STUDY DESIGN (non-experimental). power ptx
- 1. OBSERVATIONAL (NON-EXPERIMENTAL) STUDY DESIGN AUWAL KABIR IBRAHIM MSC. RADIOLOGY,(BUK) PRESENTATION. 1
- 2. • Introduction • Types: Descriptive & Analytical • Measure of risk • Bias in observational study • Conclusion OUTLINE 2
- 3. Preamble • In an epidemiological study there is one or both of the following: (a) Outcome(s) of interest (b) Exposure(s) (or potential risk factors) of interest Plus • Other exposures that may influence the outcome (potential confounders)
- 4. Introduction cont. • The definition of both the “exposure” and “outcome” depends on the question you are asking • Exposure may be an environmental hazard, chemical, or something as simple as age or a genetic factor like blood group or sickle cell trait.
- 5. Cont. • Question1: Is consumption of aflatoxin associated with increased risk of liver cancer? • Aflatoxin - Exposure Liver cancer -Outcome • Question 2: Are certain people more prone to develop liver cancer after consumption of aflatoxin than others?
- 6. Cont. • An outcome in one study could be exposure in another • e.g Is alcohol consumption during pregnancy associated with increased risk of low birth weight? • Is low birth weight associated with hypertension in later life?
- 7. • Epidemiological study design are broadly classified into two(2) major groups: 1) Observational(non experimental) 2) Interventional(experimental) Introduction . 7
- 8. RESEARCH/STUDY DESIGN INTERVENTIONAL/EXPERIMENTAL DECRIPTIVE Case Report Case series Cross-sectional study Correlational study CASE- CONTROL study COHORT study Randomize d control trial Non Randomized control trial OBSERVATIONAL ANALYTICAL 7/18/2024 8
- 9. OBSERVATIONAL • The investigator does not attempt to interfere with the process being studied. • He/she observes the events that occur in a population. • In this kind of study allocation into groups on the basis of exposure to a factor is not under the control of the investigator. • Observational studies are subclassified into: 1) Descriptive 2) Analytical 9
- 10. DESCRIPTIVE STUDIES • These types of studies are the 1st stage in understanding of the disease pattern or health problem in a given population. • The researcher is often concerned about the distribution of disease (or condition, or risk factors) in the population. • The size and nature of distribution of diseases or health related problem are described without any attempt at establishing a relationship between cause and outcome. • Distribution of the disease or health problem in terms of frequency of occurrence, who(person) is affected, indices of person include basic demographic factors example age, sex, marital status 10
- 11. DESCRIPTIVE STUDIES • Where(place) it occurs. • Characteristics of place refers to geographical distribution of the disease including variation among countries and within countries including urban and rural areas. • When(time) it occurs for instance seasonal variation in disease occurrence, disease trends over decades, occurrence in excess of expectancy(epidemic) etc… 11
- 12. DESCRIPTIVE STUDIES • Descriptive studies in general are useful for planning and prioritization of health services. • They also provide clues for the formulation of hypothesis. 12
- 13. CORRELATIONAL/ECOLOGICAL STUDIES • Use data from an entire population to compare disease frequencies between different groups during the same period of time or in the same population at different points in time. • In correlational studies, the group rather than the individual is the unit of comparison. • Disease rates in various groups are usually defined as groups living within specified geographical area example countries or communities are compared. 13
- 14. CORRELATIONAL CONTD… • The variation in rate from one area to another may be explained by correlation between this rate and risk factors in these areas. • While ecological studies are used for formulation of hypothesis, they cannot be used to draw conclusions regarding individual risk because refer to groups rather than to individuals. • Our inability to draw conclusions about individual risk from group risk is called ecologic fallacy. 14
- 15. Cont. • E.g Average intake of salt in a population and the mortality rate from stroke in that same population. • Can be done for several different countries and the data examined for a relationship between the salt intake of a population and rate of disease. • It is important to understand that the only conclusion one can draw relates to populations. • It is not possible from an ecological study to draw conclusions about exposure in the individual and the risk of the outcome.
