Epidemiology Quiz 3 Study Guide

Step 7 & 8 (outbreak)

- Collect information to develop hypothesis that can be tested - Consider factors related to time, person, and place - May change over investigation period - Should know source and mode of transmission - Must determine exposure that caused the disease by implementing a study

Sentinel

- Collection of in-depth data in hopes that it represents the total population - Inexpensive • Example: FoodNet has 3 locations across Canada to monitor enteric illnesses

2 approaches population and high risk

- Combo of both is ideal! population and high-risk

Objectives of outbreak investigations

- Halt process/spread of disease - Determine reasons for the outbreak - Institute corrective measures - Make recommendations to reduce further risk of future outbreaks

Problems with false negatives

- Possible death if disease needed to be caught early

Case-control studies

- Observational study - Selects subjects based on their DISEASE/OUTCOME STATUS - Case = O+ - Control = O- - Look back in time and assign each subject as either exposed (E+) or non-exposed (E-) - *Exposure status is unknown when subjects enroll in study 1. Purposively sample individuals from population that are O+ and O- 2. Classify each individual based on exposure status (E+ or E-) 3. Compare proportion of O+ that are exposed and non-exposed to proportion of O- that are exposed and non-exposed Advantages - Good for rare diseases, but not rare exposures • Can purposively select individuals with disease from population, but don't know exposure status when selecting subjects - Best for investigating source of outbreak • Most sick people have already been identified - Good for investigating multiple risk factors for single disease • Good for diseases with an unknown cause, and allows examination of possible factors in one study - Fast and inexpensive because they are retrospective (look back in time for E status) Disadvantages - Can't calculate disease incidence or prevalence • Selecting based on disease status, therefore setting the level of disease in study (doesn't reflect level of disease in population) • Only about determining associations between E and O - Can only examine one outcome • Select subjects on the basis of this one outcome - Most prone to bias out of the 3 observational study types • E and O have already happened, must look back in time - Temporal relationships are an issue if using prevalent cases • Associations between E and O may be attributed to subjects that already have the disease, as opposed to attributed to subjects developing the disease - Control selection may be difficult Source of cases - Must have a case definition: clear and explicit set of criteria that minimize the likelihood of missing a true case and of a non-affected person being falsely classified as a case • Symptomatic descriptions, lab confirmation, time frame (date of diagnosis), geographical location Study base: population from which cases and then controls are obtained Primary study base: population that from which the cases arise can be easily be defined, we have a good sense of the population we are selecting cases from (example: kids at a daycare, livestock on a farm) Secondary study base: group of potential subjects are one or more steps removed from the primary population, population is made up of people from a wide range of backgrounds, population is difficult to define and likely unrepresentative of surrounding population, hard to define primary population (example: patients at a hospital, lab, referral clinic). It's important you know primary population to select controls! Control selection - Controls must be representative of source population in terms of exposure and potential confounding variables - Match controls to the study base we choose from - Primary study base: can choose control subjects from community, school, friends • Use of non-hospitalized subjects as controls - "econdary study base: trickier because we don't know source population, just know they are at a given location for a specific issue (outcome positive) • Use of hospitalized patients as controls 1. Randomly select controls from list of non-cases in the population (in the same location as cases but for a different reason). Example: cases are hospitalized patients for brain tumors, controls would be hospitalized patients for heart attacks. 2. Choose controls with completely different outcomes to cases so the exposures don't overlap (example heart attacks and strokes are really similar, choose controls with broken legs) Analysis - Can't calculate relative risk for case-control studies, because you can't calculate incidence - But, you can use a 2x2 to calculate odds ratio (a*d/b*c) Bias - Selection bias: control group must represent the prevalence of exposure in the non- diseased members of the population from which cases were drawn - Information bias: recall bias is a particular problem if accuracy differs between cases and controls - Confounding bias: always a factor Summary 1. Proper selection of cases (O+) and controls (O-) 2. Ensure comparability between cases and controls 3. Accurate unbiased history of exposure for the factor of interest 4. Control for confounding

Selection bias

- Selecting cases and controls (exposed and non-exposed) so that an association is observed that in reality is not there - Involve a systematic difference between those participating and those who are not

Portals of entry

- Skin - Alimentary tract - Respiratory tract - Urinary tract

Need to know

- When exposure took place - When disease began - Length of incubation

Confounding factor

1. Must be associated with exposure 2. Must be associated with disease 3. Must not be a result of exposure

9. The units for the quantity you calculated in Question 8 could be expressed as: A. cases per 100 persons B. percent C. cases per person-year D. cases per person per year

C, D. The person-time rate presented in Question 8 should be reported as 5 cases per 250 person-years. Usually person-time rates are expressed per 1,000 or 10,000 or 100,000, depending on the rarity of the disease, so the rate in Question 8 could be expressed as 2 cases per 100 person-years of follow-up. One could express this more colloquially as 2 new cases of eye disease per 100 diabetics per year.

Chapter 11 Relative risk:

Relative risk: - Is there association between exposure to a factor and development disease? - How much more likely are you to get the disease if you are exposed vs. if you are not? - Ratio of the probability of getting the disease with exposure to probability of getting disease without exposure - Includes the numerator in the denominator! - This is good for a cohort study where you know has been exposed to risk factor and who hasn't! Disease develops No disease develops Exposed a b a+b Non-exposed c d c+d a+c b+d n = a+b+c+d RR = 1 →no difference between exposed and non-exposed RR > 1 →exposed have greater risk, positive association, may be causal RR < 1 →non-exposed have greater risk, negative association, may be preventative (ex. vaccines)

Validity of screening tests

Validity: test's ability to distinguish those who have disease from those who don't

- NAR for mortality =

anyone alive • Narrow this down to anyone with a particular factor for cause-specific

1. Population-based

applied to entire population

Causal

characteristic →disease

1. Development of disease Susceptibility

exposure to agent →disease

Carrier

individual harbors the organism without being "infected" (no antibody response), can be chronic (lasting forever), or of limited duration

Sampling frame

list of all study subjects in the source population (contact information)

Continuous variables

no positive or negative

Prevalence

number of cases of disease

Sporadic

occurs occasionally in a population

Target population

population to which it might be possible to extrapolate results

Case-fatality

proportion of cases with a disease that die of that disease

Incidence

refers to the occurrence of new cases of disease or injury in a population over a specified period of time. Although some epidemiologists use incidence to mean the number of new cases in a community, others use incidence to mean the number of new cases per unit of population

Example

relative risk of 2 means the exposed group has two times the risk of disease than the non-exposed group

Isolation can be useful method of preventing spread of disease, but

sometimes people have unknowingly spread disease during incubation period

• Outbreaks occur when

susceptible individuals > immune individuals

Negative predictive value (NPV)

the probability that a patient who tests negative actually doesn't have disease

Predictive value of a test Positive predictive value (PPV)

the probability that a patient who tests positive actually has the disease • PPV related to disease prevalence, specify when disease is infrequent • Higher prevalence = higher PPV • Best to target high-risk groups, more likely to get a true positive and more likely to seek care

