Presentation on theme: "Descriptive Studies Hui Jin"— Presentation transcript:
1 Descriptive Studies Hui Jin Department of epidemiology and health statisticsSchool of Public Health
2 A descriptive study is “concerned with and designed only to describe the existing distribution of variables, without regard to causal or other hypotheses.”
3 Descriptive studiesDescriptive studies often represent the first scientific toe in the water in new areas of inquiry.Case reports, case-series reports, cross- sectional studies, and surveillance studies deal with individuals, whereas ecological correlational studies examine populations.A frequent error in reports of descriptive studies is overstepping the data: studies without a comparison group allow no inferences to be drawn about associations, causal or otherwise.
4 Descriptive Epidemiology Traditional descriptive epidemiology has focused on three key features: person, place, and time,4 or agent, host, and environment.5An alternative approach is that of newspaper coverage. Good descriptive research, like good newspaper reporting, should answer five basic “W” questions— who, what, why, when, and where—and an implicit sixth question, so what?
5 Descriptive Epidemiology WHEREWHAT?PERSONPLACETIMEThink of this as the standard dimensions used to track the occurrence of a disease.WHY??WHOWHEN
6 Generally, stringent criteria for case definitions are desirable. WHAT-CASE DEFINITIONDevelopment of a clear, specific, and measurable case definition is an essential step in descriptive epidemiology.Generally, stringent criteria for case definitions are desirable.
7 WHYWhy did the condition or disease arise? Descriptive studies often provide clues about cause that can be pursued with more sophisticated research designs.
8 WHO is getting the disease? PersonWHO is getting the disease?Many variables are involved and studied, but factors such as sex, age & race often have a major effect.
9 Characteristics of Person AgeSexEthnic groupSocioeconomic statusNativityReligionMarital statusOccupation
13 Time WHEN does the disease occur? “Temporal”Range from hours to decadesType of disease dictates “time” element to be usedGraphic format often usedy-axis (vertical) - frequencyx-axis (horizontal) - time
17 Incidence Rates of Cancers in Women Incidence Rates of Cancers in Men Secular ChangeSecular changes (“temporal variation”) occur slowly over long periods of timeLonger than one yearIncidence Rates of Cancers in WomenIncidence Rates of Cancers in Men
18 Short-term changes occur over limited time frames Point EpidemicsShort-term changes occur over limited time framesHoursDaysWeeksMonthsUsed for short-term exposures or diseases with short incubation and/or illness durations
23 Seasonal VariationSeasonal variation can be seen for some diseases or conditions falling within a calendar year
24 Seasonal variation can be used to suggest possible etiology. Migratory Birds?
25 Time ClusteringTime clustering data can sometimes be used to trace the “beginning” to the introduction of a specific causal agentThalidomide & birth defectsFirst marketed in Europe in 1950’s as sleeping pill and to treat morning sickness in pregnant womenToxic Shock SyndromeStaphylococcus aureus infection in women using newly introduced hyperabsorbent tamponstampons卫生棉条
29 5 Criteria of Place Rate observed in all ethnic groups in the area Rate NOT observed in persons of similar groups inhabiting other areasHealthy persons entering area get ill at same frequencyPeople who leave do NOT show similar levelsSimilar levels of infestation in other species (if zoonotic disease)
30 Characteristics Relating to Place InternationalVariation within countriesUrban-ruralLocalBuilding Maps
31 Distribution of AIDS in the US 1990 PlaceDistribution of AIDS in the US 1990
34 Interactions of Time and Place Time-place clusteringMigration
35 TYPES Ecological or Correlational studies Cross-sectional or Prevalence StudiesLongitudinal or Incidence Studies
36 ECOLOGICAL STUDIESStudies conducted in specific population having specific characteristics in a specified geographical areaCauses or risk factors are studied with regard to the diseases and deaths occurred in a particular populationBoth are linked together and their co- occurrence (Correlation) is established in these studies for hypothesis formation.These are the usual initial epidemiological studies.These are usually descriptive and can also be analytical when applied to individuals in a specific group.
37 Frequency and trends of the suspected causes/risk factors ECOLOGICAL STUDIESFrequency and trends of the suspected causes/risk factorsPrevalence or incidence of the disease occuring in a particular areaCorrelated withThese are the usual initial epidemiological studies.These are usually descriptive and can also be analytical when applied to individuals in a specific group.
