Epidemiological Study Designs PDF

Epidemiological Study Designs: By Dr Hewaida el Shazly  Types  A) Descriptive Study:  Describe general characteristics of distribution In relation to time, place and person. Importance of descriptive study:  The first clue for disease determinant in order to formulate hypothesis. Desc...

Epidemiological Study Designs: By Dr Hewaida el Shazly  Types  A) Descriptive Study:  Describe general characteristics of distribution In relation to time, place and person. Importance of descriptive study:  The first clue for disease determinant in order to formulate hypothesis. Descriptive studies include:  a- Case report: It is basic type of descriptive study ft is careful detailed report by one or more clinician of profile of single patient.  it is usually report of uncommon or unusual events as drug reactions. Descriptive continued  B- Case Series:  Definition: Report of clinical characteristics from a group of clinical subject  Study question : any clinical question, usually to suggest emergency disease  For example-. Congenital anomalies of children of mother receive Thalidomide Descriptive continued  C- Cross-Sectional Survey:  is also called (incidence - prevalence study).  Methodology: a survey in which a status of an individual with respect to presence or absence of disease and exposure is assessed for point of time using questionnaire, examination, screening test.., etc. Rate of incidence and prevalence can be calculated  Since exposure and disease assessed at point of time we can't distinguish whether exposure precede disease or exposure or presence of disease affect patient exposure  D-Correlation study: b) Analytic ( explicit type of comparison ) epidemiology observational analytic  a. Case control  b. cohort study  Intervention study  A- experimntal trial  B- clinical trial CASE-CONTROL STUDY:  It is type of observational study in which a case group of patients who have disease of interest and a control group (comparison group) without disease studied are selected for investigation and exposure of each group are compared to evaluate association between certain exposure and disease.  Steps of case control study 1- case selection Establish strict diagnostic criteria for disease.. Establish case selection criteria. Hospital based (case-control).  It is the commonest because non expensive, easy. population based. The advantage of population based is to avoid selection bias. The primary concern of the study should be validity (complete and reliable information on disease and exposure) rather than generalizability.  Steps continued 2-control selection  Control selection is the most difficult issue in case-control study. The case inclusion and exclusion criteria should be included for control to be comparable. So, they should be selected to be representative for case (except for study disease manifestation) not for population. Control types  Types of control group:  1- Hospital based.  2- Population based.  3- Friends, neighbor and relatives.  Hospital hosed control;-  Are patient of same hospital from which cases are selected but with other condition than studied disease.  Advantages of hospital control:-  Easily identified.  Available in sufficient number.  Minimize effort and cost.  As they are ill, they will be aware of past exposure or events and accuracy of reporting information and low bias.  Willing to cooperate.  Disadvantages:-  As they are ill, they differ from healthy as  May not accurately represent exposure distribution in population.  If patient with disease known to be associated positively or negatively with exposure of interest should be excluded and when patient (cases chose from hospitalized and control selection from hospital is not scientifically Feasible, chose control group from general population.  Overestimation.  Population based control:-  By random selection of sample from population who are apparent healthy or not have disease of interest usually selected by random digit, telephone, population identification from population registry or voting list  Disadvantages:-  More costly, time consuming, more difficult  Population list are not always available.  Difficult to contact healthy with busy work.  Quality of information may differ from case and can't recall  Less motivated to cooperate, Types of control continued  Third source of control (friends, relatives and neighbour) :-  Advantages: -  Being general population control, healthy.  Cooperative because they are interest in case.  Offer degree of control to confounders as ethnic, socioeconomic, and environment  Disadvantages  We should put in mid that they may share exposure which leads to underestimation as diet, smoking, drug abuse.  In this situation it w+lls not an adequate group. Control selection  The use of multi-control group is also indicated when there is a concern that one selected group has specific deficiency that could be overcome by including another control group.  Case-control ratio :-  Optimal case to control ration is 1:1.  If number of cases are limited or cost of obtaining information for cases is high , sample size can alter by ratio not exceed 1 : 4 as there will be small increase in power of study ( unless extra data will be with little extra cost ). 4-Data collection  Information on disease, and exposure should be obtained from any source including:  Review of records as death certificate, case registry, ongoing surveillance, physician records, hospital admission.  Interview or mail questionnaire: information about exposure obtained by questionnaire for cases themselves or relative or spouse or mother or guardian for children.  