- 16. CASE REPORT & CASE SERIES • The case report is the most basic type of descriptive study of individuals. • It consist of a careful detailed report of the profile of a single patient who has some unusual presentations of a known or unknown disease. • Represent first clues in the identification of new diseases or adverse effects of exposures • Example, first case of Ebola, Covid-19 etc. • Such a cases could be investigated as clinical history, physical examination, laboratory investigation with discussion segment and could be reported through a journal as a case report. A collection of similar cases constitutes case series. 16
- 17. CROSS-SECTIONAL STUDIES • This study design is used for study of disease prevalence in a given community or population at a given point in time. • Are the simplest type of epidemiological studies. • The status of an individual with the present or absent of disease and risk factors are assessed at the same point in time. • It involves a single examination of a cross-section of a population(snap shot). • The disease occurrence is related to personal characteristics such as age, sex, occupation, educational status, social class, religion, ethnicity etc… 17
- 18. CROSS-SECTIONAL STUDIES CONTD… • It also describe in terms of place of occurrence using geographical location, sunlight, rainfall, vegetation, urban and rural areas, humidity and temperature. • The cross sectional study takes a snap shot at a point in time in a given population to describe the frequency of occurrence of a given disease and associated factors. 18
- 19. CHARACTERISTICS OF CROSS-SECTIONAL STUDY • They measure the prevalence of the disease. • They are useful for description of characteristics of those that are affected in terms of who(person), when(time) and where(place) • They are more useful for disease of long duration(chronic) rather than those of short duration. • They are relatively easier to execute and take a short time 19
- 20. Types of cross-sectional study • Cross‐sectional studies may be either descriptive or analytical. - Descriptive • Information is collected from individuals on outcome(s), or exposure(s), but not both. • They aim to provide estimates of prevalence of disease, traits such as smoking, behaviour, people′s attitudes, knowledge or health behaviour. - Analytical • Information is collected from individuals on both outcome(s) and exposure(s). • They aim to assess associations between different parameters.
- 21. Disadvantages • Unable to measure the incidence. • Not suitable for studying rare diseases or diseases with a short duration. • Cannot measure risk • Difficult to make a causal inference. • Prone to both non-response bias and recall bias.
- 22. LONGITUDINAL STUDIES • This study type is used to describe or monitor the occurrence of new cases of disease in a given population over a period of time. CHARACTERISTICS • They measure incidence rate of disease. • They are useful for studying natural history of disease and surveillance. • They are useful for studying diseases with short duration(acute) • They are costly to organize and time consuming. • They are easier to interpret. 22
- 23. ANALYTICAL STUDIES • They are concerned with the assessment of determinants/causes of diseases. • They are expected to provide answers as to why and how the disease occurs. • This is done by testing a given hypothesis which has linked a particular causative agent or risk factor to a given disease. 23
- 24. TYPES OF ANALYTICAL STUDIES • 1) Case control/Retrospective study • 2) Cohort/Prospective study 24
- 25. CASE CONTROL STUDY • Is aimed at testing a given epidemiological hypothesis regarding the association between a suspected etiological/risk factor and the occurrence of a disease by comparing frequency of exposure to the etiological factor in a group with the disease(cases) and another group without the disease(control). • The investigator select cases(those with the disease under study) and control that have similar characteristics with the cases but don’t have the disease under study. 25
- 26. CASE CONTROL CONTD… • He/she examines their previous experience in terms of history of exposure to the suspected risk factor. • The investigator looks backward in time or retrospectively. • Case control studies are relatively easier, quicker and cheaper compared to cohort studies. • It is reserve for rare disease and a number of risk factor can be assessed simultaneously. • However, it does not give a direct measure of risk instead it gives an approximation called odd ratio(OR) 26
- 27. • The starting point in case-control studies is the definition of a group of people with a particular disease or condition. • Suitable controls without the disease, and representing the population from which the cases originated, are also selected. • Information on the prevalence of past exposure in cases and controls is then collected.
- 28. • The source of the cases needs to be clearly defined. • In a population-based study, cases might be all possible individuals with the disease arising within a defined population within a fixed period of time. • In this situation the ‘base’ is the defined population, and there might be multiple sources of cases - surveillance, death certificates, pathology records etc. • Alternatively, the study may be hospital-based – the source of cases might be all patients fulfilling the case definition who attend one or more specific hospitals. • Here, the cases in the study may arise from a more selective population.
- 29. Measurement of Association • The statistic used to evaluate case-control studies is the Odds Ratio (OR) • The OR estimates the ratio of the forces of morbidity between persons exposed and not exposed to a risk factor
- 30. Concept of Odds • Odds of event can be defined as the ratio of the number of ways the event can occur to the number of ways the event cannot occur. • Odds = probability of event occurring probability of event not occurring
- 31. MEASURES OF RISK IN CASE CONTROL STUDIES(OR) • OR = Odds of exposure among cases Odds of exposure among controls 0R= A over B = AD C D BC 31 CASES CONTROL DISEASE PRESENT DISEASE ABSENT EXPOSED A B A+B UNEXPOSED C D C+D A+C B+D A+B+C+D
- 32. MEASURES OF RISK IN CASE CONTROL STUDY(OR) QUIZ Data from a case control study of exposure to the gonads on fluoroscopic procedure with and without protective lead apron and development of sterility among men aged 30-40years is as follows: Compute an appropriate measure of risk and interpret. 32 STERILITY PRESENT STERILITY ABSENT YES 23 304 327 NO 133 2816 2949 156 3120 3276 HISTORY OF USE OF APRON
- 33. MEASURES OF RISK IN CASE CONTROL(SOLUTION) OR= AD/BC OR=1.6 33 STERILITY PRESENT STERILITY ABSENT YES 23 304 327 NO 133 2816 2949 156 3120 3276 HISTORY OF USE OF APRON
- 34. Case-control study: Merits • They are particularly well-suited for the evaluation of rare diseases • Useful for testing hypothesis • Relatively cheap • Require fewer participants • There is no problem with losses to follow-up since this is a retrospective design
- 35. Demerits • Prone to biases, especially recall bias. • Can only look at one outcome. • Selection of appropriate control group may be difficult. • Cannot measure incidence. • Problem in sorting out the temporal relationship between exposure and disease • Not suitable for studying rare exposures.
