VIEWS: 14 PAGES: 47 POSTED ON: 8/7/2012 Public Domain
Fundamentals of Epidemiology Measures of morbidity and mortality used in Epidemiology Afolabi Olusegun MBChB (Ife), MPH (Hadassah), FMCPH Levels of measurement Nominal Ordinal Interval Ratio Absolute and Relative measures Absolute: age, weight, height Relative: BMI, Ratio of total cholesterol usually relative to size and characteristics of a population Different measures Ratios express a quantity relative to another Proportion : Numerator is part of the denominator Rate : classically is a ratio of a change in a quantity to a change in another quantity with denominator often being time Ratio Mortality: Death to case ratio Morbidity: Risk ratio; Rate ratio; Odd ratio; Period prevalence Proportion Mortality: proportionate mortality Morbidity: Attack rate, Point prevalence, Attributable proportion Rate Mortality : Crude mortality rate, case fatality rate, age specific mortality rate, infant mortality rate Morbidity: person time incidence rate RATE: Measure of the frequency with which an event occurs in a defined population during a given length of time (expressed in terms of some arbitrary constant - %, /1000, 100,000, etc) Rates may also relate special events to total events eg. CFR (not time dimensioned) Comparison of Rates Advantages Disadvantages Actual summary Differences in pop. Crude rates; compositions make Rates Readily calculable interpretation difficult Addresses Cumbersome if there Specific homogeneous are many subgroups; Rates subgroups; No summary figure Useful for PH Summary Fictional rates; Adjusted measure; Magnitude depends on Rates Permits unbiased standard pop; comparisons Hides subgroup diff’s Measures of health status ** Measures of frequency: ** Measures of association ** Measures of potential population impact How do we measure dichotomies?--diseases, disorders, death … Measures of frequency: eg morbidity— prevalence and incidence Prevalence rate is defined as: The proportion of a population-at-risk affected by a “disease” at a specific point in time Prevalence rate (P) is calculated by: # of people with the disease at a specific time x10n # of people in the population at risk at the specified time Types of Prevalence Rates • Point Prevalence rate = # persons with the condition at a point total number of persons in time • Period Prevalence rate = # persons with the condition in a specified total number of persons period of time Point Prevalence • Calendar time (ex. Nov 12, 2001) • Babies born between Jan 1 – Dec 30, 2001 – Of 20,000 born, 60 had symptoms of malformation • Women who enter a study when they show up at the clinic and are examined only once Prevalence question Interview question: Type of measure: Do you currently have • Point prevalence? asthma? • Period prevalence? • Lifetime prevalence? Prevalence question Interview question: Type of measure: Have you had asthma • Point prevalence? in the last 10 years? • Period prevalence? • Lifetime prevalence? Prevalence question Interview question: Type of measure: Have you ever had • Point prevalence? asthma? • Period prevalence? • Lifetime prevalence? Factors influencing observed prevalence rate • Longer duration of the disease • Prolongation of life without cure • Increase in new cases (incidence / risk) • Out-migration of non-diseased people • In-migration of susceptible/diseased people • Improved diagnostic &/or reporting Factors influencing observed prevalence rate • Shorter duration of the disease • Increased case-fatality rate • Decrease in new cases (incidence / risk) • In-migration of “healthy” people • Out-migration of diseased people • Improved cure rate Uses • Pretest probability • Community diagnosis • Basis for decision making and planning • Provide clues for etiology Question • A health centre needs information for use in planning a home care program for people who are too disabled to leave their houses: for example, how many cases can be expected to be under care at a given time, and what is the total number of cases that will be treated during a year? The following information is obtained from the agency that has a program in a similar neighborhood. At the beginning of 1999 the population size was 24,000, and at the end of the year it was 26,000. At the beginning of 1999 there were 96 house bound patients, 20 of these died during 1999, and 4 were moved elsewhere. Another 40 people became house bound during 1999, and 8 of them died during the year. Calculate the point prevalence rates at the beginning and end of 1999 and the period prevalence rate in 1999? Solution • Point prevalence rate was 96/24,000 *1000 At the beginning of 1999 • At the end of the year was [(96+40-20-4- 8)/26000]*1000 • To calculate Period prevalence, we either use the mid year population if given or the average of the population at the beginning and the end. (24,000+26,000)/2 =25,000 Period prevalence = [(96+40)/25,000]*1000 