QUANTIFYING THE ISSUE July 22, 2015 Maria Schymura, PhD 2 1 - - PDF document

1

July 22, 2015

Evidence‐Based Public Health: Supporting the New York State Prevention Agenda MODULE 3:

QUANTIFYING THE ISSUE

Maria Schymura, PhD

2

2

Learning Objectives

• 1. To measure and characterize disease

frequency in defined populations using principles of descriptive epidemiology and surveillance.

• 2. To find and use disease surveillance

data presently available on the Internet.

3

Epidemiology

 Study of the distribution and

determinants of health-related states or events in specified populations, and the application of this study to the control of health problems

4

3

Public Health Surveillance

 Ongoing collection and timely analysis,

interpretation, and communication of health information for public health action

 Public health surveillance systems are

important tools for collecting and disseminating descriptive epidemiologic data

5

Public Health Surveillance

Collection methods

Provide varying levels of confidence in the data

Population-based

Vital Statistics

• Birth and death

Reportable diseases Registries

• Birth defects
• Cancer
• Immunizations
• Trauma

Representative Samples

National Health Interview Survey (NHIS) National Health and Nutrition Examination Survey (NHANES) Behavioral Risk Factor Surveillance System (BRFSS) Youth Risk Behavior Survey (YRBS)

Convenience Samples

Survey at a local mall

Level of confidence high low

6

4

BRFSS

 Monitors modifiable risk factors associated with

chronic and communicable diseases

 All 50 states and DC participate  Sample based on the state’s population, not the

population of smaller geographic areas (e.g., counties)

 Adults age 18 yrs and older (non-

institutionalized)

 Random dial telephone survey – Past: landlines – Present and future: landlines (80%) and cell phone (20%)

7

BRFSS

 Raking methodology to be introduced (2011 data) – More precise estimates – Need to start new trend analyses  SMART BRFSS (Metropolitan or Micropolitan Statistical Areas) – Metro

• Lincoln (Lancaster and Seward)
• Omaha–Council Bluffs (Cass, Douglas, Sarpy, Saunders, Washington, plus IA counties)
• Sioux City (Dakota, Dixon, plus IA and SD counties)

– Micro

• Grand Island (Hall, Howard, Merrick)
• Hastings (Adams, Clay)
• Norfolk (Madison, Pierce, Stanton)
• North Platte (Lincoln, Logan, McPherson)
• Scottsbluff (Banner, Scotts Bluff)

 County-level prevalence estimates – Diabetes, obesity, physical activity (link below) – 11 indicators (2014) 8 http://apps.nccd.cdc.gov/BRFSS-SMART/index.asp http://apps.nccd.cdc.gov/DDT_STRS2/CountyPrevalenceData.aspx?StateId=18&mode=DBT

5

Obesity Trends* Among U.S. Adults—1990, 1999, 2008

* Body Mass Index (BMI) 30; or about 30 lbs. overweight for 5’4” person 1 9 9 9 2 0 0 8 1 9 9 0 Source: Behavioral Risk Factor Surveillance System 9

No data <10% 10%-14% 15%-19% 20%-24% 25%-29% ≥30%

Percent of High School Students Considered Obese, United States, 2013

Source: Youth Risk Behavior Survey 10

6

Descriptive and Analytic Epidemiology

 Descriptive epidemiology

– Frequency and distribution of risk factors in populations – Frequency and distribution of disease in populations – Can provide hypotheses for etiologic research

 Analytical epidemiology

– Study of factors associated with disease (factors that either increase or decrease risk)

11

Descriptive and Analytic Epidemiology

Thematic Example: Obesity and Cancer

Cancer site and type Summary RR from comprehensive meta-analysis and (95% CI) per given unit increase in BMI RR Overweight (BMI 25-29) vs BMI <25) RR Obese (BMI ≥ 30) vs BMI <25)

Esophagus (adenocarcinoma) 1.11 (1.07-1.15) per 1 kg/m2 increase in BMI 1.55 2.10 Colorectal 1.18 (1.14-1.21) per 5 kg/m2 increase in BMI 1.18 1.36 Pancreas 1.14 (1.07-1.22) per 5 kg/m2 increase in BMI 1.14 1.28 Kidney 1.42 (1.17-1.72) per 5 kg/m2 increase in BMI 1.42 1.84 Post-menopausal breast 1.05 (1.03-1.07) per 2 kg/m2 increase in BMI 1.13 1.25 Endometrial 1.60 (1.52-1.68) per 5 kg/m2 increase in BMI 1.60 2.20 12

Source: Eheman C, et al Annual Report to the Nation on the Status of Cancer, 1975–2008, Featuring Cancers Associated with Excess Weight and Lack of Sufficient Physical Activity. Cancer 2012; 118:2338-66.

