Lecture 3: Measuring the Occurrence of Disease Reading: Gordis – Chapter 3 Lilienfeld and Stolley – Chapter 4 Chapter 6, pp. 101-105, 109-117

Counting cases One’s knowledge of science begins when he can measure what he is speaking about and express it in numbers Lord Kelvin (1824-1907) To examine the transmission of disease in human populations, we need to be able to measure the frequency of disease occurrence and of deaths from the disease.

Measures How do we express the extent of morbidity and mortality resulting from disease? Counts Ratio a fraction with no specified relationship Proportions what fraction of the population is affected Rates how fast things are occurring

Measures Measures of morbidity Prevalence: a proportion Cumulative incidence: a rate Incidence density: a rate Measures of mortality Mortality rate: a rate Standardized mortality (SMR)

Defining case Natural course of disease Exposure  onset  symptoms  dx  outcome Recovery Death Chronic disease Incubation period Clinical Stage Subclinical Stage

Counts Prerequisite for epidemiologic investigation Simplest measure of disease frequency Frequency of affected individuals Useful for planning adequacy of health care allocation at a particular level For example: Number of West Nile virus cases

Ratio A fraction with no specified relationship between numerator and denominator Range: 0 to  A/B Examples sex ratio (M:F)

Ratio Number of men with syphilis, 1991 2,412 Number of women with syphilis. 1991 2,314 Ratio of male to females 2,412/2,314 = 1.04 (The numerator is not included in the denominator)

Proportion Type of ratio Numerator included in denominator May be expressed as percentage Percentage = proportion x 100 % Range: 0 to 1 A/(A+B) Example Prevalence

Prevalence All individuals with a disease at a given point in time Dimensionless - should not be described as a rate - may be described as a percent number of cases (A) today P = total population (A+B) today

Prevalence Proportion of individuals in a population who have the disease or condition of interest at a specific time period Utility Describe health burden of a population Status of disease in a population Estimate the frequency of exposure Project health care needs of affected individuals

Types of prevalence Point prevalence – proportion of all cases at a specific point in time Period prevalence – proportion of all cases during a period of time

Point and period prevalence Point prevalence Do you currently have asthma? Period prevalence Have you had asthma during the last five years? Every person in the numerator had the disease at some time during the period specified. Period prevalence consists of the point prevalence at the beginning of a specified period of time plus all new cases that occur during that period.

Rate A special type of proportion Unit of time in denominator A/(A+B) per time interval Always two components: New cases and time

Incidence Incidence is an important rate… It is the proportion of people (at risk) who develop diseased during a specific time period. Three key elements: Only new cases included in numerator Total population at risk in the denominator Time element – period over which new cases developed Two main types of Incidence: Cumulative Incidence Incidence Rate (a.k.a. incidence density)

Cumulative Incidence One of the most widely used measures of disease risk. Estimate of probability (risk) that an individual will develop disease during a specified period of time Cumulative Incidence = No. of new cases in a given period of time No. of people at risk during that time

Incidence rate (incidence density) Cumulative Incidence gives each individual equal weight, but different people stay in the study for different length- having different contribution. Measure of the true rate of disease development Incidence rate = No. of new cases in a given period of time total person-time of observation

Person-time 5-year (1/95-1/00) Incidence rate 1/96 1/97 1/98 1/99 1/00 Total A 3 B C 5 D 1 E 4 Total years at risk • = enter the study, X = having disease,  loss to follow-up 16 x  x x x 5-year (1/95-1/00) Incidence rate = 2/16 = 12.5/100 person-years of observation

Prevalence vs. incidence 1/95 1/96 1/97 1/98 1/99 1/00 A B C D E • = enter the study, X = having disease,  loss to follow-up, disease developing x  x x x 1/97-1/00 cumulative incidence cases: A, E 1/97-1/00 period prevalence cases: A, D, E 1/98 point prevalence: A, D

Relationship between prevalence and incidence Incidence is a proxy for “risk”, whereas prevalence is best for assessing disease burden or case load in a geographic area. There is a well known relationship between them, namely – Prevalence = Incidence x Duration of disease P = I x D

+ Incident cases Whole population at time t Prevalent cases Prevalent cases Prevalent cases Minus cures or deaths due to disease Prevalent cases

