Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Epidemiological Studies


Published on

Prepared for HSCI 330, SFU

Published in: Education
  • What are the best natural remedies for hair loss? ▲▲▲
    Are you sure you want to  Yes  No
    Your message goes here
  • Be the first to like this

Epidemiological Studies

  1. 1. STUDY DESIGNSKiffer G. Card | Health Sciences 330 | Simon Fraser University
  2. 2. OUTLINE • Midterm • Proposals • Study Designs
  3. 3. MIDTERM EXAM Find Study Buddies!
  4. 4. MIDTERM EXAM • Canvas grading didn’t seem to work correctly so Rashed did the regrading manually. • Talk to Rashed if there is any disagreement between the rubric and the points you received. • Now lets discuss commonly missed questions.
  5. 5. MIDTERM EXAM • Question 1 – Wedding cake is correct because sushi had poor explanatory power (i.e., few people who got sick ate sushi). • 45% correctly chose “wedding cake” • 36% incorrectly chose “sushi” • Question 2 – Vaccines confer immunity by instigating the adaptive immune response, not the innate response. • 72% incorrectly chose True • 28% correctly chose False • Question 5 – Mosquitoes are an effective vector, not vehicle. • 48% incorrectly chose true. • 51% correctly chose false.
  6. 6. MIDTERM EXAM • Question 7 – Predisposing and enabling factors occur before a behaviour, reinforcing factors occur after. • 62% correctly chose false • 37% incorrectly chose true (though we also gave points for this). • Question 10 – Secondary attack rate • 34% correctly answered 400 (i.e., 4/10; classmates and susceptible siblings) • 27% answered 13 (i.e., 4/303; classmates, classmates susceptible siblings, and susceptible siblings) • Definitely should not include the susceptible siblings of non-infected. • 13% answered 37 (i.e., 4/107; classmates and susceptible siblings) • “Wave 1” Infections occurred just before winter break (i.e., no more contact with classmates); • Chickenpox usually isn’t’ spread through brief contact. • Classmates are part of “wave 1” -- they were vulnerable from patient zero just like the 6. • We also accepted this response as it may not have been clear that the classmates had contact.
  7. 7. MIDTERM EXAM • Question 11 – Kissing and aerosol spread are both direct transmission. • 52% correctly chose “two of the above” • 37% incorrectly chose only the kissing mono. • Question 12 – Food-related botulism is infectious and common source. • 47% incorrectly chose only common source • 36% correctly chose “two of the above” • Question 13 – Droplet spread is Indirect, Vehicle transmission • 53% correctly chose indirect, vehicle spread • 34% incorrectly chose direct vehicle spread • Direct transmission does not use vehicles or vectors.
  8. 8. MIDTERM EXAM • Question 14 – communicable disease are transmitted between people. • 56% correctly chose “communicable” • 42% chose “two of the above” which we also accepted since communicable diseases are infectious also. • Question 15 – Person time incidence rate • 48% gave answers in the correct range between 19 and 21. • 100 patients followed for three years. • In year 1, 0 relapsed and 8 were lost. • In year 2, 2 died, 2 relapsed and 10 were lost. • In year 3, 2 died, 3 relapsed, and 13 were lost. • That’s 5 relapses (numerator). • 92 Observed in Year 1 for a full year (92) • 8 observed in year 1 for a half year (4) • 78 observed in year 2 for a full year (78) • 14 observed in year 2 for a half year (7) • 60 observed in year 3 for a full year (60) • 18 observed in year 3 for a half year (9) • 5/(92+4+78+7+60+9) = .02 x 1000 = 20
  9. 9. MIDTERM EXAM • Question 16 – Shirt as fomite or vehicle • 75% correctly chose fomite or vehicle • 10% chose reservoir • Reservoirs are usually a more general long-term source. • Question 17 – Presymptomatic stage is also called incubation • 70% correctly chose “incubation period” • 12% chose subclinical • Subclinical is more than presymptomatic – it means that the disease cant be detected using clinical screens • We also accepted subclinical.
  10. 10. MIDTERM EXAM • Question 20 – not a threshold effect because the effect is negative up until 3 drinks. • 28% correctly identified false. • 72% incorrectly identified true. • Question 24a – We accepted active, incubatory, or asymptomatic for a person with a detectable viral load because no information about symptoms were provided (most cases of HIV don’t have symptoms).