Direct measurement of the exposure if possible. Measures for cases should similar for control in same place and circumstances Case control  Disadvantages:  Selection bias: error in selection of cases and controls.  Information bias: - unavailable or inaccurate records.  Recall bias:- inaccurate recall of events in the distant past  Confounding bias: - there are variables that increase or decrease the association between the disease and the factor under study.  Yields: - only an approximation of relative risk (odds ratio) but cannot measure incidence (no population base).  Uses:  Particularly useful for etiologic study of rare diseases (high exposure)  Analysis of case-control study :- It is basically comparison between case and control regarding exposure whose potential etiologic role is being evaluated. This comparison is made by estimating relative risk as computed by odds ratio. 1. Present data in 2x2 tables. Cases Control Total Exposed a b a+b Non-exposed c d c+d Total a+c b+d A+b+c+d  Measuring risk in case control study  Odds ratio  Odds ratio (OR)= ad/bc  Attributable risk percent (AR %) AR% = OR-1/OR *100  We can't measure the Attributable risk because incidence among exposed, and non- exposed not present.  Example:- data from case-control study of oral contraceptive user and myocardial infarction in premenopausal female nurse. Of 156 women with myocardial infarction 23 were oral contraceptive user. Of 3120 control without myocardial infarction 304 were oral contraceptive user. Calculate the odds ratio- and what does it mean?  solution Son-myocardial Oral contraceptive Myocardial infarction Total infarction Exposed 23 304 327 Non-exposed 135 2816 295! Total 156 3120 3278 Odds ratio (OR)= ad/bc 232816/304133 = 1.6 Attributable risk percent=1.6-1/1.6*100 (AR %)=37% Interpretation  OR= 1.6 means exposure among cases is 1.6 times exposure among control, so it may risky.  Interpretation  If OR - 1 no difference.  If OR > 1 a causal association between the exposure and outcome.  If OR < 1 means exposure is protective. Cohort study The Second major type of observational analytic study. Definition: A group of individual are defined for being exposed or not exposed to suspected risk factor/ with all subject free from disease at time of exposure that, then followed up to assess disease occurrence. Strength : Valuable when exposure is rare Can study multiple outcome of single exposure. Can elucidate temporal relationship between exposure and outcome. If prospective low bias. Allow measure of disease incidence among exposed and non exposed. Lack of bias Types of cohort study ◦ Retrospective cohort. ◦ Prospective cohort. ◦ Case-control nested in cohort study.  Retrospective cohort study(non concurrent):-  Both exposure and outcome have already occurred when study initialed, individuals are classified on basis of exposure and followed up in the record to see occurrence of disease i.e. exposure should precede outcome.   Strength of retrospective cohort study:-  It is more quick and cheap than prospective  Efficient for investigation of long latency period.  Limitation of study: -  Since it evaluate exposure occurred many years in the past they depend on routine availability of relevant exposure data in adequate details from records.  Since these data are usually reported for other purpose, they may incomplete and possibly non comparable.  Often, the potential confounder information as diet, smoking, life style are not recorded in these records.. Prospective cohort study(concurrent):- The exposure may or may not occurred at time of study begin, and then followed into future to assess outcome and incidence of disease. Investigator use recent record or questionnaire or assess exposure directly and collect information about confounders and follow up for long period to assess disease occurrence. Limitation of study:- Time consuming and costly. Losses to follow up subject of the study may make results validity seriously affected.. Case-control nested in cohort study  For example:. show base line blood collection and store it to be evaluated for antioxidant follow up and take sample, store it, and then follow them till developing disease of sufficient number compare it to non. diseased matched sample. Then analyze blood for antioxidant for both case and selected control. This will safe money instead of analyzing blood of hundred of people. So it can be used for expensive tests.  Steps of cohort studies  1- Exposed population selection:  For relatively common exposure we could collect them for a number of possible populations. The issue is to chose group from them we can easily collect relevant data as doctors, nurses, workers in various occupation, union members, veterans, students, residents of well defined countries (whose people have stable job, residency) and allow investigator to follow up for years.  For rare exposure, select a specific group of exposure who may include individual in certain occupation or who undergo a particular medical therapy as x-ray for T.B or certain place residency (Japan suspect to pomp), certain religion as has certain diet or life style. This certain exposure group selection will:  Allow accusation of sufficient exposed individual at suitable period of time  Allow study of rare outcome as rare disease in population may be sufficiently common in special exposure group.  2-Selection of unexposed group :-  It is important to ensure that the information can be obtained from non exposed group are adequate for comparison, and the group should be similar as possible with respect to all other factors that may related to disease except the determinant under investigation..  