Home 
Contact 
Profile 
Videos 
Paper written by Professor Omar Hasan Kasule Sr.
MODULE OUTLINE
9.1 DISEASE: CLASSIFICATION, DESCRIPTION, MEASUREMENT, DIAGNOSIS, and PROGNOSIS
9.1.1 Classification of Disease
9.1.2 Disease Description
9.1.3 Disease Measurement
9.1.4 Disease Diagnosis
9.1.5 Disease Prognosis
9.2 DISEASE DETERMINANTS
9.2.1 Concepts of Disease Causation
9.3.2 Concept of Exposure
9.2.3. Disease Determinants
9.2.4 Variation of Disease Risk with Age
9.3 DISEASE CONTROL, ERADICATION, AND PREVENTION
9.3.1 Control
9.3.2 Eradication
9.3.3 Prevention
9.3.4 Preventive Medicine: Legal Basis
9.3.5 Preventive Medicine: Moderation, Balance and Equilibrium
9.4 DISEASE SURVEILLANCE
9.4.1 Definition
9.4.2 History of Surveillance
9.4.3 Objectives, Methods, and Scope
9.4.4 Surveillance System
9.4.5 Data Collection, Analysis, and Interpretation
9.5 DISEASE SURVEILLANCE
9.5.1 Definition, Objectives, Organization, And Benefits
9.5.2 Characteristics of Disease and Screening Tests
9.5.3 Epidemiologic Evaluation of Screening Programs
9.5.4 Cost Benefit Analysis of Screening Programs
9.5.5 Ethical Issues
UNIT 9.1
DISEASE CLASSIFICATION, DESCRIPTION, MEASUREMENT, DIAGNOSIS, and PROGNOSIS
Learning Objectives:
· Disease description: symptoms, signs
· Disease classification criteria
· Time characteristics of disease: trends, duration, natural history
· Place characteristics of disease: urban/rural, boundaries (political & natural), institutions
· Person characteristics of disease: age, gender, race/ethnicity, SES, marital status
· Epidemicity: epidemic, endemic, pandemic, epidemic curve, slow and acute epidemics, visibility of the epidemic
· Measures of disease occurrence and excess disease risk
Key Words and Terms:
· Disease classification
· Disorder classification
· Epidemic disease
· Ethnicity
· Etiology of disease
· Heredity
· Infection
· Morbidity
· Mortality
· Nomeclature of disease
· Nosology
· Risk ratio
· Risk difference
· Incidence rate ratio
· Incidence rate difference
· Odds ratio
· Cumulative incidence
· Morbidity
· Mortality
UNIT OUTLINE
9.1.1 CLASSIFICATION OF DISEASE
A. History of disease classification
B. Purpose of disease classification
C. Classification criteria
D. Commonly used classifications
International classification of disease
9.1.2 DISEASE DESCRIPTION
A. Purpose of disease description
B. Time characterization of disease:
C. Place characterization of disease:
D. Person characterization of disease:
E. Clustering and epidemics
9.1.3 DISEASE MEASUREMENT
A. States and Events
B. Incidence
C. Prevalence
D. Measures of Excess Disease Occurrence
E. Measures of Disease Impact & Measures of Survival
9.1.4 DISEASE DIAGNOSIS
A. Disease Identification
B. Symptoms and Signs
C. Diagnostic Tests
9.1.5 DISEASE PROGNOSIS
A. Objective Assessment of Prognosis
B. Study Design for Prognostic Factors
C. Parameters Used in Study of Prognosis
9.1.1 CLASSIFICATION OF DISEASE
A. HISTORY OF DISEASE CLASSIFICATION
In 1853 William Farr (1807-1883) introduced standard nomenclature for causes of death. In 1893 J Bertillon (1851-1922) introduced a new classification of causes of disease. In 1946 WHO introduced an International Classification of Diseases, Trauma, and Cause of Death. In 1975 the 9th revision of ICD was published. The 10th version was published in 1992.
B. PURPOSE OF DISEASE CLASSIFICATION
Disease classification serves the following five functions: explanation & description, prediction of disease course, prognosis, planning treatment, and disease prevention.
C. CLASSIFICATION CRITERIA
4 types of classification criteria can be identified: manifestational, causal, abstract, operational or pragmatic. Manifestational criteria, symptoms and signs, are used in the absence of knowledge of cause. Causal criteria are based on certain knowledge of cause examples: birth trauma, silicosis, syphilis, lead poisoning, AIDS. Abstract criteria are not natural and are a product of human intellectual effort often with little practical use. Operational/pragmatic criteria, like the International Classification of Diseases, are practically-oriented but may not have good theoretical foundation.