- 36. COHORT STUDIES • This derives its name from the term cohort, which means a group of people who share a common experience within a defined period. • Some example of cohort include birth cohort, marital cohort, occupational cohort, resident preparing for membership/fellowship examination etc… • Cohort studies are prospective studies, because they look forward in time from exposure to disease • They are used in investigating the relationship between a risk factor and disease. • In this type of study the investigator compares the incidence rate of disease in a group of individuals exposed to the factor(study cohort) compared to another group of individuals that are not exposed(control cohort). 36
- 37. Cont. • The key concept to remember is that in cohort studies, the exposure status is always determined before the outcome. • Cohort studies are not always conducted in real time – they can be conducted using historical records, or data collected in the past.
- 38. COHORT STUDIES CONTD… • The investigator starts with cohort of healthy individuals who are classified according to exposure to risk factor as study and control cohorts. • Both groups are followed up into the future(prospectively) and the incidence rate of the disease are compared between the two groups. • A cohort study has the advantage of giving direct estimation of risk exposure to a given factor. • However, cohort studies are expensive and time consuming. 38
- 39. COHORT STUDIES CONTD… • They are better suited for common diseases with shorter incubation period. • Overall, the evidence obtained from cohort study is stronger than that obtained from case control study. • The measure of risk in cohort study is relative risk(RR). 39
- 40. Cohort studies: Purpose • The goal of a cohort study is to identify the effects of an exposure with respect to a certain outcome • Research Question: • Is prior exposure related (or not) to the occurrence of an outcome ?
- 41. Developed disease Present (exposed) Yes No Absent (not exposed) Total a c b d a + b c + d Exposure or characteristic Cohort studies: Analysis RR = Incidence in exposed group Incidence in unexposed group = a/(a + b) c/(c + d) 41 7/18/2024 11:13 AM
- 42. COHORT CONTD…(SOLUTION) • Data from a cohort study of abdominal CT exposure and development of fetal anomalies among pregnant women aged 18- 35years is as follows RR = 5151 = 1.4 37114 42 FETAL ANOMALY PRESENT FETAL ANOMALY ABSENT YES 27 455 482 NO 77 1831 1908 104 2286 2390 ABDOMINAL CT EXPOSURE
- 43. ATTRIBUTABLE RISK • Is a measure of excess risk of disease in those exposed compared to the unexposed. • It is the difference between incidence rate of the disease among the exposed group(Ie) and the unexposed group(Io) divided by incidence rate of disease among exposed(Ie) multiply by 100. • It is express in percentage. • Mathematically expressed as: AR= Ie-Io x100% Ie 43
- 44. MEASUREMENT OF RISK IN COHORT STUDIES CONTD • Attributable risk(AR) AR= Incidence rate of disease among exposed(Ie) - Incidence rate of the disease among unexposed group(Io) X 100% Incidence rate of disease among exposed group • AR= Ie-Io x100% Ie AR= {27 } - {77} over 27 = answer multiply by 100%= 27.9% 482 1908 482 44
- 45. Cohort study - benefits • Incidence rates can be calculated therefore true relative risk can be calculated. • Useful for testing hypothesis • Cohort studies permit observation of many outcomes i.e. can yield association with additional disease as by-product e.g. coronary heart disease in smoking. • Less possibilities of introducing bias if good criteria and procedures for conducting the study are established in advance
- 46. Cohort study- demerits • Large number of subjects required • Usually requires long follow up periods • High attrition rate- loss of study unit • Participation in the study can affect the behavior of the exposed and unexposed subjects (Hawthorne effect) • Very costly • Changes in diagnostic criteria and methods with time
- 47. BIAS • Bias is an error in the search methodology that may cause the results of the study to deviate from the truth. • The main source of bias in radiology studies are: Patient Intervention/exposure Data gathering Interpretation/reporting approaches. 47
- 48. BIAS CONTD… • Bias can occur at any point during the study including recruitment, intervention, data collection, analysis and publication. • While there are some advanced statistical techniques that can be used to ameliorate the impact of bias introduced by errors in the study design, it is best to avoid introducing the bias as it may not be possible to minimize the impact of bias on the results once introduced. 48
- 49. Types of bias 1. Selection bias 2. Information bias 3. Comfounding
- 50. Selection bias • Is the distortion in a measure of association due to a sample selection that does not accurately reflect the target population. • It occurs when the selected study participants are systematically different in characteristics from the eligible participants who are not selected for the study. • Eg health studies that recruit participants directly from clinics miss all cases who do not attend clinics or seek care during study • Due to this, the sample and the target population differ in significant ways limiting your ability to generalized your findings. • Types of selection bias- attrition, sampling bias, non response bias etc