Incubation period

time between infection and presentation of symptoms (organism may need time to replicate)

Restriction/exclusion

when selecting participants, researchers will exclude people based on presence of one factor to ensue only one level of CFV is present

Pandemic

worldwide epidemic

Net specificity =

(spec. of A)(spec. of B)(total without) / total without

Steps of an OUTBREAK Investigation

*Don't have to happen sequentially, no two investigations are the same 1. Determine existence of outbreak 2. Confirm diagnosis 3. Prepare for field and assemble team 4. Implement immediate control measures 5. Define a case and case-finding 6. Describe data by time, place, and person 7. Develop hypothesis 8. Test hypothesis - implement studies 9. Implement prevention and control measures 10. Disseminate findings, conduct evaluation Steps 1 & 2 Two ways outbreak is detected Community-identified - Illness reported by individuals, physicians, or public health authorities - Syndromic surveillance is improving rates at which outbreaks are being community-identified Lab-based: - Ongoing surveillance of lab isolates by health departments is performed to determine if the observed number of cases exceeds the expected level - Use notifiable disease records, lab records, death records, disease registries... - Labs also monitor for unusual/rare strains of common pathogens - Changes in reporting procedures, case definitions, increased public interest, improved diagnostic tests are all important because they all allow more samples to be collected/disease to be detected - *This does not necessarily indicate an outbreak, consider false alarms

James Lind & scurvy

- "Father of nautical medicine" - Conducted randomized trials on ships in the navy when men started to become ill with scurvy - Treated groups of men in various ways - Discovered that vitamin C cured them All of these scientists didn't understand biology of disease, but took actions based on observations

Epidemiology triad

- Agent (thing causing infection) - Host (thing being infected) - Environment (how and where this happens)

Population attributable risk

- Amount of outcome in the entire population that is caused by the exposure - Prevalence/rate of outcome in population - prevalence/rate of outcome in non- exposed - Null value of 0

Infectious agents

- Bacteria - Prions (infectious proteins) - Viruses - Parasites - Fungi

Parallel testing (simultaneous) *used in clinical settings

- Both tests must be negative to indicate a negative result - More invasive and expensive - Net increase in sensitivity but decrease in specificity

Multiple tests In Series testing (two-stage)

- Both tests must be positive to indicate a positive result - Avoid false positives but miss false negatives - Net decrease in sensitivity but increase in specificity - Less invasive/more comfortable test given first - Those with positive results come back for more invasive/uncomfortable test

Problems with false positives

- Brought back for further tests - Strain on health care system - Patient anxiety - Label of "sick" never completely goes away

1. Determining who/what to sample

- Can affect validity of results if not done properly

Confirming outbreak

- Community-identified: try to lab-confirm the diagnosis by encouraging individuals to submit samples, then ensure the cases are clinically similar and epidemiologically linked - Lab-identified: clinically confirm diagnosis with standardized lab techniques (ex. serology). Consider if the number of cases is artificial (due to more reporting or better tests) or increased due to incidence of disease - Regardless of method, investigators must not wait for lab confirmation to begin investigation to minimize impact of potential outbreak Step 3 & 4

3 types of cases outbreaks

- Confirmed case: must have laboratory verification - Probable case: has typical clinical features of disease but no lab confirmation - Possible case: has fewer of the typical clinical features Step 6

Risk/rate difference

- Difference between level of risk in non-exposed and level of risk in exposed Null value = 0 for measures of effect (Exposure has no different outcome than non- exposure) RD > 1 means positive association, probably causal RD < 1 means negative association, probably preventative

Modes of transmission

- Direct: host →host - Indirect: infected/contaminated vehicle or vector →host - Horizontal: between members of same species - Vertical: mother →fetus - Ingestion - Sexual contact - Aerical/airborne - Latrogenic: from medical test or hospital

John Snow & cholera

- Discovered correlation between cholera patients and proximity to a water pump - Later concluded that it was contaminated All of these scientists didn't understand biology of disease, but took actions based on observations

Propagated outbreak

- Disease spreads slowly - Results from person-person or animal-animal - More than one incubation period - Direct contact, vehicle-borne, vector-borne - Duration is longer than any common source outbreaks with distinct incubation periods in epidemic curve

Sir Percival Pott & cancer

- First to find cause and effect relationship between carcinogens and cancer - Noticed chimney-sweepers developed testicular cancer from soot buildup in and around scrotum All of these scientists didn't understand biology of disease, but took actions based on observations

Koch's postulates

- Good criteria for infectious disease 1. Organism always found with disease 2. Not found with any other disease 3. Organism can be extracted and cultured, produces disease in a susceptible host New guidelines - Includes non-infectious diseases too 1. Temporal relationship: exposure must precede disease 2. Strength of association: more relative risk →more chance of causation 3. Dose-response relationship: higher dose of exposure →higher risk of disease 4. Replication of findings: consistent results in different studies 5. Biological plausibility 6. Alternate explanations: must rule out other non-causal possibilities 7. Cessation of exposure: decreased risk when exposure decreases (unless irreversible) 8. Consistency with other knowledge 9. Specificity of association: exposure associated with only one disease (this is a weak guideline because it's easy for one factor to cause many problems)

Attributable risk

- How much of the disease that occurred can be attributed to a certain exposure? - RD exposed - RD non-exposed / RD exposed - Proportion of disease in exposed group that is due to the exposure

Reliability

- How repeatable a test is Intrasubject variation: within individual subjects, characteristics change and so do test environments Intraobserver variation: variation in reading same test results at different times by same person, subjective factors Interobserver variation: differences between examiners

Prevention

- Identify high-risk groups to direct prevention - Identify which factors make them high risk • Non-modifiable: age, sex, race • Modifiable: weight, diet, life-style

Place & People (OUtbreak)

- Identify individuals involved and where they are physically located - Focus on descriptive epidemiology when questioning (5 W's + how) - Visualization may provide clues regarding source of agent or nature of exposure

Incidence rate ratio

- Incidence rate of disease in exposed groups / incidence rate of disease in non- exposed groups - IR of disease in exposed group is X times the IR of disease in non-exposed group Number of + outcomes Time at risk Exposed a1 t1 Non-exposed a0 t0 Null hypothesis and confidence intervals 95% confidence interval: the range of numbers that we are 95% certain contains the true value We calculate what we believe to be the value, and then the CI tells us that we are 95% sure the true value falls within Null value = 1 for measures of association *For RR, OR, and IRR, a value of 1 would indicate no association between exposure and outcome. More than 1 means a positive association, and less and 1 means a negative association. For our value to be statistically significant, the value we get must be far enough away from the null. If our CI contains the null, and our calculated value also falls within this CI, it can't be statistically significant because we aren't more than 95% sure that the null isn't the true value.

