Quantitative Measures in Epidemiology DEFINITION OF EPIDEMIOLOGY - PDF document

Quantitative Measures in Epidemiology

DEFINITION OF EPIDEMIOLOGY “The study of the distribution and determinants of disease frequency" (MacMahon, 1970) “The study of the occurrence of illness" (Cole, 1979) 2

QUANTITATIVE MEASURES OF DISEASE FREQUENCY Basic elements of epidemiologic inference are defining, counting, and summarizing disease outcomes Outcomes: expressed as either categorical (eg. Disease occurrence or severity) or continuous variables 3

NOMINAL AND ORDINAL VARIABLES • Both nominal and ordinal scale data can be summarized in frequency distributions • Nominal scale data are usually further summarized as ratios, proportions and rates • Ordinal scale data are usually further summarized with measures of central location and measures of dispersion 4

TYPES OF VARIABLES AND STATISTICS Variables Qualitative Quantitative Nominal Ordinal Continuous data Dichotomous Polychotomous (2 groups) (> 2 groups) Mean Median Mode Categorical data Range Ratio Inter-quartile range Proportion Standard deviation Rate 5

QUANTITATIVE MEASURES USED IN EPIDEMIOLOGY • Measures of disease frequency reflect the relative occurrence of the disease in a population. • Measures of association reflect the strength or magnitude of the statistical relationship between exposure status and disease occurrence. • Measures of effect: Certain measures of association involving disease incidence are also measures of the exposure effect. • Measures of impact* are used to predict the impact of an intervention on the disease occurrence in a population (extra number of cases attributable to, or prevented by, the exposure) 6

EPIDEMIOLOGIC APPROACH QUANTITATIVE METHODS • Case Definitions: • Measurement of variables – based on signs, symptoms and results of tests • Numbers and Rates • Estimation of population • Descriptive Epidemiology parameters • Analytic Epidemiology • Testing of statistical hypothesis 7

EPIDEMIOLOGIC DATA • A common form of epidemiologic data is a rectangular database. • Each row contains information about one individual-- i.e., record, observation. • Each column contains information about one characteristic--i.e., variable. • In an outbreak investigation, we usually create a database called a “line listing”. • In a line listing, each row represents a case. Columns contain identifying information, clinical details, descriptive epidemiologic factors, and possible etiologic factors. 8

Neonatal listeriosis, General Hospital A, Costa Rica, 1989 An example of “line listing”. Symptom Delivery Admitting ID Sex Date DOB Type Outcome Symptoms 1 F 6/2 6/2 Vaginal Lived dyspnea 2 M 6/8 6/2 C-section Lived fever 3 F 6/15 6/8 Vaginal Died dyspnea 4 F 6/12 6/8 Vaginal Lived fever 5 F 6/15 6/11 C-section Lived pneumonia 6 F 6/20 6/14 C-section Lived fever 7 M 6/21 6/14 Vaginal Lived fever 8 F 6/18 6/15 C-section Lived fever 9 M 6/20 6/15 C-section Lived pneumonia 10 M 6/19 6/16 Forceps Lived fever 11 M 7/21 7/21 Vaginal Died dyspnea 9 Source: Schuchat 1991

SUMMARIZING DIFFERENT TYPES OF VARIABLES When categories are used, the measurement scale is called a nominal scale. Vaccination Number Yes 76 No 125 Total 201 When points on a numerical scale are used, the 10 scale is called an ordinal scale.

FREQUENCY DISTRIBUTION • With larger databases, we usually summarize variables into tables called “frequency distribution”. • A frequency distribution shows the values a variable can take, and the number of people with each value. Example 11

Example of “frequency distribution”. Distribution of Students by Levels of Blood Sugar, n= 100 Bl. sugar (mg% ) Number Relative freq Cumulative relative freq 52-55 4 4 4 56-59 12 12 16 60-63 16 16 32 64-67 27 27 59 68-71 13 13 72 72-75 19 19 91 76-79 4 4 95 80-83 5 5 100 Total 100 100 100 12

13 In what circumstance that we should measure by counting absolute number?

(N= 38) of mumps cases, kindergarten “A” ,May–September 1999 Epidemic curve and spot map 10 0 2 4 6 8 18 - 24 M ay 25 - 31 M ay 1 child case 1 - 7 Jun 8 - 14 Jun 15 - 21 Jun 22 - 28 Jun NS 1 29 Jun - 5 Jul Distribution 6 - 12 Jul 13 - 19 Jul NS 2 20 - 26 Jul 27 Jul - 2 A ug 3 - 9 A ug 10 - 16 A ug 17 - 23 A ug 1 officer case 24 - 30 A ug 1 / 1 31 A ug - 6 Sep 7 - 13 Sep 14 - 20 Sep 21 - 27 Sep 1 / 2 28 Sep - 4 Oct 5 - 11 Oct interval Weekly 3 / 2 Laosirithaworn, 1999 2 / 2 3 / 1 2 / 1 Kit. 14

