers rule Examples of cells affected by additional SPF-SIG rule
Data Collection Death Hospitalization Ambulatory Not reported in any system Think about data availability as a pyramid Numerator – what are you counting? Denominator – who is in the target population? Are the data already available? If not, can they be collected in a systematic way?
Data Collection Numerator: Counting DWI convictions in 2004 Denominator: Determining the population at risk Example 1 – 2004 DWI Conviction Rate for New Mexico
Example DWI Conviction Rate for NM Numerator = # of DWI convictions in NM in 2004 =12,639 Denominator = # of licensed drivers in NM in 2004 =1,289,089 Time Period = 2004 Constant = 1,000 DWI Conviction Rate NM 2004 = _12,639_ X 1,000 1,289,089 = 9.80/1,000 Source: New Mexico Department of Transportation, Driving While Impaired in NM, 2004 Report
Data Collection Numerator: Counting FAS cases in NM Denominator: Determining the population at risk Example 2 – Fetal Alcohol Syndrome (FAS) Rate in New Mexico
Diagnosis of FAS Documentation of 3 dysmorphic facial features Prenatal or postnatal growth deficit in height or weight CNS abnormality Diagnosis classified on the basis of available history of confirmed or unknown prenatal alcohol exposure
Mothers Who Drank Alcohol Three Months Prior to Pregnancy and During the Last Three Months of Pregnancy New Mexico, Source: Pregnancy Risk Assessment & Monitoring System (PRAMS), NMDOH
Example 2 FAS Prevalence Rate in NM Numerator = # FAS-affected children born Denominator = # live births in NM Time Period = Constant = 10,000
Data Collection - Discussion What are some other substance use consequences in your community? Are there data already available? If data need to be collected, how would you define a case/event (numerator) and the population at risk (denominator)? Who can you call for help?
Acknowledgments Dan Green, Social Indicator Epidemiologist Jim Roeber, Alcohol Epidemiologist Substance Abuse Epidemiology Unit Corazon Halasan, Community Epidemiologist Community Health Assessment Program
Contact Information Tierney Murphy Substance Abuse Prevention Epidemiologist New Mexico Department of Health Phone:
Omitted Slides
Data Collection Example 3 – School Truancy Rate in Your Community Numerator: Counting middle and high school students who have unexcused absences Denominator: Determining the population at risk
* 5 or more drinks on one occasion Source: New Mexico BRFSS, Injury and Behavioral Epidemiology Unit, NMDOH Binge Drinking* Among Currently Drinking Adults by Age, New Mexico
Adult Binge Drinking by County Alcohol-Related MVC Death Rates by County
Calculating Rates Numerator: Counting liquor outlets in 2002 Denominator: Determining the population at risk Example 2 – 2002 liquor license density rate for Rio Arriba county
Calculating Rates – Example 3 Liquor License Density Numerator = # licensed facilities in Rio Arriba county Denominator = 2002 population of Rio Arriba county aged 20 years and older Time Period = 2002 Constant = 1,000 Liquor License Density Rate Rio Arriba County 2004 = 2.7 Source: New Mexico Alcohol and Gaming Division, Regulation and Licensing Department
Incidence Rate (Risk) Incidence = the number of NEW cases/events in a population over a given period of time Measures the probability of an event/case occurring during a period of time IR = Number of NEW events during time period Total population at risk for event
* Alcohol-related deaths are deaths from causes considered to be 100% attributable to alcohol. Sources: Bureau of Vital Records and Health Statistics, NMDOH; CDC Wonder Rates are age-adjusted to the 2000 US Standard Population Alcohol-Related* Death Rates New Mexico and US,
NM’s standard small numbers rule Examples of cells affected by additional SPF-SIG rule
Data Collection – General Principles Numerator – what are you counting Denominator – who is in the target population Is the data already available? If not, can it be collected in a systematic way?