Cross-sectional

- Is the exposure related to the outcome? - Randomly sample population of interest as a point in time, no information known about subjects prior to study, enroll them and then classify them - Compare prevalence of disease in exposed and non-exposed group 1. Randomly sample individuals from source population 2. Take measurements on each individual to determine exposure or outcome status (can collect information on multiple exposure or outcome variables) 3. Classify individuals based on exposure 4. Classify individual based on outcome 5. Compare prevalence of disease in exposed and unexposed groups Advantages - Can determine prevalence of exposure and outcome - Fast and inexpensive - Less potential for bias than case-control - Good for studying permanent factors (gender, race...) - Assess associations between multiple exposures and multiple outcomes - If random sampling is used, assess all measures of association and effect (no true rates though) Disadvantages - Not good for rare exposures or rare outcomes (because you're only looking at a single point in time, unlikely to encounter one) - Prone to selection, information, and confounding bias if not careful - Measure prevalence (number of cases at one point in time), not incidence (number of new cases) - Don't know if exposure caused outcome because of the short time frame you're looking at

Sample size for comparing two means Sample size for comparing two proportions

- Large associations are easier to detect than smaller ones Required sample size generally increases as: • The size of the difference between two means or proportions decreases (i.e., smaller difference needs to be found) • The level of power to detect a difference between two groups increases • The number of confounding variables you are controlling for in your study increases and • The number of hypotheses you are testing increases.

Field experiments

- Look at disease in real-world scenario - Evaluate whether procedure/treatment can prevent disease in those who are disease-free at the start of the study Advantages - More applicable to real-world →more relevant Disadvantages - More difficult to carry out than clinical because they involve healthy people (not diseased people) - Some may not develop disease for long periods of time - Requires more subjects and longer studies - No control over exposure/environment Designing a research question - Is this new treatment better than nothing/an older treatment - What is "better" Measuring outcomes - Need specific factors to measure success - Objective criteria is always better than subjective - Factors must be biologically/practically/economically important - Good outcome must be measured in some way Considering ethics - Is it ethical to intervene - Are negative controls ethical when some/any form of treatment has been proven beneficial? Must do so to test efficacy of treatments Need to maintain health in populations Defining populations - Need internal and external validity, because subjects are being selected and put into groups - Need strong inclusion and exclusion criteria - Need well-defined target population - Population must be representative - *These things are needed in all studies, but particularly with this type Choosing groups - Need comparison groups (control groups) to measure efficacy of intervention types Positive control groups: comparison groups receives alternate intervention • Involves current best alternate treatment Negative control groups: comparison groups receives no intervention • Is an intervention better than nothing? • Placebo Selecting and assigning subjects - Randomization is preferred approach - Reduces likelihood that unidentified factors will influence results of study - Treatment and control subjects should be alike in every way Complete randomized design: all individuals randomly assigned to treatment and control groups Randomized block design: subjects first grouped based on factor of interest Cross-over design: each subjects spends time in control and treatment group (each acts as own control) Factorial design: used to assess two or more treatments Group size - Need to have enough subjects for s statistically significant difference to exist - Subjects per group increases as Treatment effect decreases (difference between groups decreases) Variations among individuals increases (difference between groups increases) Number of other factors/covariates increases Number of groups increases Bias - Wish bias: subjects or investigators fit data into a personal scheme based on what they think ought to be Placebo effect: if a patient knows they receive treatment they may report improvements that aren't really there - Differential admission bias: form of selection bias, subjects with certain characteristics are more likely to be admitted/referred to a hospital where to the samples are being taken →unrepresentative of target population - Recall bias: difficulty for subjects to remember past information - Misclassification bias: could be due to poor sensitivity/specificity Follow up - Period when investigators determine whether outcome has occurred in subjects - Blinding is good to prevent bias Single blind: subject or researcher/person administering is blind Double blind: both Triple blind: both plus data analysts (person assessing outcome) Analysis - Statistical significance: p value less than 0.05 (probability due to chance was less than 5%) - Biologically significant: does the intervention results in outcomes that improve health

Observation to action (examples) Ignaz Semmelweis & childbed fever

- Many woman dying after childbirth from unknown causes - Mortality higher in doctors than midwives - Doctors were not washing hands, transmitting disease from cadavers to patients - Originally unsupported by medical professionals All of these scientists didn't understand biology of disease, but took actions based on observations

Information bias

- Means for obtaining info from subjects are inadequate resulting in incorrect results

Incidence

- Measure of disease or outcome occurrence over time - Must establish who is "at risk" and confirm "new cases" - For chronic disease, prevalence is much higher than incidence because the duration is so long - For short-term illness, incidence will be higher than prevalence

Social media

- Modern - Follows trends on social media via words/phrases related to illness - Can identify outbreaks 2 weeks faster than other methods when done effectively - Must consider context of searches • Example: following tweets about the flu over a year and comparing it to the incidence Issues - Underreporting: number of cases that occur in reality but are never reported • Didn't go to hospital, hospital didn't report case, ... - Lack of denominator details: we often have numerator data (number of cases) but rarely know denominator data (how many people are at risk) • Can at least maintain stable denominator with active surveillance (number of samples collected) - False alarms: improvements/changes to tests, media hype about a disease • People are more likely to report a disease if it is of public concern - Passive surveillance can have problems with lack of recent data, lack of understanding of disease, patient privacy

Sampling

- Need to ensure that your sample is representative of the large population

disease surveillance and measures of morbidity Prevalence

- Number of existing cases of disease at a given time divided by the total number of persons in population at this time - Point prevalence: specific time or date, period prevalence: during a time span - Proportion, not a risk

Cohort studies

- Observational study - Selection of subjects based on EXPOSURE STATUS - Investigators follow these subjects for a period of time to determine incidence of disease (comparing development of disease in E+ and E- subjects) - All subjects must be disease-free at the beginning of study - Can used to calculate all measures of association (relative risk, odds ratio, and incidence rate ratio) and effect (risk/rate difference, attributable risk/rate, population attributable risk/rate, and population attributable fraction) - Can be prospective or retrospective (selected before exposure) Advantages - Good for rare exposures • Sample purposively based on exposure - Can study several outcomes - Allow for a temporal sequence to be establish (prospective) - Provide incidence data (new cases) - Not as prone to bias like recall bias Disadvantages - Bad for rare outcomes - Can only study one/few exposures (except general population cohort like Framingham study) - Selection bias may be a problem • Loss of follow up, especially in prospective (death, losing track, moving...) "Are those who remain significantly different from those who withdrew?" • Non-response bias - Information bias • Misclassification bias: changing in screening diagnostic methods over time, case definition differences • Recall bias for retrospective • Need to have clear case definitions, ensure equal follow-ups for both groups - Confounding bias - Expensive and time consuming, long follow up time Ensuring validity 1. Careful selection of exposure groups: ensure samples have comparable characteristics other than just the one of interest and representative target population. Have clear case definitions and make sure everyone starts disease-free 2. Follow-up: biggest challenge! Regular follow up with both exposure groups. Goal: minimal, non-differential withdrawal 3. Objective diagnosis of disease: accurate diagnosis (specificity and sensitivity), clear case definitions Type Advantages Disadvantages Cross-sectional Can study multiple E and O hypotheses Not good for rare E or O Relatively fast and inexpensive Temporal issues (E before O?) Good for permanent risk factors (sex, blood type) Prevalence, not incidence Case-control Good for rare O Not good for rare E Often fast and inexpensive Temporal issues Most subject to bias Cohort Good for rare E Not good for rare O Establishes a temporal relationship Often long follow-up times Provides incidence data Loss to follow-up bias Generally expensive