Comparison with population rate can be used only for available rates as cancer rate, mortality rates,).  The major disadvantage of using general population as comparison group is employed people is healthier than non worker and deaths of general population are usually higher than deaths among  Data collection  interview  The pre-existing records:-  Advantage: ◦ Available for high percentage. ◦ Inexpensive ◦ Non time consuming ◦ Unbiased classification because recorded for other reason. 3-Approach to follow up in cohort study  Following all study participants from point of exposure and outcome (retrospective or prospective).  Failure to follow is major source of bias, so uninterpretable results Collecting follow up data is major challenge in cohort study with major cost and time expense.  Length of follow up depends on disease latency period for outcome of interest. For example:  Congenital anomaly and acute illness need short period  Cancer has long latency period and need long observation.  The longer the period, the more difficult to achieve complete follow up because movement, job change, so, list of relatives or friends phone, vehicle records, and employment records is important.  4- analysis of cohort study  Calculation of:  The incidence of outcome,  The incidence among exposed and non-exposed. , Outcome Positive Negative Total Exposure Exposed A B a+b Man-exposed C d c+d Total a+c b+d a+b+c+d Relative Risk (RR): Incidence among the exposed Relative Risk (RR) = Incidence among the non-exposed a/(a+b) Relative Risk (RR) = c/(c+d) Attributable Risk (AR): Attributable Risk (AR) = The incidence among exposed - The incidence among non-exposed. Attributable Risk (AR) = a / (a+b) - c / (c+d).  Example Example: The following table shows data from cohort study of oral contraceptive use in relation to bacteriurea 2390 female aged 15 - 50 years who were free of bacteriurea 482 oral contraceptive users, and 1903 non user. At second survey 27 of the users and 77 of oral contraceptive non user develop bacteriurea- Calculate the Relative Risk (RR), Attributable Risk (AR). and Attributable Risk Percent (AR %}.  Oral contraceptive Bacteriurea Non bacteriurea Total Positive 27 455 482 Negative 77 1831 1908 Total 104 2286 2390 1. Relative Risk (RR): Incidence among the exposed Relative Risk (RR)= Incidence% among the non-exposed 27  482 Relative Risk (RR)= = 1.4 77  1908 2. Attributable Risk (AR): Attributable Risk (AR) = the incidence among non ex-exposed Attributable Risk (AR)= (77482) - (77 1908) Attributable Risk (AR) = 0.0560165 - 0.04356 = 0.01566 3. Attributable Risk Percent (AR %); Attributable risk percent (AR %) Attributable Risk Percent (AR %)= Incidence among the exposed 1566 / 100000 Attributable Risk Percent (AR %)= = 27.96 % 0.0560165  Interpretation:  The incidence of the disease (outcome) is 1.4 more among the exposed than non-exposed.  Issues in interpretation in cohort study:-  Rote of bias : ◦ Selection bias:  In retrospective cohort as both exposure and outcome are occurred at start of study as knowledge about disease will affect selection of exposure or classification (not in prospective) ◦ Misclassification: i.e. some exposed considered non-exposed and visa versa. Disadvantages:  Require large number of subjects.  Long follow up period.  Attrition: - (loss of subjects from follow up) because of disinterest, migration, or death from other causes or potential loss of staff and/or funding.  Potential change of exposure status of subjects over time.  Changes in diagnostic criteria and methods over time with advances in technology.  Very costly. CLINCAL TRAILS  Advantage:.  1- Randomized clinical trial represents the gold standard for epidemiological research as it provide a degree of assurance about test validity which is not possible in any other design,  1- 2- In evaluation of treatment with small or moderate effect, it is very difficult to establish reliability by other design as treatment evidence.  INTERVENTION STUDY  It is type of cohort study.  Two types of intervention study:  a. Experimental tails,  b. Clinical trails. Types of clinical trials  Types of intervention studies (clinical trails):-  Therapeutic clinical trails  Preventive clinical trails.  Therapeutic trails:  Done among patient to reduce symptoms. Prevent recurrence and decrease disease, e.g. radical mastectomy showed no significant effect on survival from less extensive surgery.  Preventive trials:  Involve the evaluation of agent or procedures in decreasing risk of developing disease among those who are free on enrollment, so they conducted among healthy individuals. why  The purpose of clinical trial is to compare a new agent, drug or vaccine with a traditional one with regard to it's:-  Effectiveness.  Safety (toxicity and side effects).  Cost-effectiveness. Phases of Clinical Trials  Phase 1: 15-30 people ◦ What dosage is safe? ◦ How should treatment be given? ◦ How does treatment affect the body?  Phase 2: Less than 100 people ◦ Does treatment do what it is supposed to? ◦ How does treatment affect the body?  Phase 3: From 100 to thousands of people ◦ Compare new treatment with current standard  Phase 4: From hundreds to thousands of people ◦ Usually takes place after drug is approved ◦ Used to further evaluate long-term safety and effectiveness of new treatment.  steps of clinical trials  A) Selection reference population  Reference population, which is a group representative of population for generalization, the investigator expects results to be applicable.  They may be restricted to age, sex or geography.  1- Size: they should be of sufficient size to achieve requirement.  