Relative Risk (RR) Associated with Excess Weight

7

Descriptive and Analytic Epidemiology

 Descriptive epidemiology

– Frequency and distribution of risk factors in populations – Frequency and distribution of disease in populations – Can provide hypotheses for etiologic research

 Analytical epidemiology

– Study of factors associated with disease

(factors that either increase or decrease risk)

13

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

14

8

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

15

Prevalence vs. Incidence

 Prevalence is the number of existing

cases of disease in the population during a defined period

 Incidence is the number of new

cases of disease that develop in the population during a defined period

16

9

Prevalence vs. Incidence

Question Are we measuring prevalence or incidence?

 The number of persons living with HIV

in your community as of December 31, 2012

 The number of persons diagnosed with

breast cancer in your community during 2012

17

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

18

10

Incidence vs. Mortality

Question Which data are better for estimating disease rates? incidence or mortality data

19

Incidence vs. Mortality

Mortality rates are used to estimate disease frequency when

 Incidence data are not available;  Case-fatality rates are high; or  Goal is to reduce mortality among

screened or targeted populations

20

11

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

21

Role of Intermediate Outcomes

Intermediate outcomes may be used

 When it is not feasible to wait years to

see the effects of a new public health program, or

 There is sufficient type I evidence

supporting the relationship between modifiable risk factors and disease reduction

22

12

Role of Intermediate Outcomes

Long-term outcomes

 cardiovascular disease  lung cancer  breast cancer mortality  arthritis

Intermediate outcomes

 obesity, physical activity  cigarette smoking  mammography screening  ?

23

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

24

13

Small Number Issues

 Rates are often available for national

and state-wide populations

 Not always available for smaller

geographic areas or demographically defined populations

– Rates are not considered stable if fewer than 20 cases in the numerator

25

10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100

Small Number Issues

Role of standard error

numerator size relative standard error*

*RSE = 1 / cases

26

14

Small Number Issues

Possible solutions, combine…

 Years  Groups

– e.g., “other races”

 Geographic areas

– Public health department regions – Congressional districts – Program regions – School districts

27

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

28

15

Types of Rates

 Crude, or unadjusted  Standardized, or adjusted  Category-specific, or stratified

29

Types of Rates

 Crude, or unadjusted  Standardized, or adjusted  Category-specific, or stratified

30

16

Crude (or unadjusted) Rates

 Estimate the actual disease frequency

for a population

 Can be used to provide data for

allocation of health resources and public health planning

 Can be misleading if compared over

time or across populations

31

Crude (or unadjusted) Rates

Defining your population

 Define disease  Define population at risk  Select time frame

32

17

Crude (or unadjusted) Rates

Defining your population

 Define disease  Define population at risk  Select time frame

33

Breast Cancer

• Standard inclusion and exclusion

criteria (e.g., invasive, specific ICD- O-3 codes)

New York Females 2010 Crude (or unadjusted) Rates

Defining your population Define disease Define population at risk Select time frame Breast Cancer

• Standard inclusion and exclusion

criteria (e.g., invasive, specific ICD-O-3 codes)

New York Females 2010

Where do you find this data?

34

18

Crude (or unadjusted) Rates—

Variability among data sources



New York State Cancer Registry

– Incidence and mortality data

• State, NYC, NYS excl. NYC, counties

– Incidence data

• Select areas for Nassau, Rockland, Suffolk,

& Westchester Counties

• Cities: Albany, Buffalo, Rochester, Syracuse,

& Yonkers

• NYC neighborhoods
• ZIP codes

– http://www.health.ny.gov/statistics/cancer/registry/ 

New York State Vital Statistics

– Mortality data, birth data

– http://www.health.ny.gov/statistics/vital_statistics/ 

CDC Wonder

– Incidence and mortality data – National, state, regional and metropolitan statistical area (4 for NY) levels – wonder.cdc.gov – Cancer statistics:

http://wonder.cdc.gov/cancer.html 35

Crude (or unadjusted) Rates— Different options for population figures

 U.S. Census

– Decennial Census (most current: 2010) – American Community Survey (yearly estimates) – factfinder2.census.gov

 NYS Vital Statistics

– http://www.health.ny.gov/statistics/vital _statistics/

 SEER Program (Surveillance, Epidemiology and End Results)