Odds Ratio

- Odds of exposure in the diseased group / odds of exposure in the non-diseased group - Good for cohort studies: start with exposures and then look at outcome • Odds of disease in exposed are X times the odds of disease in the non- exposed - Also good for case-control: start with outcomes and look at exposure • Odds of exposure in diseased group are X times the odds of exposure in the non-diseased group *Always conclude short answer questions with a statement explaining your answer and choice of formula *Can take the inverse of the equations if your answer comes out to less than 1 (may be easier to interpret) but remember all this means is that the factor is probably preventative (like with vaccines). Easier to say "those who did not have the exposure/factor are more likely". *OR and RR will be similar with low prevalence diseases

Edward Jenner & small pox

- Once infected, those who survived became immune to small pox - Noticed milk maids exposed to cow pox were also immune - Infected child with cow pox - He became immune to cow pox - Invented vaccines All of these scientists didn't understand biology of disease, but took actions based on observations

Clinical and subclinical disease

- Only clinical illness (showing signs and symptoms) is apparent - Important to account for non-clinical disease too, these people can unknowingly spread disease! - Severity of disease is proportional to the virulence (how good it is at causing disease) of the organism

Incidence risk

- Proportion of unaffected individuals who on average develop the disease over a period of time - Number of new cases of a disease that occur during a specified time period in a population at risk for developing disease

Evidence

- Randomized trials provide stronger evidence for a treatment/preventative action than other study designs

Associations

- Real/true vs. spurious/false - Real does not mean causal!!!

Observational

- Researchers have no control over allocation of subjects to groups being compared - Good for hypothesis generation - Don't change anything in the subject's life, just observe them in their natural habitat - May be better than experimental if... • Intervention is unethical (ex. giving someone disease, only giving one group effective treatment) • Intervention is too expensive • The exposure is complex and hard to control • It's impractical to administer the exposure

Experimental

- Researchers randomly allocate subjects to groups being compared (sort subjects, generally intervene) - Good for hypothesis testing - Laboratory trials or field trials

Herd immunity

- Resistance of a group to a disease to immunity of members, once disease spreads around a community, they all become immune - If we immunize high % of population, the rest will become immune due to herd immunity - Disease must be restricted to one host species, transmission must be direct, infection must induce solid immunity, constant and random mixing - Percent of immune population for herd immunity to occur depends on disease

Incidence rate/incidence density

- Speed with which new cases of disease occur - Denominator is the sum of units of time each individual was at risk • Person week = one person followed for one week or two people followed for half a week - Exact denominator: need to know time "at-risk" for each individual • At risk: disease free with potential to become a case • Once they become a base they are no longer considered at-risk until they clear the infection

Surveillance

- The actual systems put in place to gather information about disease occurrence in a population Three definitions of surveillance CDC (Centers for Disease Control and Prevention): ongoing systematic collection, analysis, and interpretation of health data essential to the planning, implementation, and evaluation of public health practice closely integrated with the timely dissemination of these data to those who need to know PHAC (Public Health Agency of Canada): the ongoing, systematic collection, analysis and interpretation of data essential to the planning, implementation, and evaluation of public health practice, closely integrated with the timely dissemination of these data to those responsible for prevention and control WHO (World Health Organization): the continuous, systematic collection, analysis and interpretation of health-related data needed for the planning, implementation and evaluation of public health practice Surveillance cycle - Continuous and systematic collection of data - Analysis of data - Interpretation of statistics/analysis - Dissemination of findings - Taking action to reduce disease - (Begins again) - CAIDA: Surveillance allows us to... - Describe current disease frequency and distribution - Describe natural patterns of disease (ex. flu prevalence rising and falling between seasons) - Predict outbreaks/epidemics (ex. tracking Ebola) - Monitor changes in agent characteristics - Display trends in health behaviours (ex. vaccines being lacking in a certain population) - Indicate where/when intervention is needed (such as quarantine, vaccination, preventative treatment) Collecting surveillance data Passive data collection - Data acquired from other primary sources that were derived for other purposes • Collecting data from samples sent to a lab for testing • Reports by doctors to public health agencies • Hospital discharge information • Absentee information from schools Pros - Inexpensive - Useful for notifiable diseases (required to be reported to public health agencies) Cons - Time issues: data may not be current, may be sporadic - Under-recording: disease events may not be recorded/submitted - Motivation of people to ensure information gets to where it needs to be - Collection between diseases may be inconsistent Active data collection - Data is actively searched for and collected for the goal of surveillance • Sample meats for enteric pathogens • Testing hospital workers for a tuberculosis • Swabbing carcases at slaughterhouses Pros - Timely data - More accurate than passive Cons - Expensive - May require additional infrastructure - Greater time requirements - Logistically difficult Forms of surveillance Lab-based - Data produced in a clinical/public health lab - Often for confirming new cases of infectious diseases - Important for national disease surveillance - Useful for detecting changes in behaviour or presence of pathogens & for evaluating the impact of prevention and control measures • Example: DNA fingerprinting of bacteria to detect outbreaks of disease

Other measures

- These rates don't have ITC and thus are not true rates!

Outbreak investigations

- Undertaken when a higher than expected proportion of infections are occurring in a particular population/location - Process followed to determine where infection started and how it spread

Intervention studies

- Used to investigate whether or not one intervention method is better than another - A type of EXPERIMENTAL study: some sort of intervention Different from observational, in which you just observe outcome - Called randomized trials - Useful for evaluating new forms of treatment (ex. drugs) Prophylactic/preventative studies: assess efficacy of vaccines or feed additives Therapeutic studies: assess efficacy of antibiotics or anti-cancer drugs Procedure 1. Clear hypothesis 2. Define target population 3. Define study population 4. Random assignment 5. Group A treated a. Better b. Not better 6. Group B treated (same time as group A) a. Better b. Not better 7. Follow up outcomes 8. Statistical analysis 9. Interpretations and conclusions Lab experiments - Randomized clinical trials - Look at disease in a controlled, clinical environment (usually hospital) - Subjects are treated and followed up for their condition - Challenge trials: subjects exposed to a particular factor of interest Advantages - Limits confounding factors - Total control over exposure/environment Disadvantages - Doesn't take into account factors/exposures present in the real world - Further field studies usually required

Syndromic

- Uses individual and population health indicators that are available before the lab results or confirmed diagnosis to identify outbreaks health events to monitor health status in a community - Quicker response than the other methods • Example: monitoring over-the-counter drug sales

Percent agreement & Kappa statistic Percent agreement

- Value that compares how much 2 observers agree on the results of a test - Only use results that at least one observer thinks indicates a positive/disease state, ignore those that they both agree is negative - Ignore d because they both agree the result is negative Observer 1 Observer 2 Positive Negative Positive a b Negative c d

Descriptive studies

- What, when, where, how - Aim to collect information about the occurrence of a disease but don't attempt to establish a cause-effect association - Frequency and distribution of disease - Count, sample, survey, assess risk - Good for generating hypotheses - Case reports: describe rare/unusual cases - Case series: describe usual clinical course of a condition by combining observations of a series of patients with the outcome of interest - Survey: collect various information from individuals in a population with no comparison group

Analytical studies

- Why - Investigate determinants of health and disease - Statistical analysis of epidemiological data to establish relationships between causative factors and disease occurrence - Compares between groups - Good for testing hypotheses - Two types: observational and experimental

1. Which of the following are frequency measures? A. Birth rate B. Incidence C. Mortality rate D. Prevalenc

1. A, B, C, D. Frequency measures of health and disease include those related to birth, death, and morbidity (incidence and prevalence).