D. COMMONLY USED CLASSIFICATIONS
In practice the following 5 methods are used in classification or cross-classification of diseases: etiologic agent or force, disease process, organ system affected, method of transmission, and portal of entry. Etiologic agent/force of morbidity can be microbiological agents, genetic anomalies, metabolic disorders, environmental factors (water and air pollution), human behavior, stress (anxiety and depression), or traumatic injury. In practice most diseases are multi-causal. Disease process may be neoplastic, metabolic, infectious, or inflammatory. The organ system involved may be cardio-vascular, respiratory, nervous, genital, and urinary. The method of transmission may be water-borne, food-borne, air-borne, vector-borne, and STD. Portals of entry are the respiratory, GIT, and parenteral routes.
Medical test books use their own classification of diseases designed to suit their pedagogical approach and the sequence of presenting information.
E. INTERNATIONAL CLASSIFICATION OF DISEASE
The International Classification of Diseases (ICD) was evolved by the World Health Organization to be used uniformly by all countries in reporting disease statistics. It makes possible an international comparison of disease occurrence. The 10th version, called the International Statistical Classification of Diseases and Related Health Conditions (ICD-10), has 31 chapters listed as follows. Chapter I (A0-B99): Certain Infections and Parasitic Diseases. Chapter II (C00-C97): Neoplasms. Chapter III (D50-D89): Diseases of the Blood Forming organs and Certain Disorders involving the Immune Mechanism. Chapter IV (E00-E90): Endocrine, nutritional, and metabolic diseases. Chapter V (F00-F99): Mental and Behavioral Disorders. Chapter VI (G00-G99): Diseases of the Nervous System. Chapter VII (.H00-H69): Diseases of the Eye and Adnexa. Chapter VIII (H60-H95): Diseases of the Ear and Mastoid process. Chapter IX (I00-I99): Diseases of the Circulatory System. Chapter X (J00-J99): Diseases of the Respiratory System. Chapter XI (K00-K98): Diseases of the Digestive System. Chapter XII (L00-L99): Diseases of the Skin and sub-cutaneous tissue. Chapter XIII (M00-M99): Diseases of the Musculoskeletal System and Connective Tissue. Chapter XIV (N00-N99): Diseases of the Genitourinary System. Chapter XV (O00-O99): Pregnancy, Childbirth, and the Puerperium. Chapter XVI (P00-P96): Certain Conditions Originating in the perinatal period. Chapter XVII (Q00-Q99): Congenital Malformations, Deformations, and Chromosomal Abnormalities. Chapter XVIII (R00-R99): Symptoms, Signs, and Abnormal Clinical and Laboratory Findings Not Elsewhere Classified. XIX (S00-T98): Injury, Poisoning, and Certain Other Consequences of External Causes. XX (V01-Y99): External Causes of Morbidity and Mortality. Chapter XXI (Z00-Z99): Factors Influencing Health Status and Contact with Health Services. (page 46 John M Last: Public Health and Human Ecology. 2nd edition. Prentice Hall International, Inc ? year).
9.1.2 DISEASE DESCRIPTION
A. PURPOSE OF DISEASE DESCRIPTION
Disease description serves to answer the following questions: what, why, when, how, where, and who of a disease.
B. TIME CHARACTERIZATION OF DISEASE:
TWO TYPES OF TIME DESCRIPTION
Time can be described as calendar time or cohort time. Calendar time is time indicated or measured by days, weeks, months, seasons, and years. Calendar time is measured on the interval scale. The interval scale has an arbitrary zero and admits both positive and negative values. Cohort time is measured as time elapsed since a significant event. Examples of such events are: time since birth, age; time on the study; time since admission to hospital; and time since start of treatment. Cohort time is measured on the ratio scale. Zero on this scale has a biological significance; negative values have no meaning.
TIME TRENDS
Five different types of time trends can be described: biorhythm, periodicity, steps, linear, and curvilinear trends. Biorhythm is the daily alternation of biological phenomena referred to also as diurnal or nocturnal variation. Examples of the diurnal and nocturnal variations are the cycles of sleep and wakefulness, secretion of ACTH, and secretion of melatonin. Periodicity can be cyclic (monthly, annual, seasonal) or accidental. A distinction is made between secular trends that occur over several years in a predictable fashion and recent trends that may not be repeated in a predicable manner. The Fourier series of polynomial analysis can be used for analysis of cyclic events. Steps are short-term trends. Linear trends show a clear and definable relation between time and other variables. The linear trend may increase or decrease. Regression equations describe linear trends well. Curvilinear trends require more sophisticated descriptions using polynomial functions.