  11. 11. MIDTERM EXAM • Question 25a – agreement between mouse and human studies shows consistency • 38% chose consistency; 14% chose biological plausibility; 25% chose analogy • We also accepted biological plausibility. • Question 25b – All AIDS cases having HIV demonstrates specificity • 44% chose specificity; 24% chose consistency; 14% chose temporality; 15% chose strength. • Question 25c – Arsenic levels in hair and nails being associated with health problems demonstrates temporality. • 49% chose temporality; 18% chose strength; 11% chose biological plausibility. • Question 25e – Asbestos fibre toxicity shoes analogy for toxicity for carbon nano-tubes. • 35% chose analogy; 37% chose empiricism; 13% chose biological plausibility. • Question 25f – Molecular mechanisms for lead poisoning shows biological plausibility • 54% chose biological plausibility; 25% chose biological gradient
  12. 12. MIDTERM EXAM • Question 27a – 44 cases of salmonella with no identified external source indicates a propagated epidemic • 32% chose propagated (asymptomatic carrier was then the source) • 39% chose point source • 25% chose mixed epidemic • Question 27b – 38 cases of giardia over six weeks traced back to a coffee maker that is intermittently used indicates an intermittent source. • 64% chose intermittent common source • 20% chose mixed • Question 27c – 180 cases of norovirus tracked to ice indicates a point source. • 69% chose point source; 16% chose mixed
  13. 13. MIDTERM EXAM • Question 31 – Contamination traced back to dinner on the 20th for a disease with a 33 hour incubation period. • 29% chose dinner on the 20th; 59% chose breakfast on the 20th ≈33 hours ≈44 hours
  14. 14. MIDTERM EXAM • Any other questions?
  15. 15. • Last week you should have worked with your group through the “Getting Your Grant Proposal Ready” document. • Talk with your group and identify if you have any questions about the research proposal project that I can answer now. • These questions should not really be about “Is this a good research question/aim/objective” as only you – the content experts familiar with the state of the literature – can really say at this point. I will judge the research question based on how well your introduction rationalizes it within the context of recent research on the topic. • I have now posted the “Group Evaluation Form” which each of you will need to complete and submit along with your proposal. DEVELOPMENT OF RESEARCH PROPOSAL
  16. 16. SAMPLING • Population parameter. A population parameter is the true value of a population attribute. • Sample statistic. A sample statistic is an estimate, based on sample data, of a population parameter.
  17. 17. EQUAL PROBABILITY OF SELECTION • In a random sample, all members of a target population have an equal probability of selection; are selected by chance alone; and when selected they participate in the study. • Random samples have greater accuracy, precision, and representativeness. • There are many non- random sampling methods and variety of strategies used to generate random samples.
  18. 18. Expert Opinion Ecological Studies Case Studies Cross-Sectional Surveys Case Control Studies Cohort Studies Randomized Controlled Trials Meta-Analyses & Reviews Epidemiological Studies StrengthofCausalEvidence
  19. 19. In groups of two or three, use the SFU Library ( to identify a study that uses each of the following study designs. • Cross-sectional • Case-Control • Prospective Cohort • Retrospective Cohort • Nested Case-Control Based on the studies you identified try to craft a short description of what each study type entails. ACTIVITY • Case-Cohort • Double Cohort • Case-Crossover • Ecological Study
  20. 20. • A case report involves a profile of a single individual • A case series involves a small group of patients with a similar diagnosis • Provide evidence for larger scale studies (hypothesis generating) CASE-REPORT & CASE-SERIES
  21. 21. CROSS-SECTIONAL STUDIES • All variables are measured at the same time without distinction between risk factors and outcomes.