A. Incidence proportion B. Incidence rate C. Prevalence D. None of the above ________ 1. number of women in Framingham Study who have died through last year from heart disease / number of women initially enrolled in Framingham Study ________ 2. number of women in Framingham Study who have died through last year from heart disease / number of person-years contributed through last year by women initially enrolled in Framingham Study ________ 3. number of women in town of Framingham who reported having heart disease in recent health survey / estimated number of women residents of Framingham during same period ________ 4. number of women in Framingham Study newly diagnosed with heart disease last year / number of women in Framingham Study without heart disease at beginning of same year ________ 5. number of women in State A newly diagnosed with heart disease in 2004 / estimated number of women living in State A on July 1, 2004 ________ 6. estimated number of women smokers in State A according to 2004 Behavioral Risk Factor Survey / estimated number of women living in State A on July 1, 2004 ________ 7. number of women in State A who reported having heart disease in 2004 health survey / estimated number of women smokers in State A according to 2004 Behavioral Risk Factor Survey

1. A;Incidence proportion: denominator is size of population at start of study, numerator is number of deaths among that population. 2. B;Incidence rate; denominator is person-years contributed by participants, numerator is number of death among that population. 3. C;Prevalence; numerator is all existing cases. 4. A; Incidence proportion; denominator is size of population at risk, numerator is number of new cases among that population. 5. B; Incidence rate; denominator is mid-year population, numerator is number of new cases among that population. 6. C; Prevalence; numerator is total number with attribute. 7. D; None of the above; this is a ratio (heart disease:smokers)

Three options... (outbreak)

1. Conduct an epidemiological study - Retrospective cohort study • Looking backwards to identify those who were exposed and non- exposed, ill and not ill) useful for outbreak in small well-defined population (ex. wedding reception) • Calculate attack rates, compare values in exposed group to unexposed • Measure association between exposure and disease - Case-control study • Looking at cases (outcome) and controls (no outcome) • Useful for unclear population (ex. outbreak in Guelph) • Compare odds ratio of exposure in case group to odds in control 2. Conduct a trace-back, trace-forward exercise - Trace-backsteps from outcome backwards to point of origin - Trace-forward: steps from point of origin to outcome 3. Perform an environmental investigation - Collecting samples from food, surfaces, environment - Submitted for microbiological and chemical testing - Look into operating procedures and make sure it was done properly

Three types of observational studies

1. Cross-sectional: subjects who are chosen without regard to exposure or outcome 2. Cohort: subjects chosen based on exposure status 3. Case-study: subjects chosen based on outcome status

The epidemiological approach

1. Determine association between exposure to some factor or characteristic and development of disease 2. Figure out of the relationship is causation or just correlation

The death rate per 100,000 for lung cancer is 7 among nonsmokers and 71 among smokers. The death rate per 100,000 for coronary thrombosis is 422 among nonsmokers and 599 among smokers. The prevalence of smoking in the population is 55%. (20pts) 1. What is the RR of dying of lung cancer for smokers versus nonsmokers? 2. What is the RR of dying of coronary thrombosis for smokers versus? 3. What is the attributable risk of disease due to smoking among individuals with lung cancer? 4. What is the attributable risk of disease due to smoking among individuals with coronary thrombosis? 5. What is the population etiologic fraction of lung cancer due to smoking? 6. What is the population etiologic fraction of coronary thrombosis due to smoking?

1. What is the RR of dying of lung cancer for smokers versus nonsmokers? a. RR = (71/100,000)/(7/100,000) = 10.143 2 What is the RR of dying of coronary thrombosis for smokers versus? a. RR = (599/100,000)/(422/100,000) = 1.419 2. What is the attributable risk of disease due to smoking among individuals with lung cancer? a. Etiologic fraction = [(71/100,000)-(7/100,000)] / (71/100,000) = 0.901 3. What is the attributable risk of disease due to smoking among individuals with coronary thrombosis? a. Etiologic fraction = [(599/100,000)-(422/100,000)] / (599/100,000) = 0.295 4. What is the population etiologic fraction of lung cancer due to smoking? a. Etiologic fraction = [(599/100,000)-(422/100,000)] / (599/100,000) = 0.295 5. What is the population etiologic fraction of lung cancer due to smoking? a. Prevalence of smoking = .55 b. Relative risk = 10.142 c. Population etiologic fraction = {[0.55(10.14-1)] / [0.55(10.14-1)+1]} * 100 = .834 = 83% 6. What is the population etiologic fraction of coronary thrombosis due to smoking? a. Prevalence of smoking = .55 b. Relative risk = 1.419 c. Population etiologic fraction = {[0.55(1.4-1)] / [0.55(1.4-1)+1]} * 100 = .180 = 18%

10. Use the following choices for the characteristics or features listed below: A. Incidence B. Prevalence ______ Measure of risk ______ Generally preferred for chronic diseases without clear date of onset ______ Used in calculation of risk ratio ______ Affected by duration of illness

10. A. Measure of risk B. Generally preferred for chronic diseases without clear date of onset A. Used in calculation of risk ratio B. Affected by duration of illness Incidence reflects new cases only; incidence proportion is a measure of risk. A risk ratio is simply the ratio of two incidence proportions. Prevalence reflects existing cases at a given point or period of time, so one does not need to know the date of onset. Prevalence is influenced by both incidence and duration of disease — the more cases that occur and the longer the disease lasts, the greater the prevalence at any given time. Ill Well Total Ate Ill Well Total Yes 50 3 53 No 4 22 26 Total 54 25 79

Within 10 days after attending a June wedding, an outbreak of cyclosporiasis occurred among attendees. Of the 83 guests and wedding party members, 79 were interviewed; 54 of the 79 met the case definition. The following two-by-two table shows consumption of wedding cake (that had raspberry filling) and illness status. 11. The fraction 54 / 79 is a/an: A. Food-specific attack rate B. Attack rate C. Incidence proportion D. Proportion 12. The fraction 50 / 54 is a/an: A. Attack rate B. Food-specific attack rate C. Incidence proportion D. Proportion 13. The fraction 50 / 53 is a/an: A. Attack rate B. Food-specific attack rate C. Incidence proportion D. Proportion 14. The best measure of association to use for these data is a/an: A. Food-specific attack rate B. Odds ratio C. Rate ratio D. Risk ratio 15. The best estimate of the association between wedding cake and illness is: A. 6.1 B. 7.7 C. 68.4 D. 83.7 E. 91.7 F. 94.3 16. The attributable proportion for wedding cake is: A. 6.1% B. 7.7% C. 68.4% D. 83.7% E. 91.7% F. 94.3%