38 ADVANTAGES AND DISADVANTAGES Conducted at group level, not at individual level, hence relatively easy to do and quickUse existing dataGenerate and support new hypothesisEcological studies conducted over time on a specific geographical area are more convenient to perform and form hypotheses rather than studying whole populations or its samples as done in descriptive studiesDISADVANTAGEScannot be not done at individual levelhypotheses generated are subject to ecological fallacy ( see next slide)
39 ECOLOGICAL FALLACYEcological fallacy is an error in inference that occurs when association observed between variables of a group level, is assumed to exist at an individual level.
40 EXAMPLES OF ECOLOGICAL STUDIES Cancer cervix is rare in Jewish women due to male circumcisionSickle cell disease is more seen in Indian tribes镰状细胞病
41 CROSS-SECTIONAL STUDIES (PREVALENCE STUDIES) They can be of descriptive nature when one variable or each variable in a group or population is studiedorof analytical type as they are sought to provide information about the presence and strength of association.They are conducted for chronic diseases having high prevalence with very low incidence.Conducted to know about all the cases present (all the existing cases, old and new cases, all current cases) in that community i.e. Prevalence of the disease..
42 PREVALENCE STUDY METHODOLOGY It can be done at a single point of calendar time (point prevalence) when the measurement of causal relationship relate to the same point in study members’ lives or can be completed in few months or years (period prevalence).The descriptive information is obtained by means of door to door survey.Though they are usually carried out on populations or samples of population, they are individual based. They seek the information of about the individuals in a group or population.It is a one time study, but can be extended to include historical information that can be easily collected at the same time, in which house to house information was obtained about all cases present.Cross sectional studies can be conducted in community viz. on leaders, mothers, health workers, using interviews, group discussions, direct observations, and household interviews.
43 RAPID METHODS OF PREVALENCE STUDIES Cluster surveysComputer simulationRandom digit dialing for samplingComputer based interviews & use of spreadsheetWhen the cross-sectional studies are repeatedly done, they will serve the purpose of health and disease surveillance of the population.
44 DISADVANTAGES Time consuming and expensive Subjected to recall bias and confounding biasOther common biases encountered in these studies are of berkesonian, information and investigator typesUnder-represent people with a short-course of diseaseLimited to studies of causes that are of long standing natureThe validity of the associations derived of these studies depends on careful elimination of the artificial associations and confounding effects.
45 LONGITUDINAL STUDIESIn a longitudinal study design, the study is conducted at two or different points of time in the life time of individuals under study in contrast to the one time study of cross-sectional design.INCIDENCE STUDY, FOLLOW UP STUDY, PROSPECTIVE STUDYOTHER NAMES :-INCIDENCE STUDY ( AS INCIDENCE CAN BE MEASURED DIRECTLY),FOLLOW UP STUDY (AS CONTINUOUS FOLLOW UP IS ESSENTIAL)PROSPECTIVE STUDY (AS THE DIRECTION OF THE STUDY IS FORWARD-LOOKING.Cohort studies are of this design. These studies, though yield incidence directly, are time consuming and expensive when compared to cross- sectional onetime studies.It should be remembered that the follow up given during the disease management e.g. follow up of patients with tuberculosis during treatment, is not to be confused with these longitudinal studies.
46 Defining and describing the disease (Operational case definitions METHODOLOGYDefining and describing the population affected (reference population and study populations)Defining and describing the disease(Operational case definitionsTime, place and person trends)Formulation of hypothesis
47 POPULATION AT RISKIt should be remembered that entire population is not going to be studied except the relevant population .For e.g. for cancer cervix studies, we choose the women in reproductive age, not female children and certainly not males.Reference population is the one, which has to be get benefited after our descriptive study.Thus the population at risk is better one to be studied rather than the total population while formulating hypothesis as the inferences after testing hypothesis cannot be generalized to all the components of whole population except to the specific and vulnerable/susceptible portion of it.