– seer.cancer.gov/popdata

 CDC Wonder

– wonder.cdc.gov

36

19

Crude (or unadjusted) Rates

Calculation methodology

Compute disease rate for year 2010 Number of females diagnosed with breast cancer Number of females at risk for breast cancer

14,409 10,007,823

Sources: New York State Cancer Registry; CDC Wonder 37

Crude (or unadjusted) Rates

Calculation methodology

Compute disease rate for year 2010

14,409 New York females with breast cancer 10,007,823 female New York residents = 0.0014398 breast cancer cases / female NY residents / year

38

20

Crude (or unadjusted) Rates

Calculation methodology

Rates are usually expressed as whole numbers for populations at risk during specified periods:

0.0014398 breast cancer cases / female New York residents / year x 100,000 = 144.0 breast cancer cases / 100,000 female New York residents / year

39

Types of Rates

 Crude, or unadjusted  Standardized, or adjusted  Category-specific, or stratified

40

21

Standardized (or adjusted) Rates

– Removes the impact of different age distributions (or other factors) among populations – Allows for direct comparisons of those populations – Types of age standardization

• One population to another
• Using the 2000 U.S. Standard Million or

Standard Population [right]

– Multiple age category breakdowns, with 18 and 19 categories being the most common

41

Standardized (or adjusted) Rates

Example calculation from one population to another

Age (years) Deaths Persons Rate* Deaths Persons Rate* ≤29 1 100 10 20 1000 20 30–59 25 500 50 50 500 100 ≥60 100 1000 100 20 100 200 Total 126 1600 79 90 1600 56

Group A Group B * per 1,000 population per year

42

22

Standardized (or adjusted) Rates

Example calculation from one population to another

Age (years) Deaths Persons Rate* Deaths Persons Rate* ≤29 1 100 10 20 1000 20 30–59 25 500 50 50 500 100 ≥60 100 1000 100 20 100 200 Total 126 1600 79 90 1600 56

Group A Group B * per 1,000 population per year

43

Age (years) Deaths Persons Rate* Deaths Persons Rate* ≤29 1 100 10 20

1,000 100

20 30–59 25 500 50 50

500 500

100 ≥60 100 1000 100 20

100 1000

200 Total 126 1600 79 90 1600 56

Group A Group B

Standardized (or adjusted) Rates

Example calculation

* per 1,000 population per year

44

23

Age (years) Deaths Persons Rate* Deaths Persons Rate ≤29 1 100 10 20 100 x 0.02 = 30–59 25 500 50 50 500 x 0.10 = ≥60 100 1000 100 20 1000 x 0.20 = Total 126 1600 79 90

Group A Group B *Expected number of deaths based on Group A’s population distribution

Standardized (or adjusted) Rates

Example calculation

Exp* 2 50 200 252

45

Standardized (or adjusted) Rates

Age-adjusted mortality rate for Group B = (expected number of deaths / total population at risk) x 10n = (252 deaths / 1,600 persons / year) x 1,000 = 158 deaths / 1,000 persons / year

• Crude rate: 56 deaths / 1,000 persons / year

Mortality rate for Group A = 79 deaths / 1,000 persons / year

46

24

Standardized (or adjusted) Rates

Number of cancer cases, by age—New York, 2010

47

Source: CDC Wonder

2,000 4,000 6,000 8,000 10,000 12,000 14,000

Number of cases Age (years)

Standardized (or adjusted) Rates

Example calculation using a standard population; all cancers—New York, 2010

48 Age Group Count Population < 1 year 60 232,326 1-4 years 199 922,094 5-9 years 148 1,161,832 10-14 years 189 1,210,497 15-19 years 319 1,360,440 20-24 years 535 1,413,617 25-29 years 834 1,381,867 30-34 years 1,337 1,284,958 35-39 years 1,910 1,247,713 40-44 years 3,322 1,356,033 45-49 years 5,723 1,453,537 50-54 years 8,525 1,422,576 55-59 years 11,183 1,244,763 60-64 years 13,688 1,076,583 65-69 years 13,760 776,843 70-74 years 12,520 589,875 75-79 years 11,216 473,976 80-84 years 9,662 391,910 85+ years 9,068 393,766 All Ages 104,198 19,395,206

