Lecture 17 - Measuring Disease Part 2 PDF

Measuring disease – Part 2 Time = 8:30 AM  Outline of Measuring Disease Part 2 Incidence risk is the probability of becoming a case in a specified time interval Incidence rate is the number of new cases produced per animal time unit Prevalence and incidence...

Measuring disease – Part 2 Time = 8:30 AM  Outline of Measuring Disease Part 2 Incidence risk is the probability of becoming a case in a specified time interval Incidence rate is the number of new cases produced per animal time unit Prevalence and incidence related to each other via duration of infection Epidemic curves plot incidence. Epidemic curves provide information about source of infection, incubation times, and R0 Epidemic curves can demonstrate efficacy of interventions. Example of lockdowns on incidence of respiratory diseases in children in Finland Demonstrate how reproduction number (R0) can be estimated from epidemic curves. Requires definition of the generation interval (Tc) of infectious disease Estimate R0 from data of the 2003 SARS epidemic in Hong Kong  Incidence risk Number of new disease cases during time interval Incidence risk = Number of individuals at risk at start of time interval Incidence risk is the number of new cases in a time interval divided by the total population at risk at start of the time interval Incidence risk is the proportion (or percentage) of individuals in the population at risk that develop an infection over a defined time interval Incidence risk cannot be interpreted without a time interval! Incidence risk is also called the cumulative incidence  Limitations with incidence risk Incidence risk of upper respiratory tract infections (cold and flu) during the semester (January to May 2018) among the 80 students who enrolled in the class in January Over course of the semester, 20 students developed URT infections, and 5 dropped out of the class Incidence risk is 20/80 = 0.25; 25% of students in the class developed URT over the 2018 winter semester Question 1: Does incidence risk consider the 5 students that dropped out of the class? Question 2: Does incidence risk consider when the 20 students developed their URT?  Incidence rate Number of new disease cases during time interval Incidence rate = Number of animal − time units at risk during time interval Incidence rate is the number of new cases during the time interval divided by the total number of animal time units at risk Incidence rate allows animals to enter and leave the population at risk over the course of the study period. Monitor how much time each animal spends in the population at risk. Incidence rates calculated using first occurrence of the disease for each animal. Once an animal has had the disease it is no longer part of the population at risk An animal time unit is one animal for a defined time unit (dog-days, cow-months, person-years). Chosen time unit (days, years) will depend on disease dynamics  Exact method of calculating incidence rate Study to determine incidence rate of disease in group of 12 subjects over a period of 14 years (1980 to 1994) Subjects were disease-free at start of study; 5 and 7 subjects enrolled in 1980 and 1981 Subjects 2, 5, and 11 developed disease after 10, 4, and 8 years Six subjects became lost to follow-up. Only subjects 3 and 4 made it to the end of the study in 1994 12 subjects contributed 102 person-years of risk to the study Incidence rate = 3 cases/102 person-years = 0.029 cases per person-year = 2.9 cases per 100 person- years from 1980 to 1994  Example of exact method to estimate cow mortality rate Estimate incidence rate of death for a herd of 100 cows over a period of 12 months Over 12-month study period, 10 cows die 5 cows die after 2 months, 2 cows die after 5 months, 3 cows die after 8 months Use the exact method to calculate the mortality rate (incidence rate of death) Thus, 5, 2, 3, and 90 cows contribute 2, 5, 8, and 12 months of risk Sum of cow months of risk is 5*2 + 2*5 + 3*8 + 90*12 = 1124 cow months Incidence rate = 10 cow deaths/1124 cow months = 0.008896797 deaths per month Mortality rate is synonym for the incidence rate of death Mortality rate in cow herd is 88.97 deaths per 10,000 cow-months 8:40 AM  Use approximate method to estimate cow mortality rate Use the approximate method to estimate the mortality rate if we don’t have exact information on when cows died Assume that 10 deaths occurred halfway through the time interval (i.e., 6 months) 90 cows x 12 months + 10 cows x 6 months = 1140 cow months Cow mortality rate is 10 deaths/1140 months = 0.00877193 deaths per month Shortcut to calculate total animal units of risk is to multiply average population size by time interval: 