- 51. Information bias • Refers to systemic errors in measurement(misclassification) of the exposure or outcomes. • Eg wrong, incomplete or inexact recording of variables
- 52. CONFOUNDING • Confounding is a particular type of bias that results from a factor being introduced that is associated with the exposure and the outcome, but is not part of the casual pathway between the exposure of interest and the outcome, yet impacts the outcome. • Confounding is a particular concern in nonexperimental designs, whereby factors other than random chance determines which treatment a subject receives and it is these factors (confounders) that may impact outcome, rather than differences in the treatment received. 52
- 53. CONFOUNDING CONTD… This type of confounding is referred to as confounding by indication, and is a key reason why randomization , by avoiding confounding by indication, allows for unbiased assessments of treatment response. 53
- 54. METHODS OF CONTROLLING FOR SELECTION BIAS AND CONFOUNDING • Randomization • Specification • Matching • Stratification • Simple adjustment • Multiple adjustment • Best care/worse care 54
- 55. SPECIFICATION OR SAMPLE MATCHING • Specification refers to a strategy in which eligibility for entry into a study is restricted to individuals who possess a narrow range of characteristics. • This strategy simply excludes every subject who holds a characteristic(variable) distinct from the ones a priori specified 55
- 56. ANTICIPATED PROBLEM • There maybe effect modification • Restriction limits generalization • Restriction increases the level of difficulty for patients recruitment for the study 56
- 57. MATCHING • Matching is the hall mark of case-control studies. If some subjects characteristics seem strongly related to either exposure or outcome, and these confounding factors cannot be controlled by randomization, investigators should be sure that they are comparable in cases and controls, which can be achieved by the matching strategy. • Although it is usually applied in pairs of cases and controls(pair-wise matching), it can also be done in groups(frequency matching). 57
- 58. ANTICIPATED PROBLEM • Investigators cannot look at the matching variable as a predictor. • There is potential of over matching. • Challenging to match on many variables. 58
- 59. STRATIFICATION • Is a strategy that analyzes data and presents results according to subgroups of patients, or strata, of similar characteristics, if there is a potential confounding. • It ensures that only cases and controls with similar levels of a potential confounding variable are compared. • Subjects are divided into “strata” or subgroups depending on the level of a potential confounder 59
- 60. STRATIFICATION CONTD… • The relationship between the variable of interest and the outcome is then separately analyzed for each stratum. • The major problem with stratification is sample size and power, since this strategy decreases the sample by subdividing it further, resulting in lower likelihood of demonstrating a significant association when present if the study sample size is inappropriate. 60
- 61. REGRESSION ADJUSTMENT • Regression adjustment is a strategy that exists because the relationships among variables can be complex, with variables being related to one another as well as to outcome of interest. • In addition, the effect of one variable may be modified by the presence of others, and the joint effects of two or more variables can be greater than the sum of their individual effects. • The rationale for the utilization of this strategy is to investigate potential confounder variable(s) and its(their) association with the outcome, integrating this information into the analysis, which is done using multivariate regression models 61
- 62. EFFECT MODIFIER • Whereas a confounder is associated with both exposure and outcome, an effect modifier is a variable that although not associated with the outcome and differentially (positively or negatively) modifies the observed effect of an exposure(risk factor or treatment) on the outcome(disease status). • Different groups have different risk estimates when effect modification is present and, therefore, an estimate of the impact of the exposure on the outcome for the entire sample may lead to an incorrect conclusion. 62
- 63. STROBE STATEMENT • Strengthening the Reporting of Observational Studies in Epidemiology(STROBE) to improve the quality of reporting of studies. • Most journals require that reports from observational studies conform to a standardized format such as STROBE in order to be considered for publication. 63
- 64. STROBE STATEMENT • Title and abstract • Introduction • Methods • Results • Discussion • Other information. 64
- 65. Conclusion • Observational study make important contribution to the knowledge of the distribution and causes of the diseases • Observational studies are often the best approach, particularly for long period of observation, for rare effect or when experimental studies would be un ethical.
- 66. THANK YOU FOR LISTENING