Descriptive Epidemiology

1
Descriptive Epidemiology Rates -- Crude,
Specific, and Adjusted
Robert Heimer Yale University School of Public
Health April 2009
2
Overview for Today

• What is descriptive epidemiology
• Person, place and time
• Goals
• Examples breast cancer and malaria
• Useful measures in descriptive epidemiology
• Crude, specific and adjusted rates
• Standardized mortality ratios

3
Descriptive Epidemiology

• Includes descriptive statistics that provide
  information on disease patterns by various
  characteristics
• The three most obvious characteristics are
• Person
• Place
• Time.

4
Purpose

• Provide clues about disease causation and
  prevention that are usually investigated further
  in formal studies that test hypotheses.
• Assess the health status of a population.
• Allocate resources efficiently and target
  populations for education and prevention.

5
Descriptive statistics

• Routinely collected data from a variety of
  sources including
• Mortality and births from vital records
• Reportable diseases from surveillance programs
• Cancer data from registries
• Other diseases from national surveys
• Many, many others

6
Person

• Who has the disease?
• Male versus female?
• Young versus older?
• Ethnic majority versus minority?
• Rich versus poor?
• Urban versus rural?

7
Breast Cancer by Age
Age is often an important personal
characteristics, especially for chronic diseases
8
Breast Cancer by Sex

• In 2004, in the U.S.
• 186,772 women and 1,815 men were diagnosed with
  breast cancer
• 40,954 women and 362 men died from breast cancer

9
Breast Cancer Incidence by EthnicityU.S.,
1975-2004

• White women had the highest incidence rate for
  breast cancer. Black women had the second highest
  incidence of getting breast cancer, followed by
  Asian/Pacific Islander, Hispanic, and American
  Indian/Alaska Native women.

10
Breast Cancer Mortality by Ethnicity

• Despite lower incidence, black women were more
  likely to die of breast cancer than any other
  group.
• White women had the second highest rate of deaths
  from breast cancer, followed by women who are
  Hispanic, American Indian, and Asian/Pacific
  Islander.

11
Place

• Where is the disease more or less common?
• Different scales of geography
• Regions of earth, countries, states, counties,
  cities, neighborhoods
• May not include political boundaries
• Urban/rural, proximity to power lines, chemical
  plants, etc.

12
Breast Cancer IncidenceBy Country, 2005
13
(No Transcript)
14
Time

• Is the disease rate changing over time?
• Different scales here, too decades, seasons,
  days

15
Breast cancer incidence and mortality, time trends
Incidence
Mortality
16
International Example of Time Trends
Women 65 years
Women 50-64 years
Women 30-49 years
17
A Cautionary Note on Time Trends

• Might infer that some cause of the disease is
  also changing over time.
• Or, it could be an artifact (not real)
• Definition changing (AIDS)
• Diagnostic capabilities are changing (breast
  cancer)
• Screening is changing (Chlamydia, HIV in Russia)
• Reporting is changing (MDR-TB)
• Population at risk is changing (lung cancer)
• Etc.

18
Uses of Descriptive Epidemiology

• Hypothesis generation
• In the US, why is incidence higher among whites
  but mortality higher among blacks?
• Why is breast cancer incidence higher in
  industrialized nations? Is it because of diet?
  Lifestyle? Exposure to electromagnetic fields?
• In the US, why is incidence increasing (Better
  detection? More risk?) and mortality decreasing
  (Better treatment? Earlier diagnosis?)?
• In Czech republic, why is mortality decreasing
  more among young women? Earlier detection,
  changes in lifestyle, diet, environmental
  factors? If so, why?
• Public health planning
• Need for better access to treatment overall and
  among blacks and elderly in particular
• Need for better detection in developing nations

19
Another Descriptive Epidemiology Example Malaria

• Worlds most important tropical parasitic disease
• Mosquito-borne, caused by 4 species of protozoan
  parasite Plasmodium
• P. falciparum most virulent
• Clinical illness in 300-500 million people per
  year
• gt1 million deaths in 2000
• Spectrum of disease
• Asymptomatic
• Mild, non-specific febrile illness
• Severe disease results in cerebral complications,
  coma, shock, renal failure, pulmonary edema,
  respiratory distress, anemia, etc.

20
(No Transcript)
21
Trinity of Infectious Diseases
HOST
Sometimes, need a second triangle to describe
extrinsic life cycle of agent outside human host.
Alternate host/reservoir/vector
AGENT
ENVIRONMENT
Host diseased person Agent virus, parasite,
bacteria, fungi Environment setting in which
transmission occurs
Vector Environment
For malaria Host factors genetic factors Agent
factors escape mechanisms to evade host immune
response Environment human-mosquito contact
patterns
Adapted from Nelson KE et al. Infectious Disease
Epidemiology Theory and Practice 2001.
22
Person

• Children under age 5 and women in their first
  pregnancy are most vulnerable

Anemia in young children in a highly endemic area
in Kenya occurs most between 6 and 24 months.
After 24 months, anemia decreases because the
children have built up their acquired immunity
against malaria.
Pregnancy decreases immunity against many
infectious diseases. Women who have developed
protective immunity against P. falciparum tend to
lose this protection when they become pregnant
(especially during the first and second
pregnancies).
23
Place

• Where malaria is found depends mainly on climatic
  factors such as temperature (gt68F), humidity,
  and rainfalls.
• Malaria is transmitted in tropical and
  subtropical areas -- 90 of global burden of
  disease is in sub-Saharan Africa.

24
Place (continued)

• Even within tropical and subtropical areas,
  transmission will not occur
• At high altitudes
• During cooler seasons in some areas
• In deserts (excluding the oases)
• In some islands in the Pacific Ocean, which have
  no local Anopheles species capable of
  transmitting malaria
• In countries where transmission has been
  interrupted through successful eradication.

