Cohort Studies PDF

HLSC 2P27 COHORT STUDY Mostafa (Mo) Shokoohi, Ph.D. Epidemiology and Biostatistics Assistant Professor Departm...

HLSC 2P27 COHORT STUDY Mostafa (Mo) Shokoohi, Ph.D. Epidemiology and Biostatistics Assistant Professor Department of Health Sciences, Brock University 1 COHORT STUDIES A cohort is a group of similar people followed through time together. A cohort study is an observational study that follows people forward in time so that incident (new) cases of disease can be recorded. 2 2  PREVALENCE (PREVALENT CASES) Gordi Epidemiology 3 3 INCIDENCE (INCIDENT CASES) Time Gordi Epidemiology 4 4  FRAMEWORK FOR A COHORT STUDY 5 5 MEASUREMENT TIMES IN COHORT STUDIES Cohort studies have at least two measurement times: 1. At baseline to determine exposure and disease status A baseline is an initial measurement used as a benchmark for examining changes over time. 2. One or more follow-up assessments to determine the development of new (incident) diseases Incident diseases: participants who did not have the disease of interest at baseline, but the disease was developed during the study (after baseline). 6 6  TYPES OF COHORT STUDIES Prospective cohort Also known as a concurrent cohort or study longitudinal study Retrospective Historical cohort study or a cohort study nonconcurrent prospective study 7 PROSPECTIVE COHORT STUDY DESIGN Example: You are here As a researcher, you define the population of elementary school students to be followed-up prospectively (2012 as baseline). Identification of smokers and non-smokers after 10 years (2022). Development of lung cancer observed over an additional 20 years (2042). Gordis Epidemiology 8  RETROSPECTIVE COHORT STUDY DESIGN Population of interest was defined before; e.g., an old roster of elementary schoolchildren from 1982 (1982 as baseline) Identification of smokers and non-smokers after 10 years (2002). Development of lung cancer observed over an additional 20 years (2012). As a researcher, you start securitizing the already available data in 2012 and structure the data as a cohort study You are here Gordis Epidemiology 9 PROSPECTIVE COHORT VS. RETROSPECTIVE COHORT The key difference is calendar time Prospective Cohort Design: Exposure status identified as they occur during the study. Groups followed into the future for several years to measure incidence. Retrospective Cohort Design: Exposure ascertained from past records. Outcome determined at the beginning of the study. Possible to combine both designs Gordis Epidemiology 10  PROSPECTIVE VS. RETROSPECTIVE https://methods.sagepub.com/book/mono/public-health-research-methods/chpt/7-cohort-casecontrol-studies1 11 11 CASE-CONTROL VS. COHORT STUDIES 12 12  FIXED AND DYNAMIC POPULATIONS Longitudinal studies may use a fixed or dynamic population Fixed (closed) population: all participants start the study at the same time and no additional participants are added after the study’s start date. Dynamic (open) population: with rolling enrollment that allows new participants to be recruited after the study team begins collecting data 13 13 FIXED AND DYNAMIC POPULATIONS ID Enrolled on Follow-up period Notes P1 Jan 1, 2024 Jan 1, 2024 – Dec 31, 2024 Completed study Fixed (closed) population P2 Jan 1, 2024 Jan 1, 2024 – Dec 31, 2024 Completed study P3 Jan 1, 2024 Jan 1, 2024 – Jul 1, 2024 Dropped out after 6 months P4 Jan 1, 2024 Jan 1, 2025 – Sep 1, 2024 Dropped out after 8 months P5 Jan 1, 2024 Jan 1, 2024 – Dec 1, 2024 Dropped out after 11 months ID Enrolled in Follow-up period Notes P1 Jan 1, 2024 Jan 1, 2024 – Dec 31, 2024 Completed study Dynamic (open) P2 Feb 1, 2024 Feb 1, 2024 – Dec 31, 2024 Completed study population P3 Jun 1, 2024 Jun 1, 2024 – Dec 1, 2024 Dropped out after 4 months P4 Jul 1, 2024 Jul 1, 2024 – Sep 1, 2024 Dropped out after 2 months P5 Sep 1, 2024 Sep 1, 2024 – Dec 31, 2024 Completed study 14 14  LOSS TO FOLLOW-UP Loss to follow-up is the inability to continue tracking a participant in a prospective or longitudinal study because the person drops out, relocates, dies, or stops responding to study communication for another reason. Loss to follow-up is a major concern of studies that follow participants forward in time Strategies must be developed to minimize the burden of participation while maximizing interest in continuing to participate. Some studies may increase retention rates by offering participants free medical tests or other incentives in addition to regularly sharing study findings with members of the cohort so that they can see how their contributions are advancing scientific knowledge. 