  22. 22. Calculate the prevalence of red, orange, and green fruit pieces? ACTIVITY
  23. 23. CROSS-SECTIONAL EXAMPLE • Background Atherosclerosis develops from early childhood; physical activity could positively affect this process. This study’s aim was to assess the associations of objectively measured physical activity with clustering of cardiovascular disease risk factors in children and derive guidelines on the basis of this analysis. • Methods We recruited 1732 randomly selected 9-year-old and 15-year-old school children from Denmark, Estonia, and Portugal. Risk factors included in the composite risk factor score (mean of Z scores) were systolic blood pressure, triglyceride, total cholesterol/HDL ratio, insulin resistance, sum of four skinfolds, and aerobic fitness. Individuals with a risk score above 1 SD of the composite variable were defined as being at risk. Physical activity was assessed by accelerometry. • Findings Odds ratios for having clustered risk for ascending quintiles of physical activity (counts per min; cpm) were 3·29 (95% CI 1·96–5·52), 3·13 (1·87–5·25), 2·51 (1·47–4·26), and 2·03 (1·18–3·50), respectively, compared with the most active quintile. The fi rst to the third quintile of physical activity had a raised risk in all analyses. The mean time spent above 2000 cpm in the fourth quintile was 116 min per day in 9-year-old and 88 min per day in 15-year-old children. • Interpretation Physical activity levels should be higher than the current international guidelines of at least 1 h per day of physical activity of at least moderate intensity to prevent clustering of cardiovascular disease risk factors Anderson et al. “Physical activity and clustered cardiovascular risk in children: a cross-sectional study”
  24. 24. PREVALENCE RATIO Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝑃𝑅 = 𝑎/(𝑎 + 𝑏) 𝑐/(𝑐 + 𝑑) = 𝑃𝑟𝑒𝑣𝑎𝑙𝑒𝑛𝑐𝑒 𝐴𝑚𝑜𝑛𝑔 𝐸𝑥𝑝𝑜𝑠𝑒𝑑 𝑃𝑟𝑒𝑣𝑎𝑙𝑒𝑛𝑐𝑒 𝐴𝑚𝑜𝑛𝑔 𝑁𝑜𝑛𝐸𝑥𝑝𝑜𝑠𝑒𝑑
  25. 25. On Bottom (Has Factor 2) Not on Bottom (Does not have Factor 2) Green (Has Factor 1) (a) 3 (b) 5 Not Green (Does not have Factor 1) (c) 1 (d) 21 Calculate the prevalence ratio for “being green” and “being on the bottom” ACTIVITY
  26. 26. CROSS-SECTIONAL STUDY Strengths Limitations • Produces Prevalence Data • Cost and Time Efficient • Control over study population and measurements. • Several associations can be examined at the same time. • No temporal ordering of factors. • Doesn't yield relative risk or incidence. • Not feasible for rare exposures. • Potential bias from low response rate. • Potential measurement bias. • Higher proportion of long term survivors.
  27. 27. SERIAL OR PERIODIC CROSS-SECTIONAL SURVEYS • Cross-sectional surveys that are conducted on a routine basis. • Each administration is a new sample.
  28. 28. SERIAL-CROSS SECTIONAL EXAMPLE • Objective: To examine the changes over time in cardiorespiratory fitness and body mass index (BMI) of children. • Setting: Primary schools in Liverpool, UK. • Participants: A total of 15 621 children (50% boys), representing 74% of eligible 9–11- year olds in the annual school cohorts between 1998/9 and 2003/4, who took part in a 20 multi-stage shuttle run test (20mMST). • Main outcome measures: Weight, height, BMI (kg/m2 ) and obesity using the International Obesity Taskforce definition. • Results: Median (95% confidence interval) 20mMST score (number of runs) fell in boys from 48.9 (47.9–49.9) in 1998/9 to 38.1 (36.8–39.4) in 2003/4, and in girls from 35.8 (35.0–36.6) to 28.1 (27.2–29.1) over the same period. Fitness scores fell across all strata of BMI (Po0.001). Moreover, BMI increased over the same 6-year period even among children in fittest third of 20mMST. • Conclusion: In a series of uniform cross-sectional assessments of school-aged children, BMI increased whereas cardiorespiratory fitness levels decreased within a 6-year period. Even among lean children, fitness scores decreased. Public health measures to reduce obesity, such as increasing physical activity, may help raise fitness levels among all children – not just the overweight or obese. Stratton et al. “Cardiorespiratory fitness and body mass index of 9–11-year-old English children: a serial cross-sectional study from 1998 to 2004”
  29. 29. ACTIVITY With your group design a study related to your group’s research question that would utilize a serial cross-sectional study design.
