Lecture 9 Association & Validity PDF

Measures of association & Validity Community Medicine & Epidemiology 1st semester / Year 3 2024-2025 Dr. Sanabel Barakat DDOS. MSc. PhD. Week 9 JCD.  ILOs Learn about validity and reliability Under...

Measures of association & Validity Community Medicine & Epidemiology 1st semester / Year 3 2024-2025 Dr. Sanabel Barakat DDOS. MSc. PhD. Week 9 JCD.  ILOs Learn about validity and reliability Understand the difference between external and internal validity Discuss measures of association for different study designs 2  VALIDITY & RELIABILITY 3  Successful studies need to address two important dimensions: 1-Reliability 2- Validity: Internal External 4  Reliability A reliable study should be replicable, providing similar results if the same study parameters are applied (repeatability). 5  Validity Validity is concerned with the ability of the study to correctly answer the question it asks. Validity is the degree to which a test is capable of measuring what it is intended to measure. A study is valid if its results correspond to the truth; there should be no systematic error and the random error should be as small as possible. Epidemiological studies should ideally have internal and external validity. 6  Internal Validity Internal validity deals with the ability of the study to correctly infer about the relationship between the independent variables and the outcome(s) being studied. Internal validity is the degree to which the results of an observation are correct for the particular group of people being studied. E.g. periodontal examination and measurements of probing depth using a specific type of periodontal probes must distinguish healthy from patients with periodontal disease as defined in the study/ Classification systems. 7  Internal Validity For a study to be of any use it must be internally valid, Internal validity can be threatened by all sources of systematic error(Bias) but can be improved by good design and attention to details. Internal validity refers to the possibility that the conclusions of an investigation are valid for the sample, with no systematic errors or biases. Internal validity relates to the methodological and statistical dimensions of an epidemiological study.. 8  Internal Validity In order to obtain internal validity, - the comparability of the groups should be ensured (allocation), - the accuracy in the diagnostic technique (measurement tools) should be ensured, - the control over the factors that may hinder the interpretation should be ensured 9  External Validities The success of epidemiological studies relies on its inferential ability. E.g. it is hoped that a survey of caries in a sample of 12-year-old schoolchildren in a given city can produce inference for the group of 12-year-old children in the city. For this aim, it is necessary that epidemiological studies have external validity along with internal validity, ensuring that the data obtained can be extrapolated to the broader universe from which their samples were selected. 10  External Validities (Generalizability) External validity or generalizability is the extent to which the results of a study apply to people not in it External validity corresponds to the ability to generalize the results of a particular study, applying them to the population from which the sample was selected or to other populations. External validity deals with application of the findings to other observations, samples, or populations and implies generalizability of the study results. Consider with methodological and statistical aspects, such as the criteria for calculating and selecting the sample, the possibility of inference or extrapolation should be evaluated in face of the conceptual framework on the subject that is being investigated. 11  Internal and External Validities Randomised clinical trials provide a good illustration of the difficulties of epidemiological studies to present both internal and external validity. As previously mentioned, these studies have strong internal validity, due to the many methodological requirements for their accomplishment. However, these studies are often subject to external validity restrictions, depending on the specific characteristics of their samples. Most of the randomised clinical trials are conducted in high-income countries, and only individuals who met various selection requirements are researched, which prevents the results from being extrapolated to the general population. 12 Validity and reliability With low reliability but high validity the measured values are spread out, but the mean of the measured values is close to the true value. On the other hand, a high reliability (or repeatability) of the measurements does not ensure validity since they may all be far from the true value. 13  Measures of Association 14  Measures of Association Many epidemiological studies have the objective of evaluating the association between exposures (risk factors or protection) and an outcome, for this purpose, association measures that can be expressed in different forms. Contingency tables constitute a viable resource for calculating these measures. 15  Measures of Association Epidemiologic studies generally follow a series of steps that are called the “epidemiologic sequence”—a misnomer because the sequence is often disrupted. This “sequence” includes: observing by counting cases and events, relating cases and events to the population at risk, making comparisons, developing hypotheses, testing hypotheses, making scientific inferences, conducting experimental studies, intervening, and evaluating. 16  Measures of Association Ultimately, epidemiology examines the associations between sets of events, defined as outcomes, and determinants of those outcomes. An outcome in one study may be a determinant in another study. Similarly, a disease may be an outcome in one study but an exposure in another study. 17  Measures of Association Several associations may exist between factors, but not all associations are causally linked. The key questions that epidemiology tries to answer are: (1) is the observed association real or spurious? (2) is the association causal (i.e., exhibiting a cause–effect relationship)? 18  Measures of Association In epidemiology, determinants of diseases are often called “exposures,” which may be causative factors or protective factors for diseases. To find associations, construct the contingency tables (the fourfold or 2 × 2 table) which classify the population quotas according to characteristics of the exposure and the effect considered 19 CONTINGENCY TABLES (2 X 2 table) a represents the number of individuals exposed and sick; b represents the number of individuals exposed and not sick; c is the number of individuals who are not exposed and who are sick; d indicates non-exposed and non-diseased individuals. In addition to these values, the table displays the partial totals in each row or column: a + b = total of exposed individuals; c + d = total of unexposed individuals; a + c = total of sick individuals;and b + d = Total non-diseased individuals. 20  Measures of Association Based on data from contingency tables, it is possible to calculate different measures of association between variables related to exposure and disease measures 21  Measures of Association in experimental studies Calculation of the incidence measures demands the organisation of the data in time, the risk measures, as relative risk, must be estimated in prospective longitudinal studies, i.e. interventional and cohort studies. Clinical trials studies try to find What is RR, ARR, NNT? The relative risk (RR) describes the risk of the disease among the exposed Compared to the risk of the disease among the non-exposed. 22  Interventional Studies RR = incidence in the exposed to intervention/ incidence in non-exposed RR = Ie / Io RR >1 : incidence in the exposed group is higher than the incidence in the non-exposed, suggesting that exposure is a risk factor. RR or < c / (c + d )? (I.e/I.ne)? What is Risk Ratio, RR 27  RR RR = incidence in the exposed to intervention/ incidence in non- exposed RR = Ie / Io RR >1 : incidence in the exposed group is higher than the incidence in the non-exposed, suggesting that exposure is a risk factor. RR or < b/(b + d)? & Odds Ratio = ad/bc ?? 31  Periodontal No Odds Ratio (OR) smoker disease 70 disease 30 100 nonsmoker 30 70 100 OR can be used in case-control 100 100 200 OR = (a / c) / (b / d ) = ad / bc The odds ratio is the probability of occurrence of an event to that of non- occurrence. For example, if 70% of those who smoke develop periodontal disease, while 30% do not develop the condition. The odds between smokers and non-smokers for the occurrence of periodontal disease are 70/30 or 2.3. OR= 70 X70 /30X30 = 5.4 This measure indicates that the chance of developing periodontal disease is 5.4times higher among smokers than among non-smokers. 32  Association in CROSS-SECTIONAL STUDIES Odds Ratio (OR) may be calculated for groups comparisons Prevalence ratio (PR), may be calculated by comparing the measure of prevalence obtained for both groups, with and without exposure. PR is used where the outcome (dependent variable) is relatively frequent (higher than 15%), because in this case, the OR tends to overestimate the PR 33  Measures of association 34  Complete the table Study design Measure of association Randomized Clinical Trial Cohort Case-Control Cross sectional 35  Thank you & Good Luck 36