DURATION
Based on their duration, diseases can be described as acute, sub-acute, chronic, and acute exacerbations of chronic conditions. The decision of what time period is acute or chronic is arbitrary and varies from disease to disease. Acute and chronic disease: There is inaccurate perception that acute diseases are infectious and chronic diseases are non-infectious. In acute diseases, symptoms and signs are disposed of within 3 months and those who survive have complete recovery. In chronic diseases symptoms and signs continue for longer than 3 months and sometimes for life. Recovery is slow and is not complete. Acute disease can be communicable such as the common cold, pneumonia, mumps, measles, pertussis, typhoid, and cholera. They may be non communicable such as appendicitis, poisoning and traumatic injury. Chronic diseases may be communicable such as AIDS, Lyme disease, tuberculosis, syphilis, and rheumatic fever. Chronic diseases may be non communicable such as diabetes mellitus, coronary artery disease, ostoarthritis, and alcoholic liver cirrhosis.
NATURAL HISTORY
The following table summarizes the natural history of disease.
Stage | Preceding event | Contemporary events | Prevention |
Susceptibility | - | - | Primary |
Sub-clinical Disease | Exposure | Pathological changes in incubation | Secondary |
Clinical Disease | Symptom Onset | Diagnosis | Secondary |
Recovery, disability, or death | - | Treatment | Tertiary |
Source: Epidemiology Made Simple Table 2.1 page 21.
Disease can be described on a time line that traces its course of development. The time course of disease evolution is referred to as natural history. Natural history describes disease progression from the operation of the causative agent through clinical manifestations to termination. Termination may be by death, cure, or chronic complications. Disease progression is through the following stages. The pre-pathogenesis stage is the stage of operation of the risk factors. The pre-clinical stage, disease is already initiated but here are no symptoms or signs. In the clinical stage symptoms and signs appear. In the chronic stage there are complications and permanent deformities. Three time periods or intervals can be described in the natural history of disease: induction period, incubation period, latent period. The induction period is the time from causal action of the component cause to disease initiation. A co-factor may trigger the eventual disease onset. There is a different induction period with respect to each component cause. The latent period is the time between disease initiation and disease detection. The concept of latent period complicates the analysis of time-related data because it is not possible to pin -point the start of the disease process. The induction period can not be reduced. The latent period can be reduced by methods of early detection of disease.
EVENTS
Diseases can be described in relation to significant events. Significant life events include birth, death, marriage, divorce, school entry, menarche, and menopause. Point events like an earthquake or a social revolution occur for a very brief time and stop. A point event may, however, have an extended after-math like the atomic bombs whose effects were felt long after the point event itself. Some events have prolonged exposure with a start and an end. The start or end may be well defined or may be poorly defined. A good example of this type of event is air pollution.
C. PLACE CHARACTERIZATION OF DISEASE:
DISEASE AND PLACE OF RESIDENCE
Since the beginning of history, humans have known that some diseases are associated with certain places of residence. Disease can therefore be described by place of residence. The commonly used classifications are described below.
URBANIZATION
The following classifications are usually used: rural, urban, sub-urban, and slums/shacks (septic fringe). There are differences in socio-economic status, air pollution levels, nature of soil pollution, and availability of health facilities. These differences determine patterns of disease transmission.
BOUNDARIES
There are two types of boundaries: political and natural. Political boundaries separate administrative areas like the country, the district, and the city. Natural boundaries are valleys, mountains, rivers, lakes, oceans and other physical barriers to human movement. The significance of boundaries is that they restrict movement of people and therefore affect the patterns of disease transmission. Boundaries may limit the movement of disease causative agents when policies or the eco-system on one side of the boundary enhance or hinder a certain mode of disease transmission.
INSTITUTIONS
The following institutions are often used in disease description: hospital, home, school, factory, farm, and outer space. The institutional environment or the institutional activities define the type of disease found among the inmates of the institution.
MAPPING DISEASE
Geographical display and analysis can be used to relate disease phenomena to place. The following methods are used: (a) chloropeth mapping: incidence and prevalence are computed for various places and are shown on the map using different coloration (b) isopleth mapping uses contours of equal disease parameters and need not follow geo-political boundaries and can be based on sample data (c) map-on-map technique is when the map of disease distribution is superimposed on a map of risk factor distribution (d) computer displays of increasing sophistication are being developed and used.
PROXIMITY ANALYSIS
This is the of distance relations between cases of disease and the hazard. Two methods are used: the maximum ratio between the observed and the expected and the Kolmogrorov-Smirnoff statistic that tests the maximum distance between two curves.