Examples of P = I x D If the incidence of diabetes mellitus is 1% per year and its approximate duration is 5 years, then what is its expected point prevalence? Assuming equal incidence of disease, which is more prevalent: pancreatic cancer or brain cancer? Average duration of pancreatic cancer = 3 months Average duration of brain cancer = 1.5 years

Measures of mortality Annual mortality rate from all causes = Total no. of death from all causes in 1 year No.of people in the population at midyear Case-fatality rate = No. of individuals dying during a specified period of time after disease onset No. of individuals with the specified disease

Three common forms of rates Crude rates e.g. crude birth rate, crude death rate Specific rates e.g. sex-specific, age-specific, race-specific Adjusted rates e.g. age-adjusted

Crude rate: example Suppose County B recorded 4000 births and 1500 deaths in 1999. Using U.S. Census data, we find that the population size is 200,000. Crude birth rate = No. of live births in time interval T Total population = 4,000/200,000 = 20 births per 1,000 Crude death rate = No. of deaths in time interval T = 1,500/200,000 = 7.5 deaths per 1,000

Specific Rates for Mortality in Older Adults Rates for selected leading causes of death among older adults, by sex, and race -- United States, 1996* Sex Race Cause of death† Total Male Female White Black Heart disease 1,808 1,983 1,686 1,820 1,937 (612,199) Malignant neoplasms 1,131 1,442 915 1,125 1,338 (382,988) Cerebrovascular diseases 415 374 443 412 479 (140,448) * MMWR Dec 17, 1999 / 48(SS08);7-25

How do we compare rates across populations? Crude rates are not helpful because … Populations differ in their age distributions Populations differ in their racial distributions Populations differ in their SES distributions

How do we compare rates across populations? We compare rates across populations by putting them on an even playing field - that is, we either standardize one population on another or we use an outside standard and adjust our populations to that standard.

For our purposes, the most important is age-adjustment Two types of age-adjustment Direct Method Indirect Method (SMR = standard mortality ratio)

Direct method: example Population A Population B AGE N Risk Cases N Risk Cases <20 100 .1 10 500 .1 50 21-50 200 .2 40 200 .2 40 >50 500 .4 200 100 .4 40 800 250 800 130 CRUDE RISK = 250/800 = 31% 130/800 = 16% Crude risk indicates different risks of disease between populations. But age-specific rates indicate similar risks.

Direct method: example Using the total of the two populations as the standard population Population A Population B AGE Std. Risk Cases Std. Risk Cases pop. pop. <20 600 .1 60 600 .1 60 21-50 400 .2 80 400 .2 80 >50 600 .4 240 600 .4 240 1600 380 1600 380 AGE-ADJUSTED RISK = 24% 250/800 = 24%

Direct method: example Apply risks in population B to population A (using population A as the standard population. Population A Population B AGE Std. Risk Cases Std. Risk Cases pop. pop. <20 100 .1 10 100 .1 10 21-50 200 .2 40 200 .2 40 >50 500 .4 200 500 .4 200 800 250 800 250 AGE-ADJUSTED RISK = 31% 250/800 = 31%

Direct method What information is needed to calculate age-adjusted death rate, using the direct method? Standard population distributed by age Age-specific death rates in study populations The actual value of an age-adjusted rate is meaningless because it depends on the choice of the standard population. It is only meaningful in comparison to other rates which have been adjusted by the same method and the same standard population.

Indirect method Apply rates from a standard population to each age stratum in the study population to obtain expected number. This adjusted rate interpreted as: the rate that would have been experienced by the study population if their rates had been similar to the standard population.

Standardized mortality ratio (SMR) Observed number of cases per time Expected number of cases per time SMR = 0 indicates observed is not unusual SMR > 1.0 indicates morbidity (or mortality) exceeds expected SMR = 2.0 indicates two-fold increase SMR < 1.0 indicates morbidity (or mortality) is less than expected

SMR: example- death in white miners Est. pop. of white miners Death rate in general pop. Expected death Observed death Age (1) (2) (3)=(1)X(2) (4) 20-24 74598 12.26 9.14 10 25-29 85077 16.12 13.71 20 30-34 80845 21.54 17.41 22 35-44 148870 33.96 50.55 98 45-54 102649 56.82 58.32 174 55-59 42494 75.23 31.96 112 Total 534533 181.09 436 SMR = 436/181.09 = 2.41

SMR disadvantage SMR produces a ratio instead of a rate. It gives relative information but does not describe the mortality in the population. SMR depends on the choice of the standard population.