95 cows x 12 months = 1140 cow months Mortality rate for approximate method (87.72 deaths per 10,000 months) similar to exact method (88.97 deaths per 10,000 months) Question: Why is mortality rate higher for exact method (88.97 deaths per 10,000 months) versus approximate method (87.72 death per 10,000 months)? When to use incidence risk versus incidence rate Incidence risk Incidence rate Easily calculated and Considers losses to Strengths understood since it follow-up and when measures risk (probability) disease occurs Need individual follow- Does not consider losses up, which is costly and Limitations to follow-up or when time-consuming disease occurs Interpretation is not intuitive Dynamic populations. Fixed populations over Appropriate use Fixed populations over short time intervals long time intervals  Summary of measures of disease risk # of animals with disease at a point in time Prevalence = Total # of animals in the population at that point in time # of new cases of disease during observation period Incidence risk = Total # of animals at risk at beginning of observation period # of new cases of disease during observation period Incidence rate = Total observation time at risk  The epidemiologist’s bathtub The epidemiologist’s bathtub shows the relationship between the prevalence, incidence and average duration of infectious disease Water in the bathtub is the prevalence of disease in the host population. Prevalence is 20% if the bathtub is 20% full. Water entering bathtub are new cases of disease per defined time interval, which is the incidence. Infected individuals can recover from disease (evaporation) or they can die (go out the drain) Prevalence, incidence rate, and average duration of disease If population is in a steady state (gains and losses of infected individuals are constant), the relationship between prevalence (P), incidence rate (IR), and duration of infection is as follows: P/(1 – P) = IR * Average duration of infection For diseases with a low prevalence, 1 – P ~ 1 and equation becomes P = IR * Average duration of infection Question: What happens to the prevalence of disease, if the duration of disease increases? Question: What happens to the prevalence of disease, if the incidence rate of disease increases?  Change in the prevalence of HIV in Kenya HIV causes AIDS in human patients Bars show annual prevalence of HIV in Kenya from 1996 to 2006. At this time, AIDS was inevitably fatal Development of anti-retroviral drugs did not cure HIV infection, but allowed infected people to live longer (i.e., increased duration of infectious period) Prevalence of HIV increased from 1990 to 2000 because the average duration of the infection increased After 2000, availability of anti-retroviral drugs decreased, mortality rate of people infected with HIV increased, and the prevalence of HIV in the population decreased Time = 8:50 AM  Types of epidemic curves Epidemic curves graph new cases over time (i.e., they plot the incidence) Type of epidemic curve provides information about source of outbreak and duration of incubation period Epidemic curve can be used to estimate R0 for infectious disease Types of epidemic curves include: (1) point source, (2) continuous common source, (3) propagated source, and (4) intermittent source  Point source epidemic Pronounced spatiotemporal clustering of cases Common source; cases linked to an event (e.g., contaminated salad at wedding dinner) Examples include food poisoning outbreaks caused by infectious agents or their toxins Hepatitis A virus is transmitted via faecal-oral route. Infected person contaminates food Average incubation period is 28-30 days (range is 15-50 days) Cases occur from April 27 to May 21 (~25 days) Question: What factors cause variation in the incubation time?  Continuous common source epidemic Like point source epidemic -> single source of exposure Source of exposure is prolonged -> outbreak is longer Duration of epidemic exceeds the span of a single incubation period Down slope of curve might be sharp if common source is removed or exhausted No cases occur beyond one incubation period following termination of exposure Example 1: Lead poisoning outbreak where source of lead is not discovered Example 2: Contaminated water supply with continual exposure  Cholera outbreak in London Cholera outbreak in Broad Street area of London, 1854 Outbreak lasted for 3 weeks Cholera has incubation period of 1 – 3 days Duration of outbreak (21 days) was 7x incubation period (3 days) suggesting continuous exposure to common source Dr John Snow concluded Broad Street pump was common source Pump handle was removed, and outbreak ended  Propagated source epidemic The epidemic is caused by