INTERNATIONAL COMPARISONS
Different countries publish health statistics that show a wide variation in disease patterns. However gross differences especially at the country level can be misleading.
D. PERSON CHARACTERIZATION OF DISEASE:
INDIVIDUAL VARIATION
Individual variation in exposure and susceptibility varies by heredity, age, sex, SES, marital status, and ethnicity/race. Heredity operates through genetic endowment from parents ort the interaction of genetic and environmental factors in causing disease. Susceptibility to disease is high at both ends of the age spectrum due to immune incompetence. The newborn and infants have an undeveloped immune system. The immune system of the elderly in the geriatric age has undergone degenerative changes. Adolescents experience a high rate of some diseases because of their high-risk life-style and the rapid growth of the pubertal spurt. Females have disease experience different from males because of a less hazardous life style, different hormonal and reproductive functions. Females transmit X-linked hereditary diseases but do not suffer from them. Sometimes the male-female difference in disease susceptibility may be due to having different organs. Disease of the ovary and the uterus are found only in females. Disease of the prostate and testis are found only in males. SES affects disease susceptibility by determining the place of residence and hence the exposure to environmental causes of disease. Life-style, under-nutrition, and over-nutrition, type of work and hence the type of occupational exposure, and place of residence are determined by SES. Marital status is very important in that it determines psychological stability, an important ingredient in disease-related life-style. Marriage may also encourage a monogamous sexual life that prevents sexually-transmitted disease.
AGE
There are distinct disease patterns for different age groups. Diseases of childhood such as causes of IMR are different from diseases of old-age such as Alzheimer's disease. Some diseases have bimodal peaks at young and later age eg Hodgkin's Disease and testicular tumors. There is a cohort effect in disease occurrence that may translate into secular trends.
SEX
The death rate is higher for males at all ages: in utero, neonatal, and later life. The reasons for higher male mortality in later life are: smoking, accidents, firearms, and AIDS. If measured by utilization of health services, morbidity is higher for females for all diseases. The gender ratio at birth must be taken into consideration when interpreting gender-specific mortality or morbidity rates. The gender ratio at birth varies by race, birth order, and over time.
SES
Socio-economic status may be associated with high disease risk or low disease risk depending on the type of disease.
ETHNIC GROUP/RACE
The differences in disease experience among the different ethnic groups reflect socio-economic, life-style, and cultural variations. The variation of disease risk by ethnicity has no genetic basis in most cases.
MARITAL STATUS
The happily-married and their children are generally healthier than the unmarried. The married have more psychological stability. They also have more economic security. Interpretation of the effect of marital status is rather complicated because it is possible that those people who are psychologically stable and have economic security get married. Marriage therefore is a consequence and not a cause.
E. CLUSTERING and EPIDEMICS
INTERACTIONS AMONG PERSON, PLACE AND TIME CHARACTERISTICS
Sophisticated descriptions are called for when describing phenomena involving 2 or 3 dimensions at the same time. There are three usual interactions: (a) place-time (b) time-person (c) person-place. These interactions can be modeled mathematically for easier description and understanding. Sophisticated statistical methods are also available for their analysis. We will illustrate the complex interactions by study of disease clusters and epidemics.
CLUSTERING
Definition: Clustering is excessive concentration of events at a point in time or a place. Disease clustering can be described in relation to time, place, both time and place, family, and household. Clustering in place and time usually indicates an infective cause. Clustering in a family may be due to contagious infection or genetic predisposition.
Testing for clustering in time: Clustering in time is a non-linear, non-cyclical, or non-random phenomenon. The Poisson distribution is used to analyze time clustering. Clustering in time can be tested for by dividing the time of observation into short intervals and counting the number of cases of disease per time interval. The distribution should follow a Poisson distribution under the null hypothesis of no clustering. Significance deviation from the null can be tested using a chi-square test.
Testing for clustering in place: Clustering in place can be tested for by counting numbers of cases occurring per administrative unit. The number of cases per unit should follow a Poisson distribution under the null hypothesis of no clustering. Significance deviation from the null can be tested using a chi-square test.
Testing for clustering in both time and place: Description of the phenomenon of a moving cluster is quite complex because the clustering occurs in both the time and place dimensions. Testing for clustering in both time and place is more complicated. Place-time clusters can be analyzed using 2 methods: (a) the Knox method based on contingency tables (b) the Mantel-Haenszel method relating time and space intervals. Criteria of closeness are determined based on both time and place. A 4-fold table is constructed showing time (adjacent/not adjacent) and place (adjacent/not adjacent). The Poisson distribution should follow the null hypothesis of no clustering. Significant deviation from the null can be tested for using a chisquare test.
