203 Lecture 2 Theme 1 Incidence CB 2023.pptx

Thinking about populations over time: Incidence and cohort studies D R C L I O B E R RY S E N I O R L E C T U R E R I N H E A LT H C A R E E VA LU AT I O N A N D I M P R OV E M E N T C . B E R RY @ BS M S . AC . U K Learning aims To be able to explain the meaning of incidence To understand the cal...

Thinking about populations over time: Incidence and cohort studies D R C L I O B E R RY S E N I O R L E C T U R E R I N H E A LT H C A R E E VA LU AT I O N A N D I M P R OV E M E N T C . B E R RY @ BS M S . AC . U K Learning aims To be able to explain the meaning of incidence To understand the calculation of relative risk To be able to describe a cohort study, its strengths and its limitations To be able to give an example of how a cohort study contributes to our knowledge of health needs covered in the module A reminder of prevalence Prevalence measures the frequency of “cases” of a outcome in a given population at a designated time (the numerator). E.g. diagnosed asthma in children aged 5-11 years. Calculation of prevalence requires a suitable denominator (e.g. GP registered patients, schoolchildren) – the number of people who are ‘at risk’ of the outcome. Prevalence = number of people with outcome / number of people who could have outcome. Prevalence is expressed as a percentage (e.g. 70%), a proportion of 1 (0.7 is equivalent), or a proportion per unit of population (700 of every 1000 people). What is incidence? A measure of the number of new cases of a condition in some given time period – in a month or a year, for example – expressed as a proportion of a population which is at risk Often expressed as per 1000, per 10,000, or even per 1,000,000 people A group of people is followed through time and the onset of a outcome/health event is measured Relationship between prevalence and incidence •Prevalence depends on • the incidence of a outcome and • the time between onset and recovery (or death) •Prevalence = incidence x outcome duration https://youtu.be/1jzZe3ORdd8 Why might we want to know about incidence- diabetes? ◦ Understanding diabetes and its risk factors (exposures) and outcomes ◦ Accurate knowledge of disease, trends, geographical differences  health care providers, researchers and policy makers ◦ Implications for ◦ Individuals- health, happiness ◦ Society – current and future economy, labour workforce ◦ Informing prevention and public health interventions ◦ Service planning and commissioning ◦ Screening and assessment ◦ Staffing, training, resources, specialisms ◦ Identifying and prescribing targeted and indicated interventions ◦ Evaluating effectiveness of interventions Measuring incidence Incidence is the number of instances of disease/outcome case onset, in a given period in a defined population ◦The numerator is the number of new events in a population ◦The denominator is the average number of persons at risk during this period This type of incidence AKA incidence risk, cumulative/ crude incidence Incidence = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛𝑒𝑤 𝑐𝑎𝑠𝑒𝑠 𝑖𝑛 𝑎 𝑔𝑖𝑣𝑒𝑛 𝑡𝑖𝑚𝑒 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑎𝑡 𝑟𝑖𝑠𝑘 Measuring incidence: the numerator As per prevalence = caseness for the disease/outcome *Plus – clear and consistent definition as to what counts as a new case Measuring incidence: the denominator Total population at risk Must be the population truly at risk of developing the disease/condition/outcome As applies to a period of time, typically take the mid-point value, e.g. population at mid-point in a year Calculating incidence New cases of outcome Exposure Total Yes No Yes A B A+B No C D C+D A+C B+D N OUTCOME INCIDENCE •Incidence of outcome in exposed = A / (A + B) •Incidence of outcome in unexposed = C / (C + D) Calculating incidence New cases of outcome Exposure Total Yes No Yes A B A+B No C D C+D A+C B+D N OUTCOME INCIDENCE •Incidence of outcome in exposed = A / (A + B) •Incidence of outcome in unexposed = C / (C + D) Overall incidence = (A + C) / N Calculating incidence All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to 2005) Cardiovascular disease at baseline Total Yes No Yes A B A+B No C D C+D A+C HOW WOULD B + D I CALCULATE N Incidence of death amongst people newly diagnosed with type 2 diabetes? Data from Vamos et al. (2012) Calculating incidence All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to 2005) Cardiovascular disease at baseline Total Yes No Yes A B A+B No C D C+D OUTCOME INCIDENCE A+C •Incidence of outcome in exposed = A / (A + B) •Incidence of outcome in unexposed = C / (C + D) HOW WOULD B + D I CALCULATE N Incidence of death amongst people newly diagnosed with type 2 diabetes with cardiovascular disease? Data from Vamos et al. (2012) Overall incidence = (A + C) / Calculating incidence All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) A + C (25,495) B+D (100,597) WHAT IS THE Overall incidence = (A + C) / N N (126,092) Incidence of death amongst people newly diagnosed with type 2 diabetes? Data from Vamos et al. (2012) Calculating incidence All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OVERALL INCIDENCE Incidence of outcome = (A + C) / N HOW WOULD A + C (25,495) B + D I CALCULATE N (126,092) (100,597) Incidence of death amongst people newly (3535 + 21,960)/ 126,092 = 25,495 / 126,092 = 0.20 diagnosed with type 2 diabetes with Incidence of outcome in exposed = 0.20 or 200 per 1000 people in 1990 to 2005 (15 years) cardiovascular