an infectious agent that is spread from person to person or animal to animal Propagated source epidemics can last for months or even years, and are often characterized by multiple waves Classic epidemic curve for a propagated outbreak has progressively taller peaks that are spaced one incubation period apart Primary case (first infected individual) infects susceptible individuals that become secondary cases, which create tertiary cases and so on Propagated source epidemics are characterized by “build up” or “amplification” or “exponential spread”  Measles outbreak Measles outbreak starts with index case on 4 April Measles has incubation period of 10 days (range = 17 – 18 days) Person-to-person transmission Patients in secondary wave were infected by index case Successively larger peaks Peaks separated by incubation period Question: Why is the third peak smaller than the second peak? Time = 9:00 AM  Efficacy of control measures Epidemic curves can demonstrate the efficacy of control measures During COVID-19, many countries implemented lockdowns In 2020 in Finland, schools and daycares were closed from 16 March to 14 May Schools were reopened for 2 weeks followed by summer vacation Finland has excellent surveillance on common respiratory pathogens including adenovirus, influenza A and B, parainfluenza, rhinovirus, RSV, Mycoplasma pneumoniae, and Streptococcus pneumoniae Epidemic curves show that lockdowns significantly reduced incidence of many respiratory pathogens in Finnish children  Incidence of respiratory infections in 2020 Incidence of respiratory pathogens in 0- 14-year-old children in Finland in 2020 Pink boxes indicate school and daycare closures from 16 March to 14 May and summer vacation Shown are SARS-CoV-2, Mycoplasma pneumoniae, Adenovirus, Parainfluenza, Rhinovirus Respiratory infections declined in children in spring and summer of 2020 Question: What is an alternative explanation?  Compare incidence to previous years Compare incidence of respiratory infections between 2020 (red) versus 2017 to 2019 (blue) Lockdown in 2020 reduced incidence of respiratory infections compared to previous years Lockdowns did not reduce incidence of influenza A and B or RSV Epidemic curves show that lockdowns prevent respiratory infections Lockdowns have negative effects on learning and social life  Incidence, epidemic curves and R0 Incidence is number of new cases of diseases over a defined time interval Epidemic curves plot new cases over time (i.e., they plot the incidence). Reproduction number of infectious disease (R0) is number of new cases produced by single index case over the duration of the infection In class 7, we derived R0 from epidemiological models, but link with real-world data was lacking Question 1: Is there a relationship between epidemic curves and R0? Question 2: Can we estimate R0 from epidemic curves?  Epidemic curves and R0 and Rt Graph shows epidemic curve and values of R0 and Rt Epidemic only grows at R0 when the population is completely susceptible Time-varying effective reproduction number (Rt = R) is more useful measure of how epidemic changes over time When Rt > 1, epidemic is increasing and new infections per unit time > recoveries per unit time When Rt = 1, new infections per unit time = recoveries per unit time When Rt < 1, epidemic is decreasing and new infections per unit time < recoveries per unit time  Time unit for R0 9:10 AM Reproduction number (R0) is the number of secondary cases produced by the index case in completely susceptible population Here, R0 for SARS is 4.0, and index case causes 4 secondary cases Fast- and slow-moving diseases like FMD and Johne’s disease can have a similar values of R0, despite operating on very different time scales Question: What is the time unit associated with R0?  Generation interval (Tc) of an infectious disease Time interval between peaks is the generation interval (Tc), which includes the incubation period plus part of the illness period Time interval between index case on 4 April and wave 2 peak on 2 May is 28 days (2 May – 4 April), which represents 2 generations Tc = time / number of generations = 28 days/ 2 generations = 14 days We can use the generation interval (Tc) to estimate R0 from epidemic curves The relationships between time, number of generations, and generation interval are shown below 𝑡𝑖𝑚𝑒 𝑡𝑖𝑚𝑒 𝑡𝑖𝑚𝑒 = 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 ∗ 𝑇𝑐 ; 𝑇𝑐 = ; 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 = 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑇𝑐  Estimation of R0 from epidemic curves At start of epidemic, number of infected cases grows exponentially R0 can be estimated using data from exponential phase of epidemic curve