  30. 30. • Secular trends represent long-term changes in health-related states or events • Short-term trends are usually brief, unexpected increases in health-related states or events • Cyclic trends represent periodic increases and decreases in the occurrence of health-related states or events TYPES OF TRENDS
  31. 31. • Age – An effect that emerges due to the aging of a population. • Period – An effect that occurs due to the changes in a particular point in time. • Cohort – An effect that is present due to differences in birth cohorts. CAUSES OF TRENDS
  32. 32. • Some explanations for observed changes in the frequency and pattern of cases in a surveillance system, which are not due to changes in risk exposures, include • Inconsistent interpretation and application of the case definition • Change in the case definition • Change in surveillance system/policy of reporting • Improved diagnosis (e.g., new laboratory test, increased physician awareness, a new physician in town) • Change in diagnostic criteria • Change in reporting requirements • Change in the population • Change in the level and emphasis on active case detection • Random events • Increased public awareness OTHER EXPLANATIONS FOR TRENDS
  33. 33. CASE-CONTROL STUDY Presence of risk factors for cases is compared with the presence of risk factors for controls. Cases Controls Exposed Unexposed Exposed Unexposed
  34. 34. CASE-CONTROL EXAMPLE • Background: Although more than 80% of the global burden of cardiovascular disease occurs in low-income and middle-income countries, knowledge of the importance of risk factors is largely derived from developed countries. • Methods: We established a study of acute myocardial infarction in 52 countries, representing every inhabited continent. 15,152 cases and 14,820 controls were enrolled. The relation of smoking, history of hypertension or diabetes, waist/hip ratio, dietary patterns, physical activity, consumption of alcohol, blood apolipoproteins (Apo), and psychosocial factors to myocardial infarction are reported here. Odds ratios and their 99% CIs for the association of risk factors to myocardial infarction and their population attributable risks (PAR) were calculated. • Findings: Smoking (odds ratio 2·87 for current vs never, PAR 35·7% for current and former vs never), raised ApoB/ApoA1 ratio (3·25 for top vs lowest quintile, PAR 49·2% for top four quintiles vs lowest quintile), history of hypertension (1·91, PAR 17·9%), diabetes (2·37, PAR 9·9%), abdominal obesity (1·12 for top vs lowest tertile and 1·62 for middle vs lowest tertile, PAR 20·1% for top two tertiles vs lowest tertile), psychosocial factors (2·67, PAR 32·5%), daily consumption of fruits and vegetables (0·70, PAR 13·7% for lack of daily consumption), regular alcohol consumption (0·91, PAR 6·7%), and regular physical activity (0·86, PAR 12·2%), were all significantly related to acute myocardial infarction (p<0·0001 for all risk factors and p=0·03 for alcohol). These associations were noted in men and women, old and young, and in all regions of the world. Collectively, these nine risk factors accounted for 90% of the PAR in men and 94% in women. • Interpretation: Abnormal lipids, smoking, hypertension, diabetes, abdominal obesity, psychosocial factors, consumption of fruits, vegetables, and alcohol, and regular physical activity account for most of the risk of myocardial infarction worldwide in both sexes and at all ages in all regions. This finding suggests that approaches to prevention can be based on similar principles worldwide and have the potential to prevent most premature cases of myocardial infarction. Yusuf et al. “Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study”
  35. 35. ODDS RATIO Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝑂𝑅 = 𝑎/𝑐 𝑏/𝑑 = 𝑎 ∗ 𝑑 𝑏 ∗ 𝑐
  36. 36. ATTRIBUTABLE RISK IN EXPOSED Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝐴𝑅 = 𝑎 𝑎 + 𝑏 − 𝑐 𝑐 + 𝑑 = 𝑟𝑖𝑠𝑘 𝑖𝑛 𝑒𝑥𝑝𝑜𝑠𝑒𝑑 − 𝑟𝑖𝑠𝑘 𝑖𝑛 𝑛𝑜𝑛 − 𝑒𝑥𝑝𝑜𝑠𝑒𝑑
  37. 37. ATTRIBUTABLE RISK PERCENT IN EXPOSED Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝐴𝑅% = 𝑎 𝑎 + 𝑏 − 𝑐 𝑐 + 𝑑 𝑎 𝑎 + 𝑏 ∗ 100
  38. 38. POPULATION ATTRIBUTABLE RISK Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝑃𝐴𝑅 = 𝑎 + 𝑐 𝑎 + 𝑏 + 𝑐 + 𝑑 − 𝑐 𝑐 + 𝑑
  39. 39. POPULATION ATTRIBUTABLE RISK PERCENT Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝑃𝐴𝑅% = 𝑎 + 𝑐 𝑎 + 𝑏 + 𝑐 + 𝑑 − 𝑐 𝑐 + 𝑑 𝑎 + 𝑐 𝑎 + 𝑏 + 𝑐 + 𝑑 ∗ 100
  40. 40. CASE-CONTROL STUDY Strengths Limitations • Effective for rare outcomes. • Provides partial time sequencings. • Relatively less expensive than cohorts. • Yields odds ratios (which are a good estimate of relative risk when an outcome is rare). • Limited to one outcome. • No incidence, relative risk, or natural history (i.e., latency and induction). • Potential selection/survival bias. • Potential recall and interview bias. • Vulnerable to misclassification. • Does not yield incidence or prevalence.