Source: CDC Wonder

25

Age Group Count Population Crude Rate < 1 year 60 232,326 1-4 years 199 922,094 5-9 years 148 1,161,832 10-14 years 189 1,210,497 15-19 years 319 1,360,440 20-24 years 535 1,413,617 25-29 years 834 1,381,867 30-34 years 1,337 1,284,958 35-39 years 1,910 1,247,713 40-44 years 3,322 1,356,033 45-49 years 5,723 1,453,537 50-54 years 8,525 1,422,576 55-59 years 11,183 1,244,763 60-64 years 13,688 1,076,583 65-69 years 13,760 776,843 70-74 years 12,520 589,875 75-79 years 11,216 473,976 80-84 years 9,662 391,910 85+ years 9,068 393,766 All Ages 104,198 19,395,206 0.005372

0.005372 x 100,000 = 537.2 cases / 100,000 people (crude rate) / =

49

Standardized (or adjusted) Rates

Example calculation using a standard population, all cancers—New York, 2010

Source: CDC Wonder

50

Age Group Count Population Crude Rate US Standard Million < 1 year 60 232,326 13,818 1-4 years 199 922,094 55,317 5-9 years 148 1,161,832 72,533 10-14 years 189 1,210,497 73,032 15-19 years 319 1,360,440 72,169 20-24 years 535 1,413,617 66,478 25-29 years 834 1,381,867 64,529 30-34 years 1,337 1,284,958 71,044 35-39 years 1,910 1,247,713 80,762 40-44 years 3,322 1,356,033 81,851 45-49 years 5,723 1,453,537 72,118 50-54 years 8,525 1,422,576 62,716 55-59 years 11,183 1,244,763 48,454 60-64 years 13,688 1,076,583 38,793 65-69 years 13,760 776,843 34,264 70-74 years 12,520 589,875 31,773 75-79 years 11,216 473,976 26,999 80-84 years 9,662 391,910 17,842 85+ years 9,068 393,766 15,508 All Ages 104,198 19,395,206 0.005372 1,000,000

Standardized (or adjusted) Rates

Example calculation using a standard population, all cancers—New York, 2010

Source: CDC Wonder

26

Age Group Count Population Crude Rate US Standard Million Expected

< 1 year 60 232,326 0.000258 13,818 4 1-4 years 199 922,094 0.000216 55,317 12 5-9 years 148 1,161,832 0.000127 72,533 9 10-14 years 189 1,210,497 0.000156 73,032 11 15-19 years 319 1,360,440 0.000234 72,169 17 20-24 years 535 1,413,617 0.000378 66,478 25 25-29 years 834 1,381,867 0.000604 64,529 39 30-34 years 1,337 1,284,958 0.001041 71,044 74 35-39 years 1,910 1,247,713 0.001531 80,762 124 40-44 years 3,322 1,356,033 0.002450 81,851 201 45-49 years 5,723 1,453,537 0.003937 72,118 284 50-54 years 8,525 1,422,576 0.005993 62,716 376 55-59 years 11,183 1,244,763 0.008984 48,454 435 60-64 years 13,688 1,076,583 0.012714 38,793 493 65-69 years 13,760 776,843 0.017713 34,264 607 70-74 years 12,520 589,875 0.021225 31,773 674 75-79 years 11,216 473,976 0.023664 26,999 639 80-84 years 9,662 391,910 0.024654 17,842 440 85+ years 9,068 393,766 0.023029 15,508 357 All Ages 104,198 19,395,206 1,000,000 4,821 x x x x x x x x x x x x x x x x x x x = = = = = = = = = = = = = = = = = = =

51

Standardized (or adjusted) Rates

Example calculation using a standard population; all cancers—New York, 2009

Source: CDC Wonder

Count Crude Rate* Standardized Rate*† All Sites Combined 104,198 537.2 482.1 Prostate 14,680

156.4

147.6 Female Breast 14,409

144.0

123.6 Lung and Bronchus 13,301

68.6

61.5 Colon and Rectum 9,311

48.0

42.9 Urinary Bladder 4,792

24.7

22.1 Non‐Hodgkin Lymphoma 4,519

23.3

21.1 Thyroid 3,643

18.8

17.8 Melanoma of the Skin 3,487

18.0

16.5 Corpus Uteri 3,464

34.6

28.7 Kidney and Renal Pelvis 3,387

17.5

15.6

Standardized (or adjusted) Rates

Comparison between crude and standardized rates for the ten leading cancer types—New York 2010

* per 100,000 population per year; †Standardized using the 2000 U.S. Standard Million

52

Source: CDC Wonder

27

Standardized (or adjusted) Rates

Changes in age distribution—United States and New York

53

Source: CDC Wonder

0% 1% 2% 3% 4% 5% 6% 7% 8% 9%

Age (years) New York 2000 New York 2010 US Standard Million

c4

Types of Rates

 Crude, or unadjusted  Standardized, or adjusted  Category-specific, or stratified