48 REFERENCE POPULATIONS The population which is at risk and the health action has to be initiated after the study is over is the reference population.OrThe population in which a particular disease or exposure has occurred and is to be investigatedReference population is the one, which has to be get benefited after our descriptive study.For example, if we are studying Scabies incidence in a sample of school children, all the school children comprise our reference population. It is the population to which ,we apply or extrapolate the end results or inferences obtained after testing our hypothesis.These populations may be as narrow and small as exposed groups or as big and universal as for example, all reproductive women.It usually serves as denominator to calculate rates and ratios to quantify the disease/exposure load
49 A representative portion of it (sample). STUDY POPULATIONthis is the population on which the hypothesis is actually studied and tested.It may be the entire reference population itself (if it is small and feasible for study)ORA representative portion of it (sample).This serves as a refined denominator for calculating and quantifying the disease load.The study population characteristics have to be described in detail for proper formulation and elucidation of hypothesis.
50 DESCRIBING THE STUDY POPULATION AgeSexOccupationsSocioeconomic statusLiteracy profileSocial customs, habitsSpecific lifestylesKnowledge of health facilities available and their utilization
51 VITAL REQUIREMENTS OF STUDY POPULATION 1. Its representativeness to the parent reference population2. Its optimum sizeThen only the results of hypothesis tested on any study population can be generalized to the reference population.
52 DEFINING AND DESCRIBING THE DISEASE Defining the disease (case definition)An operational working definition to make uniform and unbiased counting in populations
53 APPLICATIONS To formulate causal hypothesis. Indicate the disease load and frequency alterations and thereby help to make future projections.Diagnosing and telling the prognosisDisease clustering offers the good opportunity to study the natural history of disease and also for controlling it e.g. Observation by John Snow of cholera clustering at Broad street in London square led to understanding of mode of transmission and its control.Disease mapping is an useful tool to know the geographical pathology and for health planning basing on local conditions and resources
54 JURISDICTIONThe dividing line between the descriptive and analytical studies is not very sharp.as individuals, groups , communities and the total population and their characteristics are interrelated and intermingled and any study conducted on them is flexible and not typical.Ecological study can be a group based analytical study in addition to be of descriptive nature.Cross-sectional study can be both descriptive as well as analytical one, descriptive when one variable or each variable in a group or population is studied, analytical as it provides information about the presence and strength of association. (oxford textbook of public health)
55 ECOLOGICAL AND INDIVIDUAL LEVELS OR CONTEXTS The hypothesis to be studied will be more meaningful when it is applicable at both ecological and individual levels or contexts and also explains the ecological (group) to individual correlation.The line between the ecological and individual contexts is the dividing line between descriptive and analytical studies and is not sharp and often overlapping.e.g cross sectional studies can be both descriptive and analytical. One practical distinction between the descriptive and analytical studies is that former are based on the study and analysis of routinely collected data while the later from ad hoc collection of information.
56 FORMULATION OF HYPOTHESIS Sources to form hypothesis?1) Descriptive studies2) Ecological studies on specific groups3) Keen observation of the data/information available4) Inductive reasoning (MILL’S CANNONS)5) Deductive reasoningSeveral hypotheses can be made suspecting the time, place, person variables obtained from descriptive studies as causes for the disease occurrence or its altered frequency.This is the very usual way of getting ideas about hypothesis formation. If one thinks and analyses why a particular disease has occurred at a particular time or place and why it has affected people of certain ages and of certain groups only, then a number of hypotheses will emerge.Ecological studies, the studies conducted on specific groups with specific characteristics, also provide useful information for forming the hypothesis. Examples are that Cancer cervix is rare in Jewish women due to male circumcision and Sickle cell disease is more seen in Indian tribes.
57 DATA AND HYPOTHESISKeen observation of any data, collected for specific purpose, and purpose or without purpose can also yield information for hypotheses forming.Political & Natural observation of Bills of Mortality in London “ by John Graunt yielded lots of hypotheses to be verified.It is often observed in the development of science that many unexpected discoveries were made while observing for some thing else.Data treatment in terms of tables , graphs, rates and ratios will provide material for hypothesis formulation.
58 SUMMARYDescriptive epidemiology is hence rightly called the hypothesis forming stage of epidemiological sequence as descriptive epidemiology is very useful in providing immense information regarding the various variables like time, place, person, clustering, etc to form the hypothesis.As Descriptive epidemiology is very useful in providing immense information regarding the various variables like time, place, person , clustering etc to form the hypothesis.
59 Question? Assigned readings, session 2: Grimes DA, Schultz KF. Descriptive studies: what they can and cannot do. Lancet 2002;359:145-9.