  41. 41. ACTIVITY With your group design a study related to your group’s research question that would utilize an case-control study design.
  42. 42. PROSPECTIVE COHORT STUDY Exposed Unexposed Time Outcome Follow-up over TimeRecruitment Data Analysis Outcome • Participants are selected based on exposure status and followed across time to monitor outcome events.
  43. 43. PROSPECTIVE COHORT EXAMPLE • Objective: To determine the independent risk factors for atrial fibrillation. Design.p=m-Cohort study. • Subjects: A total of 2090 men and 2641 women members, free of a history of atrial fibrillation, between the ages of 55 and 94 years. • Main Outcome Measures: Sex-specific multiple logistic regression models to identify independent risk factors for atrial fibrillation, including age, smoking, diabetes, electrocardiographic left ventricular hypertrophy, hypertension, myocardial infarction, congestive heart failure, and valve disease. • Results: During up to 38 years of follow-up, 264 men and 298 women developed atrial fibrillation. After adjusting for age and other risk factors for atrial fibrillation, men had a 1.5 times greater risk of developing atrial fibrillation than women. In the full multivariable model, the odds ratio (OR) of atrial fibrillation for each decade of advancing age was 2.1 for men and 2.2 for women (P<.0001). In addition, after multivariable adjustment, diabetes (OR, 1.4 for men and 1.6 for women), hypertension (OR, 1.5 for men and 1.4 for women), congestive heart failure (OR, 4.5 for men and 5.9 for women), and valve disease (OR, 1.8 for men and 3.4 for women) were significantly associated with risk for atrial fibrillation in both sexes. • Conclusion: In addition to intrinsic cardiac causes such as valve disease and congestive heart failure, risk factors for cardiovascular disease also predispose to atrial fibrillation. Modification of risk factors for cardiovascular disease may have the added benefit of diminishing the incidence of atrial fibrillation Benjamin et al. “Independent Risk Factors for Atrial Fibrillation in a Population-Based Cohort”
  44. 44. RISK RATIO Outcome (Myocardial Infarction) No Outcome (No Myocardial Infarction) Exposed (Sedentary) (a) 89 (b) 468 Unexposed (Physically Active) (c) 70 (d) 487 0 2 4 6 8 10 12 14 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 NumberofParticipantswithMyocardial Infarction Risk Ratio = 𝑎 𝑎 + 𝑏 𝑐 𝑐 + 𝑑 = 89 89 + 468 70 70 + 487 = 89 557 70 557 = 0.15978 0.12567 = 1.27 Conclusion: Relative to physically active participants, sedentary participants have 1.27 time higher risk of heart attack. Numbers are simulated, and do not reflect actual data.
  45. 45. n3 PREVALENCE VS. INCIDENCE n4 n5 n8 n7 n6 n9 n12 n10 Time n1 n2 n11 ExposedUnexposed “Diseased” Died “Healthy” Cross-Section Key Outcome No Outcome Exposed 1 1 Unexposed 4 2 Outcome No Outcome Exposed 5 1 Unexposed 4 2 Cross-Sectional Cohort Cohort
  46. 46. RATE RATIO Case Control Was Exposed (a) (b) Was Not Exposed (c) (d) 𝑅𝑎𝑡𝑒 𝑅𝑎𝑡𝑖𝑜 = 𝑎 𝑝𝑒𝑟𝑠𝑜𝑛 𝑡𝑖𝑚𝑒 𝑜𝑓 𝑑𝑖𝑠𝑒𝑎𝑠𝑒 𝑓𝑟𝑒𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑢𝑝 𝑐 𝑝𝑒𝑟𝑠𝑜𝑛 𝑡𝑖𝑚𝑒 𝑜𝑓 𝑑𝑖𝑠𝑒𝑎𝑠𝑒 𝑓𝑟𝑒𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑢𝑝
  47. 47. PROSPECTIVE COHORT STUDY Strengths Limitations • Establishes temporal sequencing of events • Avoids prevalence-incidence and Berkson’s bias. • Several outcomes can be assessed from shared exposures. • Better handles natural history of disease. • Yields, incidence, relative risk, and attributable risk. • Often requires large sample sizes to detect sufficient disease. • Expensive and time consuming. • Not feasible for rare outcomes. • Limited to only 1 explanatory factor. • Bias due to loss to follow-up.