11. B, C, D. The fraction 54 / 79 (see bottom row of the table) reflects the overall attack rate among persons who attended the wedding and were interviewed. Attack rate is a synonym for incidence proportion. 12. D. The fraction 50 / 54 (under the Ill column) is the proportion of case-patients who ate wedding cake. It is not an attack rate, because the denominator of an attack rate is the size of the population at the start of the period, not all cases. 13. A, B, C, D. The fraction 50 / 53 (see top row of table) is the proportion of wedding cake eaters who became ill, which is a food-specific attack rate. A food-specific attack rate is a type of attack rate, which in turn is synonymous with incidence proportion. 14. C. Investigators were able to interview almost everyone who attended the wedding, so incidence proportions (measure of risk) were calculated. When incidence proportions (risks) can be calculated, the best measure of association to use is the ratio of incidence proportions (risks), i.e., risk ratio. 15. A. The risk ratio is calculated as the attack rate among cake eaters divided by the attack rate among those who did not eat cake, or (50 / 53) / (4 / 26), or 94.3% / 15.4%, which equals 6.1. 16. D. The attributable proportion is calculated as the attack rate among cake eaters minus the attack rate among non-eaters, divided by the attack rate among cake eaters, or 94.3 - 15.4) / 94.3, which equals 83.7%. This attributable proportion means that 83.7% of the illness might be attributable to eating the wedding cake (note that some people got sick without eating cake, so the attributable proportion is not 100%).

Description: Three people became ill in June. Of them, two recovered in July and one in August. Five additional persons became ill in July; two of them recovered that month and three recovered in August. Return to text. 17. What is the prevalence of disease during July? A. 3 / 700 B. 4 / 700 C. 5 / 700 D. 8 / 700 18. What is the incidence of disease during July? A. 3 / 700 B. 4 / 700 C. 5 / 700 D. 8 / 700

17. D. A total of 8 cases are present at some time during the month of July. 18. C. Five new cases occurred during the month of July.

19. What is the following fraction? Number of children < 365 days of age who died in Country A in 2004/ Number of live births in Country A in 2004 A. Ratio B. Proportion C. Incidence proportion D. Mortality rate

19. A, D. The fraction shown is the infant mortality rate. It is a ratio, because all fractions are ratios. It is not a proportion because some of the children who died in early 2004 may have been born in late 2003, so some of those in the numerator are not in the denominator. Technically, the mortality rate for infants is the number of infants who died in 2004 divided by the estimated midyear population of infants, so the fraction shown is not a mortality rate in that sense. However, the fraction is known throughout the world as the infant mortality rate, despite the technical inaccuracy.

Use the following choices for Questions 2-4. E. Ratio F. Proportion G. Incidence proportion H. Mortality rate 2. ____ # women in Country A who died from lung cancer in 2004/ # women in Country A who died from cancer in 2004 3. ____ # women in Country A who died from lung cancer in 2004/ # women in Country A who died from breast cancer in 2004 4. ____ # women in Country A who died from lung cancer in 2004/ estimated # women living in Country A on July 1, 2004

2. A, B. All fractions are ratios. This fraction is also a proportion, because all of the deaths from lung cancer in the numerator are included in the denominator. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is not a mortality rate because the denominator is not the estimated midpoint population. 3. A. All fractions are ratios. This fraction is not a proportion, because lung cancer deaths in the numerator are not included in the denominator. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is not a mortality rate because the denominator is not the estimated midpoint population. 4. A, D. All fractions are ratios. This fraction is not a proportion, because some of the deaths occurred before July 1, so those women are not included in the calculation. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is a mortality rate because the denominator is the estimated midpoint population

20. Using only the data shown below for deaths attributed to Alzheimer's disease and to pneumonia/influenza, which measure(s) can be calculated? A. Proportionate mortality B. Cause-specific mortality rate C. Age-specific mortality rate D. Mortality rate ratio E. Years of potential life lost

20. E. The data shown in the table are numbers of deaths. No denominators are provided from which to calculate rates. Neither is the total number of deaths given, so proportionate mortality cannot be calculated. However, calculation of potential life lost need only the numbers of deaths by age, as shown.

21. Which of the following mortality rates use the estimated total mid-year population as its denominator? A. Age-specific mortality rate B. Sex-specific mortality rate C. Crude mortality rate D. Cause-specific mortality rate

21. C, D. Only crude and cause-specific mortality rates use the estimated total mid-year population as its denominator. The denominator for an age-specific mortality rate is the estimated mid-year size of that particular age group. The denominator for a sexspecific mortality rate is the estimated mid-year male or female population.

22. What is the following fraction? Number of deaths due to septicemia among men aged 65-74 years in 2004/ Estimated number of men aged 65-74 years alive on July 1, 2004 A. Age-specific mortality rate B. Age-adjusted mortality rate C. Cause-specific mortality rate D. Sex-specific mortality rate

22. A, C, D. The fraction is the mortality rate due to septicemia (cause) among men (sex) aged 65-74 years (age). Age-specific mortality rates are narrowly defined (in this fraction, limited to 10 years of age), so are generally valid for comparing two populations without any adjustment.

23. Vaccine efficacy measures are: A. The proportion of vaccinees who do not get the disease B. 1 - the attack rate among vaccinees C. The proportionate reduction in disease among vaccinees D. 1 - disease attributable to the vaccine

23. C. Vaccine efficacy measures the proportionate reduction in disease among vaccinees.

24. To study the causes of an outbreak of aflatoxin poisoning in Africa, investigators conducted a case-control study with 40 case-patients and 80 controls. Among the 40 poisoning victims, 32 reported storing their maize inside rather than outside. Among the 80 controls, 20 stored their maize inside. The resulting odds ratio for the association between inside storage of maize and illness is: A. 3.2 B. 5.2 C. 12.0 D. 33.3

24. C. The results of this study could be summarized in a two-by-two table as follows: Strored Maize inside Cases Controls Total Yes a = 32 c = 20 52 No b = 8 d = 60 68 Total 40 80 120 The odds ratio is calculated as ad/bc, or (32 x 60) / (8 x 20), which equals 1,920 /160 or 12.0.

25. The crude mortality rate in Community A was higher than the crude mortality rate in Community B, but the age-adjusted mortality rate was higher in Community B than in Community A. This indicates that: A. Investigators made a calculation error B. No inferences can be made about the comparative age of the populations from these data C. The population of Community A is, on average, older than that of Community B D. The population of Community B is, on average, older than that of Community A

25. C. The crude mortality rate reflects the mortality experience and the age distribution of a community, whereas the age-adjusted mortality rate eliminates any differences in the age distribution. So if Community A's age-adjusted mortality rate was lower than its crude rate, that indicates that its population is older.