Cluster investigations are made difficult by several factors: rarity of events, vague case definitions, lack of a population base for rate computations, weak associations, multiple risk factors, long induction periods, and multiple comparisons. Cluster investigation starts by careful data collection about disease occurrence. This is followed by careful epidemiological investigation which includes defining the geographical area to be covered, the case definition, confirmation of case diagnosis by using pathological and clinical data, computing disease rates and ratios, and a study design to investigate etiology.
Comparison of outbreak and cluster investigation: Outbreaks involve infectious diseases with definable transmissible agents. Clusters are non-communicable conditions with several risk factors some of which are not well established. Outbreak investigations are short measured in days and weeks whereas cluster investigations are longer measured in weeks or months. Exposure levels and effect estimates are high in outbreaks and low in disease clusters. Exposure period in outbreaks is acute being measured in hours of days whereas disease clusters are associated with chronic exposures measured in years or decades. Laboratory confirmation of disease and exposure is easy in outbreaks but difficult in disease clusters. The cause-effect relation is easy to establish in disease outbreaks and difficult in disease clusters. Disease outbreaks are usually investigated using retrospective cohort studies whereas disease clusters are investigated using case control studies.
DISEASE OUTBREAK
A disease outbreak is usually excessive diseased occurrence of a lesser degree than an epidemic. Investigation of an outbreak starts with determining that there is an excessive occurrence based on comparison of current disease rates with surveillance data. Disease diagnosis is then confirmed using laboratory techniques. A case definition is then developed so that a definitive count of cases can be made based on the definition. Disease occurrence is then described by its place, time, and person characteristics. The susceptible population at risk of the disease is then determined. Explanatory hypotheses on causality are developed and are tested in a systematic way. A comprehensive report is prepared. Control and preventive measures are instituted. Control may involve the pathogen, the chain of transmission or the host response. Control aimed at the pathogen involves removing the source of contamination, removing persons from exposure, inactivating the pathogen, treating isolating the infected persons. Transmission is interrupted by sanitation, sterilization, disinfection, vector control, and hygiene. Host response is modified by immunization or using prophylactic therapy.
EPIDEMICS
Definition: En epidemic represents a time-person-place interaction. An epidemic is defined as excessive frequency. Most epidemics in history occurred as acute dramatic events which made many forget the existence of slow but serious epidemics. Three terms should not be confused: endemic, epidemic, and pandemic. An endemic disease is one with high prevalence in an area. An epidemic is excessive incidence over a given usually brief period of time. The following epidemics were recorded in the US: polio epidemic in in 1916, St Louis encephalitis in 1975 that affected 1815 persons, Legionnaire’s disease in 1976 that affected 235 persons, the Guillain Barre syndrome epidemic following swine virus immunization in 1976, Toxic exposure from the Love Canal in 1979, the toxic shock syndrome in 1980 that affected 877 persons, plague in 1983 that affected 40 persons, Lyme disease in 1982-1983 that affected 872 persons, the encephalomyalgia syndrome in 1990, and the ongoing AIDS epidemic starting in 1981. A pandemic is an epidemic occurring simultaneously in widely separated geographical regions. An example of a pandemic was the influenza of 1918-1919 which started in France and spread to Spain, England and the rest the rest of Europe, China, and West Africa. The current AIDS epidemic is becoming an pandemic.
Types of epidemics: An epidemic may be point source if the origin is one person or one place. It is common source if more than one origin is involved. Transmission may be person to person or the outbreak may be propagated. Epidemic progression can be described as propagated when new infections are occurring or limited when infection is limited to the initial cases.
Visibility of an epidemic: Visibility of the epidemic in increased under the following five scenarios: the cases are few but the disease is rare, a large number of cases, a distinctive illness, a swift shift from the non-epidemic to the epidemic status, and disease of very short duration. Cholera, smallpox, and plague are examples of visible epidemics. An invisible epidemic may be missed in the following five situations: a slow epidemic, disease manifesting in an unfamiliar form, lack of systematic search for the disease (eg the thalidomide disaster), and inattentiveness (e.g. the London fog of 1952 that killed 4000 people in 10 days), and not knowing the background pre-epidemic incidence. Coronary heart disease and lung cancer are examples of slow epidemics.