disease? Data from Vamos et al. (2012) Diabetes: numerator and denominator Changing/varied/complex diagnostic criteria, e.g. ◦different tests / changing thresholds within tests / no test i.e. - self-reported doctor diagnosis ◦self-reported diabetes medication usage – e.g. metformin ◦Type 2 - slow onset + usual presentation without acute metabolic disturbance in type 1 complicates identifying time of onset Diabetes: numerator and denominator Changing screening, management and recordkeeping practices ◦Quality and Outcomes Framework (2004) - improved screening and management record-keeping ◦e.g. Primary Care Trusts screen for cardiovascular risk in 40 75 years often with glucose test ◦greater NHS patient choice = greater movement between / out of NHS Trusts makes it harder to follow patients Why does it matter? Affects incidence estimates ◦False positives and negatives ◦Increased/decreased numerator and denominator Affects coverage of findings ◦Exclude sub-populations of interest from numerator/denominator Affects identification of causal factors and outcomes ◦Obscure identification/ascertaining impact of exposures, and of temporal or geographical trends Affects service planning and commissioning ◦Unable to meet demand/wasted resources ◦Inappropriate targeting of resources Affects treatment ◦Under or over-investigation/over-treatment of individuals How might we find data to measure incidence? The cohort study • Focus on identification of exposures relevant to particular outcome • Group of individuals free from outcome selected, usually at random • Participants selected into exposed and non-exposed group for exposure/s of interest • Occurrences of outcome (incidence) are recorded • Comparisons are made between groups with respect to incidence rates Types of cohort study A prospective cohort is preferred, following a population forward over time However a retrospective cohort can be used, especially with high quality routine data such as electronic health records Other types of cohort study could include: ◦Studying the entirety of a population ◦Studying the entirety of a population with a specific period (e.g. consecutive babies born in 2010) ◦Studying individuals with exposure/outcome to identify risk of poor outcome/ death Strengths and weaknesses of the cohort study More than one outcome related to single exposure Can offer some evidence of cause – effect relationship Good when exposure is rare Can calculate incidence and relative risk Potential for losses to follow-up Often requires large sample Less suitable for rare outcomes If prospective, long time to complete, expensive If retrospective, data availability and quality may be poor Vulnerable to confounding Confounding and the cohort study A confounder is a variable that influences both the exposure and outcome causing a spurious association. A cohort study offers more protection against confounding than a cross-sectional study because it establishes temporal precedence But it does not avoid the problem of both exposure and outcome being influenced by other variables Selection bias-systematic differences between groups (outset and/or during) Information bias-systematic differences in measurement between groups Sources of cohort study data? Seconda ry Primary + Cheap + If anonymous, minimal ethical/governance approval needed - Limited by what data already gathered - -Poor accuracy and missing data • Mortality registers • Hospital/medical records • Census data • Survey data + Gather additional data - Difficult to achieve representative sample - More expensive Sources of cohort study data? Seconda ry Primary • Mortality registers • Hospital/medical records • Census data • Survey data Measuring incidence: Relative risk Clear advantage of the cohort study is that we don’t just have to measure (crude) incidence – we can use the cohort study to compare incidence between the two groups Relative Risk or Risk Ratio (RR) is the risk of developing an outcome in exposed group compared to developing an outcome in unexposed group Incidence among exposed = __a _ (a+b) Incidence among non-exposed = __c _ (c+d) Exposed Outcom e No outcome Total Yes a b a+b No c d c+d RR = incidence of outcome among exposed = a/(a+b) Incidence of outcome among non-exposed c/(c+d) Cohort study data analysis: Relative risk Strength of association: Direction of association: <1.0 RR=1.0 Risk in exposed group less than the risk in non-exposed group. Risk in exposed group equal to the risk in non-exposed group. The exposure may be protective against the outcome (negative association). The exposure is not associated with the outcome (no association) >1.0 Risk in exposed group greater than the risk in non-exposed group. The exposure may be a risk factor for the outcome (positive association). - RR of 1.5  risk of outcome 50% higher in exposed than unexposed group - RR of 3.0  risk in exposed group is three times as high as unexposed - RR of 0.8  risk of outcome 20% lower in exposed than unexposed group Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OUTCOME INCIDENCE Incidence among exposed = __a _ (a+b) WHAT IS A + C (25,495) B+D N (126,092) Relative(100,597) risk of death amongst people newly Incidence among non-exposed = __c _ (c+d) RR = incidence of outcome among exposed = a/(a+b) incidence of outcome among non-exposed c/(c+d) diagnosed with type 2 diabetes with cardiovascular disease compared to without? Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OUTCOME INCIDENCE Incidence among exposed = 0.29 Incidence among non-exposed = _21,960_ (113,713) RR = incidence of outcome among exposed = WHAT IS A + C (25,495) B+D N (126,092) Relative(100,597) risk of death amongst people newly = 0.19 incidence of outcome among non-exposed 0.29 0.19 diagnosed with type 2 diabetes with cardiovascular disease compared to without? Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OUTCOME INCIDENCE Incidence among exposed = 0.29 Incidence among non-exposed = _21,960_ (113,713) RR = incidence of outcome among exposed = WHAT ISB + D A + C (25,495) N (126,092) Relative(100,597) risk of death amongst people newly incidence of outcome among non-exposed diagnosed with type 2 diabetes with cardiovascular disease compared to without? 0.29 0.19 RR= 1.53 in 1990 to 2005 (15 years) Cohort study data analysis: Relative risk Strength of association: Direction of association: <1.0 RR=1.0 Risk in exposed group less than the risk in non-exposed group. Risk in exposed group equal to the risk in non-exposed group. The exposure may be protective against the outcome (negative association). The exposure is not associated with the outcome (no association) >1.0 Risk in exposed group greater than the risk in non-exposed group. The exposure may be a risk factor for the outcome (positive association). - RR of 1.5  risk of outcome 50% higher in exposed than unexposed group - RR of 3.0  risk in exposed group is three times as high as unexposed - RR of 0.8  risk of outcome 20% lower in exposed than unexposed group Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OUTCOME INCIDENCE Incidence among exposed = 0.29 Incidence among non-exposed = _21,960_ (113,713) RR = incidence of outcome among exposed = incidence of outcome among non-exposed WHAT ISB + D A + C (25,495) N (126,092) Relative(100,597) risk of death amongst people newly diagnosed with type 2 diabetes with cardiovascular >1.0 disease compared to without? 0.29 0.19 Risk in exposed group greater than the RR= 1.53risk in non-exposed group. The exposure may be a risk factor for the outcome (positive association). Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) OUTCOME INCIDENCE Incidence among exposed = 0.29 Incidence among non-exposed = _21,960_ (113,713) RR = incidence of outcome among exposed = incidence of outcome among non-exposed WHAT ISB + D A + C (25,495) N (126,092) Relative(100,597) risk of death amongst people newly 0.29 0.19 diagnosed with type 2 diabetes with cardiovascular disease compared to without? Risk in exposed group is 53% RR= 1.53greater than the risk in non-exposed group. Calculating relative risk All‐cause mortality in patients newly diagnosed with type 2 diabetes, with and without established cardiovascular disease Death in study period (1990 to Total 2005) Cardiovascular disease at baseline Yes No Yes A (3535) B (8844) A + B (12,379) No C (21,960) D (91,753) C+D (113,713) B+D N (126,092) e.g. RR of 3.0 could = 0.003 / 0.001 i.e. 3 vs 1 person in 1000 (100,597) (absolute risk difference of 2 people) An issue with relative risk: or A + C (25,495) RR of 3.0 could = 0.90 / 0.30 i.e. 900 vs 300 people in 1000 (absolute risk difference of 600 people) Therefore, researchers should always provide absolute difference in risk in addition to relative risk A cohort study of diabetes Lin et al. (2010): Depression and Advanced Complications of Diabetes: A prospective cohort study ◦ Prospective cohort of 4623 primary care patients with type 2 diabetes ◦ Exposure = major depressive disorder ◦ Outcomes = microvascular events e.g. blindness, macrovascular events e.g. stroke. ◦ Results ◦ Major depressive disorder = significantly increased risk of adverse micro and macro-vascular outcomes ◦ Even after adjusting for diabetes severity and self-care activities ◦ Contributions ◦ Important and treatable modifier of diabetes outcome ◦ Prevention of microvascular and macrovascular events key personal, clinical and economic importance https://doi.org/10.2337/dc09-1068 Summary Incidence is a measure of the number of new cases of a condition in some given time period – in a month or a year, for example – expressed as a proportion of a population which is at risk Prevalence depends on the incidence of a outcome and the time between onset and recovery (or death) ◦Prevalence = incidence x outcome duration Incidence = number new cases in a given time period / total population at risk Summary Cohort studies are used to measure incidence ◦ Group free from outcome identified at baseline and followed up to record incidence ◦ Often selected at baseline into a group with exposure of interest and a group without ◦ The two groups are then compared with respect to incidence – to test association between exposure and outcome Relative risk is a way of quantifying the exposure-outcome association – typically used in cohort studies ◦ Ratio of incidence in exposed compared to non-exposed group ◦ Calculated as incidence in exposed group / incidence in non-exposed group ◦ Provides both strength and direction of association Cohort studies offer some protection against confounding as exposures measured before outcomes – but are still vulnerable to effects of other variables confounding exposure-outcome association Next time… Collecting data about people and comparing the health of groups D R C L I O B E R RY S E N I O R L E C T U R E R I N H E A LT H C A R E E VA LU AT I O N A N D I M P R OV E M E N T C . B E R RY @ BS M S . AC . U K