7. Many of the students at the boarding school, including 6 just coming down with varicella, went home during the Thanksgiving break. About 2 weeks later, 4 siblings of these 6 students (out of a total of 10 siblings) developed varicella. The secondary attack rate among siblings was, therefore,: A. 4 / 6 B. 4 / 10 C. 4 / 16 D. 6 / 10

7. B. The secondary attack rate is calculated as the number of cases among contacts (4) divided by the number of contacts (10).

Ratios

A dimensionless, unit-less comparison of one count to another (ex. 1:4, 2:70)

Proportion Definition of proportion

A proportion is the comparison of a part to the whole. It is a type of ratio in which the numerator is included in the denominator. Measures of Risk You might use a proportion to describe what fraction of clinic patients tested positive for HIV, or what percentage of the population is younger than 25 years of age. A proportion may be expressed as a decimal, a fraction, or a percentage.

There was a flu vaccine clinic held and the based on the dates of birth the following ages of the attendees were 63, 41, 53, 62, 41, 57, 62, 41, 65, 70, 38,

A. Arrange in ascending order: (38, 41,41, 41,53,57,62,62,63,65,70) /11= mean B. Identify a. Mode =41 b. Mean=53.909 c. Median =57

Unless otherw ise instructed, choose ALL correct choices for each question. 1. Which of the following are frequency measures? A. Birth rate B. Incidence C. Mortality rate D. Prevalence Use the following choices for Questions 2-4. E. Ratio F. Proportion G. Incidence proportion H. Mortality rate 2. ____ # women in Country A who died from lung cancer in 2004 # women in Country A who died from cancer in 2004 3. ____ # women in Country A who died from lung cancer in 2004 # women in Country A who died from breast cancer in 2004 4. ____ # women in Country A who died from lung cancer in 2004 estimated # women living in Country A on July 1, 2004 5. All proportions are ratios, but not all ratios are proportions. A. True B. False 6. In a state that did not require varicella (chickenpox) vaccination, a boarding school experienced a prolonged outbreak of varicella among its students that began in September and continued through December. To calculate the probability or risk of illness among the students, which denominator would you use? A. Number of susceptible students at the ending of the period (i.e., June) B. Number of susceptible students at the midpoint of the period (late October/early November) C. Number of susceptible students at the beginning of the period (i.e., September) D. Average number of susceptible students during outbreak Measures of Risk Page 3-57 7. Many of the students at the boarding school, including 6 just coming down with varicella, went home during the Thanksgiving break. About 2 weeks later, 4 siblings of these 6 students (out of a total of 10 siblings) developed varicella. The secondary attack rate among siblings was, therefore,: A. 4 / 6 B. 4 / 10 C. 4 / 16 D. 6 / 10 8. Investigators enrolled 100 diabetics without eye disease in a cohort (follow-up) study. The results of the first 3 years were as follows: Year 1: 0 cases of eye disease detected out of 92; 8 lost to follow-up Year 2: 2 new cases of eye disease detected out of 80; 2 had died; 10 lost to follow-up Year 3: 3 new cases of eye disease detected out of 63; 2 more had died; 13 more lost to follow-up The person-time incidence rate is calculated as: A. 5 / 100 B. 5 / 63 C. 5 / 235 D. 5 / 250 9. The units for the quantity you calculated in Question 8 could be expressed as: A. cases per 100 persons B. percent C. cases per person-year D. cases per person per year 10. Use the following choices for the characteristics or features listed below: A. Incidence B. Prevalence ______ Measure of risk ______ Generally preferred for chronic diseases without clear date of onset ______ Used in calculation of risk ratio ______ Affected by duration of illness Measures of Risk Page 3-58 Use the following information for Questions 11-15. Within 10 days after attending a June wedding, an outbreak of cyclosporiasis occurred among attendees. Of the 83 guests and wedding party members, 79 were interviewed; 54 of the 79 met the case definition. The following two-by-two table shows consumption of wedding cake (that had raspberry filling) and illness status. Ill Well Total Ate wedding cake? Yes 50 3 53 No 4 22 26 Total 54 25 79 Source: Ho AY, Lopez AS, Eberhart MG, et al. Outbreak of cyclosporiasis associated with imported raspberries, Philadelphia, Pennsylvania, 2000. Emerg Infect Dis 2002;l8:783-6. 11. The fraction 54 / 79 is a/an: A. Food-specific attack rate B. Attack rate C. Incidence proportion D. Proportion 12. The fraction 50 / 54 is a/an: A. Attack rate B. Food-specific attack rate C. Incidence proportion D. Proportion 13. The fraction 50 / 53 is a/an: A. Attack rate B. Food-specific attack rate C. Incidence proportion D. Proportion 14. The best measure of association to use for these data is a/an: A. Food-specific attack rate B. Odds ratio C. Rate ratio D. Risk ratio 15. The best estimate of the association between wedding cake and illness is: A. 6.1 B. 7.7 C. 68.4 D. 83.7 E. 91.7 F. 94.3 Measures of Risk Page 3-59 16. The attributable proportion for wedding cake is: A. 6.1% B. 7.7% C. 68.4% D. 83.7% E. 91.7% F. 94.3% Use the following diagram for Questions 17 and 18. Assume that the horizontal lines in the diagram represent duration of illness in 8 different people, out of a community of 700. Description: Three people became ill in June. Of them, two recovered in July and one in August. Five additional persons became ill in July; two of them recovered that month and three recovered in Augus 17. What is the prevalence of disease during July? A. 3 / 700 B. 4 / 700 C. 5 / 700 D. 8 / 700 18. What is the incidence of disease during July? A. 3 / 700 B. 4 / 700 C. 5 / 700 D. 8 / 700 19. What is the following fraction? Number of children < 365 days of age who died in Country A in 2004 Number of live births in Country A in 2004 A. Ratio B. Proportion C. Incidence proportion D. Mortality rate Measures of Risk Page 3-60 20. Using only the data shown below for deaths attributed to Alzheimer's disease and to pneumonia/influenza, which measure(s) can be calculated? A. Proportionate mortality B. Cause-specific mortality rate C. Age-specific mortality rate D. Mortality rate ratio E. Years of potential life lost Table 3.16 Number of Deaths Due to Alzheimer's Disease and Pneumonia/Influenza — United States, 2002 Age Group (years) Alzheimer's disease Pneumonia/ Influenza < 5 0 373 5-14 1 91 15-24 0 167 <34 32 345 35-44 12 971 45-54 52 1,918 55-64 51 2,987 65-74 3,602 6,847 75-84 20,135 19,984 85+ 34,552 31,995 Total 58,866 65,681 Source: Kochanek KD, Murphy SL, Anderson RN, Scott C. Deaths: Final data for 2002. National vital statistics reports; vol 53, no 5. Hyattsville, Maryland: National Center for Health Statistics, 2004. 21. Which of the following mortality rates use the estimated total mid-year population as its denominator? A. Age-specific mortality rate B. Sex-specific mortality rate C. Crude mortality rate D. Cause-specific mortality rate 22. What is the following fraction? Number of deaths due to septicemia among men aged 65-74 years in 2004 Estimated number of men aged 65-74 years alive on July 1, 2004 A. Age-specific mortality rate B. Age-adjusted mortality rate C. Cause-specific mortality rate D. Sex-specific mortality rate Measures of Risk Page 3-61 23. Vaccine efficacy measures are: A. The proportion of vaccinees who do not get the disease B. 1 - the attack rate among vaccinees C. The proportionate reduction in disease among vaccinees D. 1 - disease attributable to the vaccine 24. To study the causes of an outbreak of aflatoxin poisoning in Africa, investigators conducted a case-control study with 40 case-patients and 80 controls. Among the 40 poisoning victims, 32 reported storing their maize inside rather than outside. Among the 80 controls, 20 stored their maize inside. The resulting odds ratio for the association between inside storage of maize and illness is: A. 3.2 B. 5.2 C. 12.0 D. 33.3 25. The crude mortality rate in Community A was higher than the crude mortality rate in Community B, but the age-adjusted mortality rate was higher in Community B than in Community A. This indicates that: A. Investigators made a calculation error B. No inferences can be made about the comparative age of the populations from these data C. The population of Community A is, on average, older than that of Community B D. The population of Community B is, on average, older than that of Community A