Identification of an epidemic: The occurrence of an epidemic can be ascertained in 3 ways: studying changes in trend over time, comparing incidence in epidemic and non-epidemic places, and comparing disease incidence among population sub-groups of the same country. Investigating an epidemic must follow certain procedures. The diagnosis must be established. A case definition must be agreed on. Based on the diagnosis and case definition a determination is made whether an epidemic exists. The epidemic is then characterized by place, time, and person. A spot map showing cases can be drawn. Incidence rates can be computed by location. Clusters can be identified but care must be taken to distinguish real from chance clusters. Hypotheses are developed about the source and route of infection. The hypotheses are tested by laboratory studies and case control studies. The pattern of spread may be common source, propagated or a mixed pattern. A point source exposure may be one-time or sporadic. A common source exposure may be continuous. Control measures are then instituted and may include sanitation, prophylaxis, diagnosis and treatment, and vector control. Surveillance is continued in the post epidemic period.
EPIZOOTIC
An epizootic is an epidemic disease in animal populations. Epizootics can become epidemics in human populations for example the St Louis encephalitis started as an epizootic condition that became an epidemic. An enzootic is an endemic disease among animals.
EPIZOODEMIC
This is an epidemic involving both human and animal populations.
9.1.3 DISEASE MEASUREMENT
A. STATES and EVENTS
State is static measured as prevalence (point or period prevalence). Event involves a time dimension. An event is a change of state. Events are measured as incidence. The term incidence is used to refer to incidence rate and cumulative incidence.
B. INCIDENCE
INCIDENCE TIME, INCIDENCE NUMBER, and ABSOLUTE RATE
Incidence is a dynamic concept that measures risk, the force of morbidity or hazard. It is a measure that deals with random events and change in status. Incidence time is the time at which new disease occurs in a population for example age at death, time on treatment. Incidence time is always measured from a zero time. Incidence number is the number of new cases detected in the time interval. The term absolute rate is also used to refer to the number of new cases in a given time interval.
INCIDENCE RATE
The incidence rate (IR) is a basic measure of disease occurrence. Incidence rate (IR) = incident number/ total person-time. Person time is defined as the sum of the product of number of persons observed by the length of time that they are observed. IR can be computed based on the first ever episode. It can alternatively be computed as episode incidence which involves the first and subsequent events. The incident number considers only the first occurrence of disease. Person-time measures the time at risk. The time dimension can be chronological or can be the duration of time on study. Use of the concept of the steady state enables us to define IR using the mid-year population as follows: IR= # newly reported cases of a disease in a year / mid-year population. The incidence rate has a lower bound of zero but has no interpretable upper bound. The upper bound depends on the unit of measurement used for example an incidence rate of 100 per person-year is the same as 10,000 per person-century. The plot of the logarithm of the incidence rate against time indicates disease trend well. The definition of the incidence rate for an open population is different from that of a closed population. The incidence rate for an open population is referred to as incidence density, person-time rate, person-time incidence rate, force of morbidity, hazard rate, disease intensity, instantaneous risk, instantaneous probability, force of mortality, and force of morbidity. The situation of an open population is depicted in the figure below:
The computation of the 95% confidence intervals for the incidence rate varies according to the number of cases, n. If n<75, the Poisson distribution is used. If n < 75 <100, either the poisson of the normal distribution is used. For n>100 the normal distribution is used. There are alternatives to the incidence rate as defined above. For example we can describe the accident rate without reference to time as the number of accidents per person-mile. The reciprocal of an incidence rate is the waiting time until occurrence of the event of interest. The average waiting time until death, for example, is the waiting time until death.
CUMULATIVE INCIDENCE
Cumulative incidence, also called incidence proportion or attach rate, is defined over a given interval of time as the number of incident cases divided by the total number of the cohort observed at the start of the observation interval. Stated otherwise it is the proportion of those who become cases among those who entered the given time interval. It can also be defined as the number of new cases as a proportion of the susceptible population at the start of the observation. Cumulative incidence = summation over time of IR. CI is a probability and a measure of risk and can be considered as the average risk over the time interval.
COMPARISON OF INCIDENCE RATE AND CUMULATIVE INCIDENCE
Both the cumulative incidence (CI) and prevalence rates (P) are based on the incidence rate (IR) but are not reliable unless follow-up is for a very short period of time. When CI is low <0.1, CI is approximately equal to IR. IR is superior to CI in that competing causes of death operate on both the numerator and denominator of IR but only on the numerator of CI. IR is suitable for study of dynamic populations. IR has the advantage that person-time can be adjusted for the effects of censoring. CI can also be used for censored data if lifetable and actuarial methods are resorted to. CI is suited to study of acute diseases with restricted risk periods. It is also suitable for study of fixed cohorts.