  48. 48. ACTIVITY With your group design a study related to your group’s research question that would utilize an prospective cohort study design.
  49. 49. RETROSPECTIVE COHORT Determine Incidence Exposed Time Outcome Identify Patients with Exposure OutcomeUnexposed Look through patient records Find Data Source A cohort study conducted using data or records that were already collected.
  50. 50. RETROSPECTIVE COHORT EXAMPLE • Background Although CT scans are very useful clinically, potential cancer risks exist from associated ionising radiation, in particular for children who are more radiosensitive than adults. We aimed to assess the excess risk of leukaemia and brain tumours after CT scans in a cohort of children and young adults. • Methods In our study, we included patients without previous cancer diagnoses who were first examined with CT in National Health Service (NHS) centres in England, Wales, or Scotland (Great Britain) between 1985 and 2002, when they were younger than 22 years of age. We obtained data for cancer incidence, mortality, and loss to follow-up from the NHS Central Registry from Jan 1, 1985, to Dec 31, 2008. We estimated absorbed brain and red bone marrow doses per CT scan in mGy and assessed excess incidence of leukaemia and brain tumours cancer with Poisson relative risk models. To avoid inclusion of CT scans related to cancer diagnosis, follow-up for leukaemia began 2 years after the first CT and for brain tumours 5 years after the first CT. • Findings During follow-up, 74 of 178 604 patients were diagnosed with leukaemia and 135 of 176 587 patients were diagnosed with brain tumours. We noted a positive association between radiation dose from CT scans and leukaemia (excess relative risk [ERR] per mGy 0·036, 95% CI 0·005–0·120; p=0·0097) and brain tumours (0·023, 0·010–0·049; p<0·0001). Compared with patients who received a dose of less than 5 mGy, the relative risk of leukaemia for patients who received a cumulative dose of at least 30 mGy (mean dose 51·13 mGy) was 3·18 (95% CI 1·46–6·94) and the relative risk of brain cancer for patients who received a cumulative dose of 50–74 mGy (mean dose 60·42 mGy) was 2·82 (1·33–6·03). • Interpretation Use of CT scans in children to deliver cumulative doses of about 50 mGy might almost triple the risk of leukaemia and doses of about 60 mGy might triple the risk of brain cancer. Because these cancers are relatively rare, the cumulative absolute risks are small: in the 10 years after the first scan for patients younger than 10 years, one excess case of leukaemia and one excess case of brain tumour per 10 000 head CT scans is estimated to occur. Pearce et al. “Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study”
  51. 51. RETROSPECTIVE COHORT STUDY Strengths Limitations • Cheaper and faster than a traditional cohort • Less control over data and collections methods.
  52. 52. ACTIVITY With your group design a study related to your group’s research question that would utilize an retrospective cohort study design.
  53. 53. NESTED CASE-CONTROL STUDY Exposed to Exposure 1 Not Exposed to Exposure 1 Outcome Outcome n1 n2 n3 n4 n5 n6 n7 n8 n9 Cases (i.e., Developed Outcome) Had Exposure 2 Didn’t Have Exposure 2Controls (i.e., No Outcome) A case-control study within a cohort study, often looking at a different exposure than the one recruited on.
  54. 54. NESTED CASE-CONTROL EXAMPLE • Background. Controversy has surrounded the question about whether high-dose rofecoxib increases or naproxen decreases the risk of serious coronary heart disease. We sought to establish if risk was enhanced with rofecoxib at either high or standard doses compared with remote non-steroidal anti-inflammatory drug (NSAID) use or celecoxib use, because celecoxib was the most common alternative to rofecoxib. • Methods. We used data from Kaiser Permanente in California to assemble a cohort of all patients age 18–84 years treated with a NSAID between Jan 1, 1999, and Dec 31, 2001. Cases of serious coronary heart disease (acute myocardial infarction and sudden cardiac death) were risk-set matched with four controls for age, sex, and health plan region. Current exposure to cyclo- oxygenase 2 selective and non-selective NSAIDs was compared with remote exposure to any NSAID, and rofecoxib was compared with celecoxib. • Findings. During 2 302 029 person-years of follow-up, 8143 cases of serious coronary heart disease occurred, of which 2210 (27·1%) were fatal. Multivariate adjusted odds ratios versus celecoxib were: for rofecoxib (all doses), 1·59 (95% CI 1·10–2·32, p=0·015); for rofecoxib 25 mg/day or less, 1·47 (0·99–2·17, p=0·054); and for rofecoxib greater than 25 mg/day, 3·58 (1·27– 10·11, p=0·016). For naproxen versus remote NSAID use the adjusted odds ratio was 1·14 (1·00– 1·30, p=0·05). • Interpretation. Rofecoxib use increases the risk of serious coronary heart disease compared with celecoxib use. Naproxen use does not protect against serious coronary heart disease. Grahamet al. “Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs”
  55. 55. NESTED CASE-CONTROL STUDY Strengths Limitations • Takes advantage of existing study framework to conduct study cheaply. • Control subjects may not be representative of original cohort due to death or loss to follow- up.