Here N0 and Nt are the number of new infected cases at time t and time 0 Nt is the product of N0 and R0 raised to the number of generations that the disease has been circulating 𝑁𝑡 = 𝑁0 ∗ 𝑅0 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑁𝑡 𝑅0 = 𝑁0 1Τ𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑇 𝑐 Τ𝑡𝑖𝑚𝑒 𝑁𝑡 𝑁𝑡 𝑅0 = → 𝑅0 = 𝑁0 𝑁0  Calculate R0 for SARS-CoV First SARS epidemic started in Hong Kong in 2003 15 February, index case occurred in Hong Kong 28 March, 425 cases in Hong Kong at 41 days after index case Contact tracing found that the mean generation interval (Tc) was 8.4 days R0 was calculated to be 3.455 Question: why is it important that these data were collected before lockdown measures were imposed? 𝑇 𝑐 Τ𝑡𝑖𝑚𝑒 8.4Τ41 𝑁𝑡 425 𝑅0 = = = 4250.204878 = 3.455 𝑁0 1  Relationship between R0 and epidemic curve index secondary tertiary day01 day02 day03 day04 day05 day06 day07 day08 day09 day10 day11 day12 day13 day14 day15 day16 day17 day18 day19 day20 I1 E E E E E I I I S1 E E E E E I I I S2 E E E E E I I I S3 E E E E E I I I T1 E E E E E I I I T2 E E E E E I I I T3 E E E E E I I I T4 E E E E E I I I T5 E E E E E I I I T6 E E E E E I I I T7 E E E E E I I I T8 E E E E E I I I T9 E E E E E I I I Total 1 1 1 3 3 3 9 9 9 Interval 1 X X X X X X X Interval 2 X X X X X X X Relationship between R0 and epidemic curve Infectious disease has an R0 of 3.0 Latency/incubation period = 5 days; Infectious/illness period = 3 days Transmission to susceptible hosts occurs on 2 nd day of illness Mean generation interval (Tc) is 6.0 days Experimentally infect index host on day 1 Index case peaks on day 7, tertiary cases peak on day 19 Time interval is 19 – 7 = 12 days, which is 2 generation intervals 𝑇 𝑐 Τ𝑡𝑖𝑚𝑒 6Τ12 𝑁𝑡 9 𝑅0 = = = 90.5 = 3.0 𝑁0 1  Summary of Measuring Disease Part 2 Understand the differences between prevalence, incidence risk, and incidence rate, and know how to calculate these measures for examples Understand how prevalence, incidence and duration of infection are related. Recognize the different types of epidemic curves and understand the underlying diseases that generate them Give examples of how epidemic curves can demonstrate the efficacy of an intervention Understand that R0 is defined for the generation interval (Tc) of the disease and that it is not measured on an absolute time scale Understand how R0 can be estimated from epidemic curves. Be able to calculate R0 from epidemiological data if provided with the formula Time = 9:20 AM End of course # 17  Example calculation of incidence risk A cattery with 100 cats is experiencing an outbreak of feline viral rhinotracheitis (FVR). During week 1, 20 cats develop FVR. During week 2, 8 more cats develop FVR Calculate incidence risk for week 1, week 2, and weeks 1 and 2 combined Incidence risk for week 1 is 20/100 = 0.20 per week Incidence risk for week 2 is 8/80 = 0.10 per week Incidence risk for 2 weeks is 28/100 = 0.28 per 2 weeks Notice calculation in week 2, excludes cats that became infected in week 1 Notice incidence risk is reported with time interval (1 week, 2 weeks)  Incidence risk of death during childbirth In 19th century, women experienced very high mortality during childbirth due to unhygienic medical practices and puerperal fever Puerperal fever is a bacterial infection of the female reproductive tract following childbirth or miscarriage Monthly incidence risk of death during childbirth at Vienna General Hospital from 1841 – 1849 shows percent of women that died during childbirth Doctors used same tools to perform autopsies and deliver babies with no sterilization or hand washing In 1847, Dr Ignaz Semmelweis introduced a chlorine handwash and incidence risk in maternity clinic dropped from 18% to < 2% This is an example of where the incidence risk is calculated for a specific event (childbirth) and so there is no associated time interval  Dr Ignaz Philipp Semmelweis (1818-1865) Ignaz Semmelweis (1818 – 1865) was a Hungarian doctor at Vienna Hospital who noticed association between poor hygiene and puerperal fever In 1861, he published a book on his findings. Medical establishment ridiculed his findings. He died in an asylum in 1865 at age of 47 years 20 years later, Semmelweis was vindicated by Pasteur and medical establishment recognized the importance of hygiene The Semmelweis effect or Semmelweis reflex is the human tendency to reject new evidence or ideas that contradict established beliefs  Examples of familiar incidence rates Annual death rate. Number of deaths per 1000 inhabitants per year. Canada has 8.2 deaths/1000 inhabitants whereas the Ukraine has 18.6 deaths/1000 inhabitants. Total fertility rate per woman. The average number of live births a