2- Number of outcome; they should be expected to experience sufficient number of outcome.  3- Accurate information there must be a likelihood of obtaining accurate, complete follow up information for period of trial i.e high follow up rate. Steps continued  Eligibility and consent  Screen population for eligibility and explain if excluded.Those who are willing and eligible are the actual study population.  We compare data age, sex, SES of those willing and not willing to give idea that they didn't differ and there is no non respondent bias.  If they differ it may help in assessing this difference. These data can obtain from records, sometimes it is difficult.  2-Allocation of study regimen :  Assignment for study group should be done at random.  If outcome will be affected by certain subgroups as sex. SES, disease stage, do block randomization on every participant is classified into block and then chose from each block at random.  Advantage of randomization and block randomization:-  Removal of potential selection bias (as physician may allow seriously ill to receive treatment while, other receive placebo, this will lead to underestimation of drug benefit.  Adjust for confounders. as they will be equally distributed (both known and unknown confounders).  Favorable impression of design for those reading published paper.  Maintenance and assessment of compliance:  Be sure that participant didn't deviate from protocol due to forgetting treatment or side effect or chose alternative treatment or non compliance due to long period of treatment.  Try to enhance compliance by select people with stronger motivation and by frequent visit, phones, mail. Monitor compliance because non compliance reduces power of study.  Also implementation of run in strategy in which before randomization all participant either receive treatment or placebo for week or months to exclude those who can't comply before randomization, and there will be no effect on study. However restrict study on compiler who may differ from population.  How to monitor compliance:  Self report.  Count pills.  Biochemical parameter (expensive).  Bias and clinical trial:  I- Observational bias:  The potential for observational bias in ascertainment of outcome in an interventional study can exist as  Knowledge of participant treatment may influence reporting of relevant event and this is related to subjectivity of outcome.  If outcome is objective and can't be affected by knowledge as death, observational bias will be minimal.  If outcome is subjective like clear cut and may influence by clinician knowledge of treatment, assignment as severity of illness, decreased pain, frequent side effect..etc, there will be systematic difference in ascertainment of outcome.  Solution Jo this bias:-  1- Keep study investigators and participant blinded until data collection completed.  In single blind study: only participant are blinded to treatment received.  In double Mind study: both investigator and participant are blinded i.e. neither investigator nor participants know to which treatment group an individual has been assigned Unblind or single blind are much easier to be executed than double blind Also in certain design, it is difficult to use blindness as diet use, exercise, smoking for evaluating program (β carotene and yellow coloration of skin, operative and medical treatment group... etc).  In triple blind study: the statisticians i.e who do the statistical analysis also blind in addition to both investigator and participant.  Other problems in interventional trial:  Tendency for individuals to report favorable response to any therapy regardless the physiologic efficacy and what they receive This phenomenon is referred to placebo effect so you should use placebo control.  Person taking drugs or undergoing a medical procedures may be sensitized to their condition and describe any symptoms or unusual occurrence to their treatment  Example: From individuals who received Aspirin 24% develop GIT symptoms, peptic ulcer, and from those who received placebo 15% develop G(T symptoms.  You found 24% attributed to aspirin if no placebo and only 9% (24%-15%) attributed to aspirin when placebo is used, that is the placebo control.  When to terminate intervention study: slopping rules or modification rules. ◦ If data indicate that one treatment is clearly harmful. ◦ If data indicate clear and extreme benefit on primary end point as when treatment is beneficial and-this appears clearly, it is not ethically to continue holding treatment from placebo.  Issues in analysis:  Decrease bias as discussed by :  a. Randomization leads to minimize allocation group bias  b. Sufficient sample size.  c. Blindness leads to minimize observation outcome bias.  Confounder randomization:  a. Distribute confounder equally in treatment group, also sample size lead to decrease risk of chance.  b. Matching in base line characteristics.  c. Never exclude patients after randomization this lead to bias.  d. In analysis study the effect among compiler, non compiler and 9  e. Keep losses to follow up to minimum.  Ethics in experimental studies  There should be a written consent with brief summary of purpose, nature of experiment, likely risks and benefits of experiment, procedure listed particularly emphasis is given to potential danger.  A guarantee guard the privacy of medical results is provided and voluntary nature of participation and right to withdraw at any time from the study. There should be Institution Review Board (IRB) or human subject committee composed of medical personnel, a , purpose review research proposed for both scientific and ethical problems. IRB approval requires appropriate consent form. Thank you