  56. 56. ACTIVITY With your group design a study related to your group’s research question that would utilize an nested case-control study design.
  57. 57. Subcohort (Controls) CASE-COHORT STUDY Exposed to Exposure 1 Not Exposed to Exposure 1 Outcome Outcome Random Selection into Subcohort Cohort Eligible for Case-Cohort Cases Cases in Subcohort Exposed? Exposed? A comparison of cases to a randomly selected subset of controls within the context of a cohort study.
  58. 58. CASE-COHORT EXAMPLE • Background Amyloid β peptides (Aβ) are important components of plaques in Alzheimer's disease. Plasma concentrations of Aβ1–40 and Aβ1–42 rise with age and are increased in people with mutations that cause early-onset Alzheimer's disease. However, Aβ1–42 concentrations may decrease early in the dementia process. We postulated that concentrations of Aβ1–40 and Aβ1– 42 in plasma are associated with risk of dementia. • Methods Embedded within the population-based Rotterdam Study (n = 6713 at risk for dementia), a random sample of 1,756 people was drawn. During follow-up (mean 8·6 years), 392 incident dementia cases were identified. We investigated the association between plasma Aβ concentrations and risk of dementia and its subtypes using Cox proportional hazard models. • Findings High concentrations of Aβ1–40 but not Aβ1–42 at baseline were associated with an increased risk of dementia. Compared with the first quartile of Aβ1–40, age and sex-adjusted hazard ratios for dementia for the second, third, and fourth quartiles were 1·07 (95% CI 0·72– 1·58), 1·16 (0·78–1·70), and 1·46 (1·01–2·12). People with an increased Aβ1–42/Aβ1–40 ratio had a reduced risk of dementia. Compared with the first quartile of the Aβ1–42/Aβ1–40 ratio, hazard ratios for the second, third, and fourth quartiles were 0·74 (0·53–1·02), 0·62 (0·44–0·88), and 0·47 (0·33–0·67). Associations were similar for Alzheimer's disease and vascular dementia. • Interpretation High plasma concentrations of Aβ1–40, especially when combined with low concentrations of Aβ1–42, indicate an increased risk of dementia. A potential role of plasma Aβ concentrations as a marker of incipient dementia warrants further investigation. Van Oijen et al. “Plasma Aβ1–40 and Aβ1–42 and the risk of dementia: a prospective case-cohort study”
  59. 59. CASE-COHORT STUDY Strengths Limitations • Saves resources by only assessing additional exposures among cases and a relatively small sub-cohort. • Can study multiple outcomes. • Reduces selection bias (same population for cases and controls). • More complex data analysis and management of error. • Exposure information collected at different time for cases and Subcohort.
  60. 60. ACTIVITY With your group design a study related to your group’s research question that would utilize an case-cohort study design.