woman would have by age 50 (time interval is reproductive lifespan). Canada has 1.5 births per woman, Niger has 6.6 births per woman, South Korea is 0.9 births per woman Annual intentional homicide rate. Number of intentional homicides per year (excludes war-related deaths, suicide, non-intentional homicide). Canada has 2.273 homicides per 100,000 inhabitants, whereas Haiti has 40.845 homicides per 100,000 inhabitants. Annual lung cancer rates. Number of patients diagnosed with lung cancer per year (standardized for age structure). China has 40.8 cases per 100,000 inhabitants, which is considerably higher than the world average of 23.6 cases per 100,000 inhabitants Annual incidence of tuberculosis in Canada by population Incidence of slow-moving infectious diseases like tuberculosis (TB) is often expressed per year. Compare annual incidence of TB among population subgroups. Annual incidence of TB in Inuit (188.7 cases per 100,000) is 472x higher compared to Canadian-born, non-Indigenous populations (0.4 cases per 100,000) Reasons include poverty, crowded and poor-quality housing, malnutrition, barriers to health care, mistrust of Western medicine  Incidence rates have a time unit Incidence rate is the number of new cases per unit of animal time Incidence rate measures the rapidity with which new cases develop over time Example: 50 cats in a cattery experience 10 cases of FVR in a one-year period Activity: calculate the incidence rate per year, per month, and per week IR = 10 cases/(50 cats x 1 year) = 10 cases/(50 cat years) = 0.20 cases/cat year IR = 10 cases/(50 cats x 12 mo) = 10/(600 cat months) = 0.0167 cases/cat month IR = 10 cases/(50 cats x 52 wk) = 10/(2600 cat weeks) = 0.00385 cases/cat week Magnitude of the incidence rate depends on the time unit used! Time = 8:40 AM  Epidemic curve, common intermittent source Epidemic curve for common source outbreak with intermittent exposure Intermittent access to a play yard contaminated with roundworm and hookworm eggs Case numbers peak at irregular times corresponding to earlier exposures X-axis is day of onset of clinical signs and y-axis is number of cases  FMD epidemic in UK in 2001 2001 FMD epidemic in UK lasted 6 months Feb 20: Confirmation of FMD at an abattoir for pigs Feb 23: National movement ban on animals March 5-21: Disease control strategy becomes more aggressive March 23: Most aggressive strategy March peak: 300 new cases/week May: 50 cases per week  Map of FMD epidemic in UK in 2001 In June 2000, UK had 40 million sheep, 9 million cattle, and 6 million pigs Culled 6.45 million animals: 5.25 million sheep (13.1%), 758,000 cattle (8.4%), 449,000 pigs (7.5%) Movement ban, biosecurity, and culling saved many animals 2000 farms had cases of FMD Most cases occurred in the northern counties near the border with Scotland Why this geographic distribution of cases?  Movement of asymptomatic sheep in Feb 2001 Index case was a pig farm in northern England Farmer fed untreated restaurant swill to his pigs Sheep on neighbouring farms became infected 13 Feb: sheep sent to market; sheep were dispersed over a large area in UK Pan-Asian strain causes obvious disease in pigs and cattle but is asymptotic in sheep 20 Feb: Lab confirmation. FMDV was already seeded on 57 farms in 15 different counties! 23 Feb: Movement ban: 120 farms with cases in 22 different counties  Generation interval (Tc) of disease Symptoms and infectious period Index case Incubation and latency period Symptoms and infectious period Secondary case Incubation and latency period 1 2 3 4 5 6 7 8 9 10 11 12 Time (days)  Generation interval (Tc) of disease continued … For convenience, latency period = incubation period = 5 days Duration of illness = duration of infectious period = 1 day Veterinarians counting cases will see ill animals on days 6 and 11 Primary and secondary cases are separated by period of 5 days Mean generation interval (Tc) = mean interval between illness of primary case and illness secondary cases For fast-moving infections like FMD, Tc < 1 week For slow-moving infections like Johne’s disease, Tc ~ several years  Major take home messages R0 is the number of secondary cases by an index case in a completely susceptible population Effective reproduction number (Rt) is a more useful measure Epidemic is increasing (Rt > 1), peaking (Rt = 1), waning (Rt < 1) R0 can be estimated from epidemic curves Mean generation interval (Tc) is the time between peak of secondary cases and peak of primary case R0 does not have time units; increase over 1 generation interval of pathogen Tc for Ebola and Johne’s disease is weeks vs years, but R0 could be same When measured on units of time, Ebola is much faster than Johne’s disease