SURROGATE MEASURES OF INCIDENCE
Mortality rates are readily available and are used as surrogates for incidence rates. Mortality is related to the important demographic parameter of life expectancy
C. PREVALENCE
Prevalence is a static concept that is a measure of state. It is a still-picture of the disease situation at a given point in time. Whereas incidence relates to events, prevalence relates to disease states at a point in time. The prevalence number is the number of cases of disease existing at the particular point in time. The prevalence proportion = # cases of illness at a particular time / # of individuals in the population at the same time. Prevalence proportion is also called prevalence rate or point prevalence. Prevalence is not a rate but a proportion; however the term prevalence rate has become so popular in medical literature that it will take long for this error to be corrected. Prevalence is measured in cross-sectional studies. Only one observation at one point in time is needed in the determination of prevalence.
Three types of prevalence are described in epidemiological literature: point, period, and lifetime prevalence. Point prevalence is a theoretical concept that assumes ability to count cases of illness at an infinitesimal short period of time. Period prevalence refers to counting the number of illnesses over a practically reasonable length of time. This must not be so long that there is a change in the status quo by death of cases or incidence of new ones. Period prevalence is more stable and therefore more useful than point prevalence. Life-time prevalence is a special type of period prevalence referring to the whole of a person's prior life. Cumulative prevalence includes all disease conditions (cured, or resolved, continuing, and dead) in a given period of time.
There is a relation among incidence, prevalence and duration. Prevalence proportion = incidence rate x average duration of disease. A different reformulation of this formula by Richard Monson is PR = (IR - CR - MR)D. where PR=prevalence rate, CR= cure rate ie cases cured per unit time, MR= mortality rate i.e. cases dying per unit time, and D=duration in time.
Prevalence is useful for administrative purposes. It is rarely used for etiological studies except for conditions in which incidence is difficult to measure such as congenital malformations, non lethal degenerative diseases, and sero-conversion. Prevalence is not good for etiological studies for the following reasons: (a) It can not distinguish the contribution of incidence from that of disease duration (b) The time sequence is not obvious; disease and exposure are studied at the same time. PR may only give a clue about IR if IR is not available. In the extreme cases of rapidly fatal diseases, Prevalence may indicate true incidence. Average disease duration = 1R/termination rate per P-Y. Prevalence ratio (p/1-p) = IR x average duration. If p is small, p = IR x Average duration
Change in prevalence is due to: (a) Change in incidence (b) Change in duration: due to dearth or recovery (c) Both change in IR and duration.
It is sometimes useful to compute 95% confidence intervals for a proportion. The exact binomial is used of the number of cases is less than 75. The normal approximation is used for numbers above 75.
D. MEASURES OF EXCESS DISEASE OCURRENCE
Measures of excess disease occurrence, also called measures of effect, are based on measures of association. Excess disease risk is measured as an absolute effect (Rate Difference or Risk Difference) or a relative effect (Relative Risk, Rate Ratio, Risk Ratio, Prevalence Ratio, Cumulative Incidence Ratio, Incidence density Ratio, Odds Ratio, and Standard Mortality Ratio). There is no consistent relation between RD and RR. RD may be large for a small RR and vice versa. In the same way one measure may show heterogeneity of stratum-specific measures whereas the other does not. The rate ratio, the risk ratio and the odds ratio may be approximately equal under certain conditions. The rate ratio and odds ratio are commonly used in epidemiology. The odds ratio is a good estimate of the rate ratio under three conditions: the cases of the disease are representative of all diseased people in the population, the controls or comparison group are representative of all healthy people in the population, and the disease is rare. In general risk ratio > rate ratio > odds ratio. Various parameters discussed below are defined from the 2 x 2 contingency table
Disease + | Disease - | Person-time | ||
Exposure + | a | b | a + b | PT+ |
Exposure - | c | d | c + d | PT- |
a + c | b + d | N | PT |
Two odds ratios can be defined, the disease odds ratio and the exposure odds ratio. The disease odds ratio is defined as a/b ¸ c/d. The exposure odds ratio is defined as b/d ¸ a/c. It can be shown mathematically that the disease odds ratio = exposure odds ratio = ad/bc. The standard deviation used for computation of 95% confidence intervals for the odds ratio is given by OR exp [+/1 1.96 (1/a + 1/b + 1/c + 1/d)]1/2.