  61. 61. • Distinct from conventional cohort studies in that two distinct populations are involved with different levels of an exposure of interest • Double cohorts are employed when the exposure is rare and a relatively small number of people are affected DOUBLE COHORT STUDY
  62. 62. DOUBLE COHORT EXAMPLE • Giant cell arteritis is the most frequent vasculitis. Cardiovascular events such as cerebrovascular accident or ischemic heart disease may occur in patients with giant cell arteritis. However, their real incidence, as well as their relative risk compared to the general population, remains unknown. To assess in a prospective the incidence of cardiovascular events in giant cell arteritis patients compared to controls, after controlling for cardiovascular risk factors. We included on predefined criteria 432 newly diagnosed patients with giant cell arteritis, each assigned to sex- and age-matched controls randomly selected from the general population. Cardiovascular risk factors (high-blood pressure, diabetes, smoking, hypercholesterolemia and preexisting peripheral vascular disease) were collected at inclusion. During the 24-month follow-up, all cardiovascular events were collected. After stratification for cardiovascular risk factors, a log-rank test was performed to compare cases and controls. A parametric survival model was used for multivariate analysis. Cardiovascular events all combined were significantly increased in patients with giant cell arteritis (RR = 2.15 [1.21-3.81], P = 0.009), and were mainly associated with age (P = 0.0001), past history of cardiovascular disease (P = 0.023) but also with giant cell arteritis (P = 0.009). However, each subset of cerebrovascular accident (RR = 2.42 [0.84-7]) or ischemic heart disease (RR = 1.67 [0.72- 3.89]) increased but did not significantly. Cardiovascular events incidence is increased in patients with giant cell arteritis, and prescription of preventive antiagregant treatment may be discussed. Page et al. “Incidence of cardiovascular events in giant cell arteritis: preliminary results of a prospective double cohort study”
  63. 63. DOUBLE COHORT Strengths Limitations • Useful when distinct cohorts have different or rare exposures • Must control for time-related factors
  64. 64. CASE-CROSSOVER STUDY Slide 64 n5 n2 n1 n3 n6 n4 Between-Subject Time “Case” “Control” Key Assessment n1 n2 n3 n4 n5 n6 Time Within-Subject Exposure frequency during a window period prior to an event is compared with exposure frequency during a control time at an earlier period.
  65. 65. CASE-CROSSOVER EXAMPLE • OBECTIVE: To evaluate the association between sleep and wakefulness duration and childhood unintentional injury. • DATA COLLECTION: Two hundred ninety-two injured children who presented at the Children’s Emergency Center of Udine, Italy, or their parents were interviewed after a structured questionnaire. Information was collected concerning sociodemographic variables, participant’s habits, and injury characteristics, including a brief description of the accident dynamics. • ASSESSMENT: Sleep or wakefulness status of the child was assessed retrospectively for each of the 48 hours before injury. For each child, we compared the 24 hours immediately before the injury (hours 1–24; case period) with hours 25 to 48 (control period). • ANALYSIS: Nonparametric tests were conducted to compare the difference of sleep duration between case and control periods. In addition, we conducted intrapersonal conditional logistic regression analyses and estimated relative risks (RRs) and 95% confidence intervals (CIs). • CONCLUSIONS: Overall, more children had longer hours of sleep during the control period than during the case period. However, this difference was significant for boys only. A direct association between injury risk and sleeping. Valent et al. “A Case-Crossover Study of Sleep and Childhood Injury”
  66. 66. CASE-CROSSOVER STUDY Strengths Limitations • Controls for individual-level confounders. • Good for studying short-term exposures on the risk for acute events • Potential confounding bias when sampling two populations
  67. 67. ACTIVITY With your group design a study related to your group’s research question that would utilize an case-crossover study design.
  68. 68. • Utilize exposure and outcome data at the population-level rather than individual-level, to draw an association between two factors. • Often used to make international or intranational comparisons. ECOLOGICAL STUDIES
  69. 69. ECOLOGICAL EXAMPLE • Idiopathic autism, suspected to be caused by exposure of genetically susceptible individuals to unknown environmental triggers, has increased dramatically in the past 25 years. The objectives of our study were to determine, using a linear regression model, whether the county prevalence of autism in the Pacific Northwest of the United States was associated with the source of drinking water for that county and whether this relationship was dependent on the level of environmental pollutants and meteorological factors in the county. We found the previously reported relationship between precipitation and autism in a county was dependent on the amount of drinking water derived from surface sources in the county. We also found a positive association between the EPA’s risk of neurological disease and autism, but this relationship was only present in warm areas. Our study provides evidence for the hypothesis that environmental factors are associated with autism and that meteorological factors play a role in this relationship. St-Hilaire et al. “An ecological study on childhood autism”
  70. 70. Countries Individuals within Countries ? ECOLOGICAL STUDIES
  71. 71. ECOLOGICAL STUDIES Strengths Limitations • Inexpensive • Less time consuming • Examines community-, group-, or national- level trends and data. • Subject to ecological fallacy (individual-level data may not be well represented by clustered-data. • Difficult to detect complex etiologies.
  72. 72. ACTIVITY With your group design a study related to your group’s research question that would utilize an ecological study design.