25
Time Seasonality
26
Seasonality (continued)
27
Uses of Descriptive Information

• Determining sufficient resources in sub-Saharan
  Africa
• Diagnostic and treatment supplies
• Health education
• Training of health care personnel
• Prevention targeted to children and pregnant
  women
• Bednets
• Anti-malarial drugs for intermittent preventive
  treatment
• Insecticide spraying in homes

28
Crude, Specific and Adjusted Rates

• Crude rates
• A summary measure calculated by dividing the
  total number of cases in the population by the
  total number of individuals in that population at
  a specified time period
• Category specific rates
• Rates specific to some particular sub-population
  for example, ethnicity-specific or sex-specific
•  Problems comparing crude rates among
  populations
• Groups differ with respect to underlying
  characteristics that affect overall rate of
  disease (for example, age) and so you may be
  making an unfair comparison
• Things may be more different than same

29
Race Comparison of All-Cause Mortality Rates
among Females, U.S., 1998
Crude mortality rate for whites 903.7 deaths
per 100,000 pop per year Crude mortality rate for
blacks 746.4 deaths per 100,000 pop per year
30
Age Distribution of White and Black Females, 1998
They are different black population has a
younger age distribution.
31
Age Distribution of White and Black Females,1998
32
Comparing Rates in Two or More Populations

• How do we get around this problem of differences
  in age structures?
• The difference in the age structure between the 2
  populations of black women and white women make
  this an unfair comparison
• We can compare age-specific rates, but this will
  be cumbersome.
• Calculate a summary rate for each population
• age-adjusted or age-standardized rate
• uses a procedure designed to minimize the effects
  of differences in age composition
• Age-adjustment can be used for rates, risks, or
  prevalences.

33
Age-Adjustment or Standardization of
Rates(Direct Method)

• We want a summary number for both whites and
  blacks that allows for comparison with each other
  accounting for differences in age.
• These estimates will answer the question What
  would the death rate be in each racial group if
  the populations in each group had identical age
  distributions?
• We can answer this question by forcing the two
  populations to have the same age distribution as
  a standard population.
• What age distribution?
• Sometimes the population in a census year is
  used.
• Another good choice is the combined populations
  under study.

34
Direct Method Continued

• Method overview
• Compute a weighted average of the age-specific
  rates.
• Weights the age distribution of a specified
  standard population.
• Apply the weights to create age-adjusted
  mortality rates
• Adjusted mortality rates -- what the populations
  of interest would have experienced if they each
  had the same age-distribution as the standard
  population.
• Summary rate that accounts for age difference
  between populations.
• Any differences between age-adjusted rates cannot
  be attributed to age.

35
Standard Populations

• U.S. population in 1940, 1970, 2000
• Soviet census in 1989 Russian census in 2002
• Planned census of U.S. (2010) and Russia (2010)
• World population
• European population
• Combined populations (if convenient)

36
Standard Population for Any One Million
People1940 U.S. Population
37
Age-Adjusted, All-Cause Mortality Rate
For U.S. White Women Unadjusted (Crude)
Rate 903.7
For U.S. Black Women Unadjusted (Crude)
Rate 746.4
38
Alternative Method

• Mathematically equivalent to
• Estimate age-specific rates in each population
• Apply age-specific rates for each age group in
  each population to the standard population to
  estimate the expected number of deaths in the
  standard population at each populations rate
• (standard population) (rate)
• Divide the expected deaths in each population by
  the standard population total to get the
  age-adjusted rate for each population

39
Explaining Age-Adjustment

• Mortality rates rise steeply with age.
• 0-34 years -- 100 deaths/100,000 women/year
• 35-54 years -- 200 deaths/100,000 women/year
• 55-74 years -- 1,500 deaths/100,000 women/year
• 75 years -- 8,000 deaths/100,000 women/year
• White females had a older age distribution than
  black females.
• White females had a higher mortality rate than
  black females simply because they were older.
• Age confounded the relationship between race and
  mortality
• This is a common problem because age is a major
  determinant of disease and mortality

40
Comments on Age-Adjusted Rates

• Standard population is arbitrary
• Does not measure the rate in the standard
  population
• Allows us to compare mortality rates across
  populations with different age structures if
  each population had the age distribution of the
  standard population, how would their rates
  compare?
• An age-adjusted rate for a single population is
  not a meaningful quantity.
• Hypothetical applies to no one
• Age-adjustment only takes on meaning when 2 or
  more populations or groups are being compared

41
Standardized Mortality Ratio(Indirect Method)

• Purpose is to compare mortality experience
  (rates) of an exposed group with that of general
  population.
• Used in occupational epidemiology to answer the
  question Do people who work in a certain
  industry have higher mortality than people of the
  same age in the general population?
• Can be used if you do not know the numbers of
  deaths for each age-specific stratum in
  population of interest.
• Involves computation and comparison of expected
  and observed mortality counts.

42
Computing a Standardized Mortality Ratio

• Age-specific rates for a known population
  (usually the general population) are applied to
  the population of interest
• Yields number of expected deaths in population of
  interest if this population had the mortality
  experience of the known population
• The ratio of total number of observed to expected
  is calculated.
• SMR observed deaths
• expected deaths

SMR for tuberculosis in miners aged 20-59, US,
1950
SMR 436/181.09 2.41 Conclude the observed
number of deaths exceeds the expected.
43
Summary

• Descriptive epidemiology is interesting and
  useful.
• Useful measures in descriptive epidemiology
  include crude, specific, and adjusted rates
• Can be for incidence or mortality
• Adjustments can be for age, sex, race/ethnicity,
  time, etc to make things more same than
  different for fair comparisons