Example: Investigation of increasing death from unintentional fall, Thailand • The injury surveillance (IS) data from Lampang regional hospital showed increasing number of death from unintentional fall after 1998 • FETP was notified and went to investigate • IS report and medical records were reviewed and relatives of the deaths were interviewed 15 Source: Jiraporn Plaitho, 2002

Number of deaths from unintentional fall by year and age-group, Lampang hospital 1997-2001 Age 1997 1998 1999 2000 2001 0-14 yr 1 0 1 1 1 15-59 12 7 15 12 20 yr > = 60 15 10 16 17 22 yr 16 Source: Jiraporn Plaitho, 2002

Population of Lumpang province 1997-2001 Age 1997 1998 1999 2000 2001 0-14 161,221 162,430 159,550 158,160 153,343 yr 15-59 479,933 493,553 500,146 509,278 513,012 yr > = 60 90,002 93,184 94,983 82,914 99,441 yr 17 Source: Jiraporn Plaitho, 2002

Number of deaths from unintentional fall by age- group, Lampang hospital 1997-2001 Number 25 0-14 yrs 15-59 yrs >=60 yrs 0-14yr 20 3% 15 14-59yr 44% >=60yr 53% 10 5 0 N=150 1997 1998 1999 2000 2001 18 Source: Jiraporn Plaitho, 2002

Death rate of unintentional fall by age-group, Lumpang hospital 1997-2001 r ate ? Rate per 100,000 pop 25 0-14 yrs 15-59 yrs >=60 yrs 20 0-14yr 3% 15 14-59yr 44% >=60yr 10 53% 5 0 N=150 1997 1998 1999 2000 2001 19 Source: Jiraporn Plaitho, 2002

DISEASE FREQUENCY Disease frequency is usually measured as a proportion or rate in which: • Numerator reflects the number of cases or events of interest • Denominator reflects the size of a population from which those cases or events are identified 20

TYPES OF FREQUENCY MEASURES Distinguished by type of numerator • Incidence: the numerator reflects the number of new cases identified during a given period. • Prevalence: the numerator reflects the number of existing cases identified at a point in time. 21

22 and Prevalence Incidence

23 Prevalence is the frequency of existing cases PREVALENCE MEASURES

Prevalence is calculated by: Number of people with the disease or condition at a specific time P = Total population at a specific time 24

1 existing case at a point in time in a population of 5 babies 1 Prevalence = = 0.2 = 20% 5 • Point prevalence is: The proportion of the population affected by a disease at a specific point in time 25

PERIOD PREVALENCE • Period prevalence is calculated by: Number of incident and prevalent cases identified during a given period • P = Size of the total population during the period 26

3 existing case during a period of time in a population of 5 babies 3 Period revalence = = 0.6 = 60% 5 • Period prevalence is: The proportion of the population affected by a disease anytime during a given period 27

INTERPRETATION OF PREVALENCE • Because prevalence reflects both incidence rate and disease duration, it is not as useful as incidence for studying causes of disease. • It is useful for measuring disease burden on a population, especially if those who have the disease require specific medical attention. 28

RELATIONSHIP BETWEEN PREVALENCE AND INCIDENCE Prevalence is less useful than incidence in etiologic studies, because it is a function of incidence rate ( ) and duration of disease ( ) I T ( ) P = I T − 1 P Assumption: prevalence, incidence rate and mortality rate remain constant over time, no in- and out-migration 29

RELATIONSHIP BETWEEN PREVALENCE AND INCIDENCE If the disease is rare, ( ) P ≈ I T = mean duration of disease T Assumption: prevalence, incidence rate and mortality rate remain constant over time, no in- and out-migration 30

FACTORS INFLUENCING OBSERVED PREVALENCE Increase/Decrease Longer duration of the disease High case fatality rate Decrease in incidence Improved diagnostic facilities Better reporting Improved cure rate In-migration of healthy people Out-migration of cases 31 Source: WHO, 1994

The proportion of infants who are born alive with a defect of the ventricular septum of the heart is a prevalence or incidence? 32

33 Risk and Rate

INCIDENCE MEASURES: RISK AND RATE Distinguished by type of denominator • Risk (cumulative incidence, incidence proportion): probability of the event • Incidence rate (incidence density): rate estimate expresses the “rate” at which the events occur in the population at risk at any given point in time 34