54

Slide 53 c4 At first glance, I'm not sure what the line graphs represent.

cthomaskutty, 5/9/2012

28

Category-specific (or stratified) Rates

 Can be used for valid comparison of

populations

 Can be cumbersome if there is a large

number of categories to compare

55

Category-specific (or stratified) Rates

Two general categories

 Age-specific: crude rates across different

age groups

 “Other”-specific: crude or standardized

rates across different groups

• Person: sex, race / ethnicity, education,

income, health insurance status

• Place: geographic unit (e.g., county), urban /

rural, population density

• Time: short or long-term trends, cyclic trends,

cohort effects

56

29

Category-specific (or stratified) Rates

Age-adjusted Rates for Colorectal Cancer, Both Males and Females, by County, New York, 2008-2012

57

Source: New York State Cancer Registry

Category-specific (or stratified) Rates

Age-Adjusted Colorectal Cancer Rates by Race/Ethnicity and Gender, New York, 2008–2012

10 20 30 40 50 60 Male Female White Non-Hispanic Black Non-Hispanic Hispanic Asian/PI

58

Source: New York State Cancer Registry

30

Category-specific (or stratified) Rates

Trends in Colorectal Cancer Incidence by Race/Ethnicity, New York, 1990-2012

10 20 30 40 50 60 70 Age-adjusted rate Year of diagnosis Non-Hispanic White Non-Hispanic Black Non-Hispanic Asian/Pacific Islander Hispanics

Category-specific (or stratified) Rates

Female Breast Cancer Rates* for Northeastern States, 2007-2011

60

Crude Age-Adjusted State Rate Rank Rate Rank Connecticut 165.2 1 136.6 1 Maine 164.7 2 126.4 9 Massachusetts 159.6 5 135.6 2 New Hampshire 161.8 3 134.1 3 New Jersey 152.4 8 129.6 5 New York 149.0 9 128.5 7 Pennsylvania 158.3 6 126.8 8 Rhode island 156.2 7 130.1 4 Vermont 161.4 4 129.1 6

Source: CDC Wonder & SEER Public Use File

*Rates are per 100,000

31

New York State Community Health Indicator Reports

61

New York State Community Health Indicator Reports

62

32

63

Source: 2010-2012 Vital Statistics Data as of February 2014 Source: 2010-2012 SPARCS Data as of June, 2014

Cardiovascular Disease Mortality Rate*, 2010-2012

(* per 100,000 Adjusted to 2000 US Population)

Cardiovascular Disease Hospitalization Rate*, 2010-2012

(*per 10,000 Adjusted to 2000 US Population) 64

Source: 2010-2012 Vital Statistics Data as of February 2014 Source: 2010-2012 SPARCS Data as of June, 2014

Diabetes Mortality Rate*, 2010-2012

(* per 100,000 Adjusted to 2000 US Population )

Diabetes Hospitalization Rate* (primary diagnosis), 2010-2012

(* per 10,000 Adjusted to 2000 US Population)

33

Descriptive Epidemiology Terminology and uses

 Prevalence vs. incidence  Incidence vs. mortality  Role of intermediate indicators  Small number issues  Types of rates  Estimate error and confidence intervals

65

Estimate Error and Confidence Intervals (CI)

Population-based

Vital Statistics

• Birth and death

Reportable diseases Registries

• Birth defects
• Cancer
• Immunizations
• Trauma

Representative Samples

National Health Interview Survey (NHIS) National Health and Nutrition Examination Survey (NHANES) Behavioral Risk Factor Surveillance System (BRFSS) Youth Risk Behavior Survey (YRBS)

Convenience Samples

Survey at a local mall

Level of confidence high low

66

34

Estimate Error and Confidence Intervals (CI)

Population-based Representative Sample Subject to sampling error? No Yes Impacted by random variation? Yes, especially when looking at rates for rare events or among small geographic areas Yes CIs* used to describe the range

• f that variation?

Yes, random variation Yes, both

*95% CIs are typically calculated to provide a range of values in which if one repeated a study 100 times, 95 of the intervals would include the true rate

67

Category-specific (or stratified) Rates

Age-adjusted Rates for Male Stomach Cancer, by County, New York, 2008-2012

68

Source: New York State Cancer Registry

35

Public Health Surveillance Loop

69

Data Program Interpretation Evaluation Data Information Program Analysis Dissemination Implementation Data Program Collection Planning