Relative Risk can be defined as a rate ratio based on incidence rates or as a risk ratio based on cumulative incidence or prevalence. The rate ratio is defined as {a / PT+} / {c/PT-}. The risk ratio is defined as {a/N} / {b/N}. A relative risk of >4 indicates strong association. A relative risk of 2-4 indicates moderate association. A relative risk of 1-2 indicates weak association. The prevalence ratio is defined as {a/(a + b)} / {a/(c +d)}. The exposure ratio is defined as {a/(a + c)} / {c/(c + d)}. For cumulative incidences < 0.05 the odds ratio is equal to the cumulative incidence ratio. As the time interval over which CI is measured decreases, the cumulative incidence ratio approximates the incidence density ratio.
Cordis gives the formula for computing the 95% confidence interval for the risk ratio as 95% CI = RR exp [+/1 1.96 {var(lnRR)}1/2 where var(lnRR) = [{(1 - a) / (a + c)} / {a} + {(1-b) / (b + d)} / {b}]1/2
The following interpretations of the odds ratio and risk ratio were given by Greenberg RS 1986: Prospective Studies in: Encyclopedia of Statistical Sciences Vol 7 p 315-319 eds S. Kutz and NL Johnson. John Wiley and Sons, New York).
i. – 0.3 strong benefit
0.4 – 0.5 moderate benefit
0.6 – 0.8 weak benefit
0.9 – 1.1 no effect
1.2 –1.6 weak hazard
1.7 –2.5 moderate hazard
>=2.6 strong hazard
The approximate equality of the OR and RR can be proved algebraically. RR = Pr(D+|E+) / Pr (D+|E-). By substituting Pr(D+|E+) = {Pr(D+) Pr(E+|D+)} / {Pr(D+) Pr(E+|D+) + Pr(D-) Pr(E+|D-)} and Pr(D+|E-) = {Pr(D+) Pr(E-|D+)} / {Pr(D) Pr(E-|D+) + Pr(D-) Pr(E-|D-)} and crossing out like terms, RR= {Pr(E+|D+) / Pr(E-|D+)} / {Pr(E+|D-} / Pr(E-|D-)} = OR.
Two rate differences can be defined, the prevalence difference and the exposure difference. The prevalence difference is defined as {a/(a + b – c)} / {a/(c + d)}. The exposure difference is defined as {a/(a + c – b)} / {b/(b + d)}.
Four measures of excess risk are used. The excess risk among the exposed is defined as {a/(a + b – c) } / {c/(c+d)}. The population excess risk is defined as {(a + c) / (n – c)} / {c + d). The attributable fraction among the exposed is defined as [{a/(a + b)} – {c/(c + d)}] / [a/(a + b)] = {(prevalence ratio – 1) / (prevalence ratio)} x 100. The attributable fraction for the population is defined as [{(a + c) / n} / n} – {c/(c +d)}] / [(a + c) / n]. This is equivalent to {(prevalence ratio –1) x (exposure rate in population)} / {prevalence ratio – 1}.
The proportion of disease due to a particular exposure is measured by various parameters of attributable rate (AR). The attributable rate takes into consideration the population at risk. It is determined as the proportion of cases in the total population that is attributable to the risk factor. AR is a measure of the impact of eliminating a particular risk factor on disease risk. There are several formulations of AR: (a) Attributable Risk (AR) = IR(exposed) - IR (unexposed) = Pe (RR-1) / 1+ Pe (RR-1) where Pe = proportion of the population that is exposed. AR <1 indicates that the risk factor under consideration is protective. (b) AR% = AR/IR (exposed). (c) Etiologic fraction/attributable fraction (d) Population Attributable Risk (PAR).
Proportional mortality studies are used to compare the proportion of deaths among the exposed to the proportion of deaths death among the non-exposed. The proportional mortality ratio has some weaknesses. It can not distinguish between causative and preventive exposures. It can also not determine to what extent the purported exposure contributed to death. The vagueness in PMR can be resolved by treating such studies as case control studies.
E. MEASURES OF DISEASE IMPACT & MEASURES OF SURVIVAL
MEASURES OF DISEASE IMPACT
A common measure of disease impact is the years of potential life lost (YPLL).
MEASURES OF SURVIVAL
If R = incidence proportion, then the survival proportion, S = 1-R. The incidence odds is R/S = R / 1-R. If R is small, S approximates 1.0 and S/R approximates R therefore the incidence odds will approximate the incidence proportion. The case fatality rate is a type of incidence poroportion. The Kaplan-Meier formula helps compute the survival proportion over several consecutive time intervals thus S = Pk=I (Nk - Ak) / Nk where Nk = number at risk and Ak = number of cases. Cumulative survival could also be expressed using an exponential formula thus S = exp (- S Ik Dtk) where Ik = incidence in sub-interval and Dtk = length of the sub-interval. The product limit and exponential formulas do not operate well in the presence of competing risks.
<Read more>
<Read more>