Incidence 1. Counts 2. Risk [Cumulative incidence] 3. Rate [Incidence density, person-time incidence] Cumulative incidence: Is a proportion— N of new cases (numerator) in a fixed group at risk (denominator) over defined time period i.e. risk - attack rate (eg food poisoning) - event rate—new + recurrent cases - lifetime incidence, eg incidence to age 75 Incidence density or person-time incidence: a ratio, a true rate (not a proportion) N of new cases in a changing population observed for varying length of time expressed as N of cases / person-time Eg NSAID-Alzh NEJM 2001;345:1022-9, HRT-CHD Ann Intern Med 2001;135:1-8 Rate analagous to speed of car (rate of change, eg 100 km/hr) --Instantaneous --Potential --Average Risk analagous to distance covered --1/2 of the distance --1/10 of the distance Incidence (cont’) How do we calculate incidence? *Cumulative incidence (CI) CI = I / N0 CI = I / [N0 -(w/2)] ie survival analysis CI ~ 1 - e (-PTI* t ) ie calculated from rate (based on relationship between rate and risk—see ‘sea’ of person-time) Incidence (cont’) Incidence density (ID), person-time incidence (PTI) PTI = I / ti ti = PT PTI = I / N (t) ‘ordinary’ incidence PTI = ID = average incidence density or average rate for a given period Incidence (cont’) ID (PTI) is a ratio not a proportion CI is a proportion Infant mortality example: Is infant mortality a cumulative incidence or person-time incidence measure? Incidence (cont’) Is infant mortality a cumulative incidence or person-time incidence measure? IM as usually calculated is a PTI rate: N of deaths within 12 months of birth divided by the N of live births in a calendar year—say 2003. Why is this a rate and not a proportion? Hint--What have we estimated in the denominator? PT How would you determine infant mortality as a CI measure? Incidence (cont’) How would you determine infant mortality as a CI measure? IM obtained by following a cohort of babies from birth to 1 year is a CI measure --a proportion ie the numerator is part of the denominator Calculations of incidence from example So we can estimate risk in 3 ways: 1. Fixed cohort, no attrition CI=I/N0 2. Actuarial method CI= I/(N0-w) 2 3. Incidence density CI 1 - e -ID * t = PTI·t / [(PTI · t/2)+1] CI ~ ID · t when this product is small ~ < 0.10-- 0.20 CI ~ ID in py when t (duration) is 1 yr Uses of incidence - Prediction individuals - Clinical dx - Etiologic studies (cause) - Community dx population - Prevention/evaluation outcome evaluation of a program ie change in incidence of new or recurrent events Relationship between incidence and prevalence Gordis Relationship between incidence and prevalence Relationship between incidence and prevalence Prevalence - Incidence Screened Prevalence Annual Duration population (per 1000) Incidence (years) (per 1000) Onetown 100 Twotown 60 Prevalence - Incidence Screened Prevalence Annual Duration population (per 1000) Incidence (years) (per 1000) Onetown 100 4 Twotown 60 20 Prevalence - Incidence Screened Prevalence Annual Duration population (per 1000) Incidence (years) (per 1000) Onetown 100 4 25 Twotown 60 20 3 Prevalence - Incidence Screened Prevalence Annual Duration population (per 1000) Incidence (years) (per 1000) Onetown 100 4 25 Twotown 60 20 3 Prevalence Incidence x Duration Mortality studies Basic and commonly available information for characterizing a population and community dx Sources: *death notifications (DC) [numerator] *counts, census in defined areas [denominator] Mortality rates 1. Proportional mortality denominator - all death 2. Case fatality denominator - all cases of disease i. tells you how important - relatively ii. tells you how dangerous 3. Crude mortality—if ‘ordinary incidence’ then the denominator is the average population, typically during a year Mortality rates (cont’) 4. Stratum specific mortality eg age, sex specific - infant mortality 5. Disease specific mortality denominator - same as 3 (above) 6. Prevalence at death denominator - total number of autopsies Proportional mortality--examples Mortality* from cancer in black mine workers (S. Africa) and US blacks US Black Bantu Site 3.0 12.7 liver 61.5 1.3 other 64.5 14.0 all * rates per 100,000 PY, not age adjusted Proportional mortality (cont’) Proportional mortality ratio = 19 (91% of cancer in Bantu due to PHC vs 4.7% in US blacks) Rate ratio = 12.7 / 3 = 4.2 • Case Fatality rate (percent)= No. of individuals dying during a specified period of time after disease onset or diagnosis *100 No. of individuals with a specified disease • Crude Death Rate= Total no. of deaths from all causes in 1 year *1000 No. of persons in the population at midyear Uses of Mortality data • Index of severity of disease • Index of risk of a disease 1. When case fatality rate is high 2. When duration of disease (survival) is short Problems • Underlying cause of death • Quality of data • Cross comparability over time