15 15 LOSS TO FOLLOW-UP EXAMPLE Scenario: Tracking Smoking Habits Over 5 Years Study Objective: Examine changes in smoking habits among participants over 5 years. Participants Enrolled: 5 individuals (open cohort) ID Enrolled on Follow-up period Follow-up duration Status P1 Jan 1, 2020 Jan 1, 2020 – Jan 1, 2025 Full 5 years Completed all follow-ups P2 Feb 1, 2020 Feb 1, 2020 – Feb 1, 2023 3 years Lost to follow-up (moved another city) P3 Jun 1, 2020 Jun 1, 2020 – Jun 1, 2024 4 years Lost to follow-up (due to health reasons) P4 Jul 1, 2020 Jul 1, 2020 – Jul 1, 2022 2 years Dropped out (No longer interested in study) P5 Sep 1, 2020 Sep 1, 2020 – Sep 1, 2021 1 year Lost to follow-up (due to death) 16 16  DROPOUT VS LOSS TO FOLLOW-UP Dropout A participant actively withdraws from the study. Participants may explicitly state they no longer wish to participate, due to reasons like lack of interest, time constraints, or personal objections. Example: Explicitly decided to leave the study due to disinterest. Loss to Follow-up A participant stops responding or is unreachable for follow-up assessments, often without clear communication. Due to moving, health issues, or becoming unavailable without informing the study team. Example: Due to health/death reasons (but did not formally withdraw) 17 17 2*2 TABLES IN COHORT STUDIES Disease status Disease+ Disease- Total Exposed A B N1 = A+B Exposure Unexposed C D N0 = C+D status Total A+C B+D N Exposed individuals who develop the disease (A) Exposed individuals who won’t develop the disease (B) Unexposed individuals who develop the disease (C) Unexposed individuals who won’t develop the disease (D) 18 18  “RISK RATIO” IS COHORT STUDIES Exposure Disease status Risk and Risk Ratio (RR) groups Disease+ Disease- Total Risk of Risk Ratio disease Exposed A B N1 = A+B A / N1 𝐴/(𝐴 + 𝐵) = 𝐶/(𝐶 + 𝐷) Unexposed C D N0 = C+D C / N0 --- 𝐴 𝑟𝑖𝑠𝑘 𝑜𝑓 𝑑𝑖𝑠𝑒𝑎𝑠𝑒 𝑎𝑚𝑜𝑛𝑔 𝑒𝑥𝑝𝑜𝑠𝑒𝑑 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙𝑠 𝑅𝑖𝑠𝑘 𝑟𝑎𝑡𝑖𝑜 𝑅𝑅 = = 𝐴+𝐵 𝑟𝑖𝑠𝑘 𝑜𝑓 𝑑𝑖𝑠𝑒𝑎𝑠𝑒 𝑎𝑚𝑜𝑛𝑔 𝑢𝑛𝑒𝑥𝑝𝑜𝑠𝑒𝑑 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙𝑠 𝐶 𝐶+𝐷 19 19 COHORT STUDIES Study Design A group of 1000 individuals who are regular smokers (exposed) 1000 individuals who have never smoked (unexposed). Follow-up Period: Follow both cohorts for 10 years to observe the development of lung cancer. Exposure Disease status Risk and Risk Ratio (RR) groups Lung ca+ Lung ca- Total Risk of disease Risk Ratio Non-smokers 10 990 1000 ? --- Smokers 50 950 1000 ? ? 20 20  RELATIVE RISK (RISK RATIO) RR = 1 There is no difference in the risk of the outcome between the exposed and unexposed cohorts. The exposure is not associated with any change in the risk of the outcome This is often referred to as the null value or the reference value for the relative risk. RR > 1 The exposed cohort has a higher risk of developing the outcome compared to the unexposed cohort. Suggesting a positive association between the exposure and the outcome. E.g., RR = 1.5: the exposed cohort has a 1.5 times higher risk of the outcome compared to the unexposed cohort. RR < 1 The exposed cohort has a lower risk of developing the outcome compared to the unexposed cohort. The exposure is associated with a reduced risk of the outcome (possibly protective) E.g., RR = 0.8, the exposed cohort has an 0.8 times lower risk of the outcome compared to the unexposed cohort Alternatively, the exposed cohort is 20% (i.e., 1 – 0.8 = 0.2) less likely to develop the disease vs. unexposed cohort 21 21 RR, 95% CI, STATISTICAL SIGNIFICANCE 22 22  INTERPRETATION OF RR BASED ON 95% CI When the entire 95% CI is less than 1, the RR is statistically significant The exposure is deemed to be protective in the study population. When the entire 95% CI is greater than 1, the RR is statistically significant The exposure is deemed to be risky in the study population. When the 95% CI overlaps RR = 1, the RR is NOT statistically significant The lower end of the CI is less than 1, suggesting protection, while the higher end of the CI is greater than 1, suggesting risk. In this situation, the exposure and disease are deemed to have no statistical association in the study population. 23 23 CASE-CONTROL AND COHORT STUDIES: ADVANTAGES Case-Control Less expensive (vs. cohort studies and RCTs) Appropriate for diseases with long induction Effective when the disease is rare Allows simultaneous evaluation of multiple exposures Temporal sequence can be established Cohort Provide clearer indication of the temporal sequence Incidence of outcome can be measured Valuable for investigating the effects of rare exposures Allows studying multiple diseases associated with each exposure 24 24  CASE-CONTROL AND COHORT STUDIES: LIMITATIONS Case-Control Inefficient for rare exposures, which might lead to small samples. Potential for inaccurate exposure information (recall bias as an information bias) Incidence is hard to be directly estimated Focus on a single outcome (limiting examination of multiple outcomes) Cohort Expensive and time-consuming Inefficient for rare diseases (or diseases with long latency) Loss to follow up is involved, thus, cohort attrition is possible Changes in exposure and outcome definitions over time 25 25 APPENDIX 26 26  COHORT STUDIES Exposure Disease status Risk and Risk Ratio (RR) groups Lung ca+ Lung ca- Total Risk of disease Risk Ratio Non-smokers 10 990 1000 0.01 (1%) --- Smokers 50 950 1000 0.05 (5%) 0.05 / 0.01 = 5 Interpretation The risk ratio of developing lung cancer among smokers compared to non-smokers is 5. This indicates that smokers have a five times higher risk of developing lung cancer compared to individuals who have never smoked. 27 27 RELATIVE RISK (RISK RATIO) https://www.youtube.com/watch?v=felIAwyaGFM 28