Answers to Self-Assessment Quiz 1. A, B, C, D. Frequency measures of health and disease include those related to birth, death, and morbidity (incidence and prevalence). 2. A, B. All fractions are ratios. This fraction is also a proportion, because all of the deaths from lung cancer in the numerator are included in the denominator. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is not a mortality rate because the denominator is not the estimated midpoint population. 3. A. All fractions are ratios. This fraction is not a proportion, because lung cancer deaths in the numerator are not included in the denominator. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is not a mortality rate because the denominator is not the estimated midpoint population. 4. A, D. All fractions are ratios. This fraction is not a proportion, because some of the deaths occurred before July 1, so those women are not included in the calculation. It is not an incidence proportion, because the denominator is not the size of the population at the start of the period. It is a mortality rate because the denominator is the estimated midpoint population. 5. A. All fractions, including proportions, are ratios. But only ratios in which the numerator is included in the denominator is a proportions. 6. C. Probability or risk are estimated by the incidence proportion, calculated as the number of new cases during a specified period divided by the size of the population at the start of that period. 7. B. The secondary attack rate is calculated as the number of cases among contacts (4) divided by the number of contacts (10). 8. D. During year 1, 92 returning patients contributed 92 person-years; 8 patients lost to follow-up contributed 8 x ½ or 4 years, for a total of 96. During the second year, 78 disease-free patients contributed 78 person-years, plus ½ years for the 2 with newly diagnosed eye disease, the 2 who had died, and the 10 lost to follow-up (all events are assumed to have occurred randomly during the year, or an average, at the half-year point), for a total of 78 + 14 x ½ years, for another 85 years. During the third year, returning healthy patients contributed 60 years; the 3 with eye disease, the 4 who died, and the 11 lost to follow-up contributed 18 x ½ years or 9 years, for a total of 69 years during the 3rd year. The total person-years is therefore 96 + 85 + 69 = 250 person-years. 9. C, D. The person-time rate presented in Question 8 should be reported as 5 cases per 250 person-years. Usually person-time rates are expressed per 1,000 or 10,000 or 100,000, depending on the rarity of the disease, so the rate in Question 8 could Measures of Risk Page 3-63 be expressed as 2 cases per 100 person-years of follow-up. One could express this more colloquially as 2 new cases of eye disease per 100 diabetics per year. 10. A. Measure of risk B. Generally preferred for chronic diseases without clear date of onset A. Used in calculation of risk ratio B. Affected by duration of illness Incidence reflects new cases only; incidence proportion is a measure of risk. A risk ratio is simply the ratio of two incidence proportions. Prevalence reflects existing cases at a given point or period of time, so one does not need to know the date of onset. Prevalence is influenced by both incidence and duration of disease — the more cases that occur and the longer the disease lasts, the greater the prevalence at any given time. Ill Well Total Ate wedding cake? Yes 50 3 53 No 4 22 26 Total 54 25 79 11. B, C, D. The fraction 54 / 79 (see bottom row of the table) reflects the overall attack rate among persons who attended the wedding and were interviewed. Attack rate is a synonym for incidence proportion. 12. D. The fraction 50 / 54 (under the Ill column) is the proportion of case-patients who ate wedding cake. It is not an attack rate, because the denominator of an attack rate is the size of the population at the start of the period, not all cases. 13. A, B, C, D. The fraction 50 / 53 (see top row of table) is the proportion of wedding cake eaters who became ill, which is a food-specific attack rate. A food-specific attack rate is a type of attack rate, which in turn is synonymous with incidence proportion. 14. C. Investigators were able to interview almost everyone who attended the wedding, so incidence proportions (measure of risk) were calculated. When incidence proportions (risks) can be calculated, the best measure of association to use is the ratio of incidence proportions (risks), i.e., risk ratio. 15. A. The risk ratio is calculated as the attack rate among cake eaters divided by the attack rate among those who did not eat cake, or (50 / 53) / (4 / 26), or 94.3% / 15.4%, which equals 6.1. 16. D. The attributable proportion is calculated as the attack rate among cake eaters minus the attack rate among non-eaters, divided by the attack rate among cake eaters, or 94.3 - 15.4) / 94.3, which equals 83.7%. This attributable proportion means that 83.7% of the illness might be attributable to eating the wedding cake (note that some people got sick without eating cake, so the attributable proportion is not 100%). 17. D. A total of 8 cases are present at some time during the month of July. Measures of Risk Page 3-64 18. C. Five new cases occurred during the month of July. 19. A, D. The fraction shown is the infant mortality rate. It is a ratio, because all fractions are ratios. It is not a proportion because some of the children who died in early 2004 may have been born in late 2003, so some of those in the numerator are not in the denominator. Technically, the mortality rate for infants is the number of infants who died in 2004 divided by the estimated midyear population of infants, so the fraction shown is not a mortality rate in that sense. However, the fraction is known throughout the world as the infant mortality rate, despite the technical inaccuracy. 20. E. The data shown in the table are numbers of deaths. No denominators are provided from which to calculate rates. Neither is the total number of deaths given, so proportionate mortality cannot be calculated. However, calculation of potential life lost need only the numbers of deaths by age, as shown. 21. C, D. Only crude and cause-specific mortality rates use the estimated total mid-year population as its denominator. The denominator for an age-specific mortality rate is the estimated mid-year size of that particular age group. The denominator for a sexspecific mortality rate is the estimated mid-year male or female population. 22. A, C, D. The fraction is the mortality rate due to septicemia (cause) among men (sex) aged 65-74 years (age). Age-specific mortality rates are narrowly defined (in this fraction, limited to 10 years of age), so are generally valid for comparing two populations without any adjustment. 23. C. Vaccine efficacy measures the proportionate reduction in disease among vaccinees. 24. C. The results of this study could be summarized in a two-by-two table as follows: Cases Controls Total Stored maize inside? Yes a = 32 c = 20 52 No b = 8 d = 60 68 Total 40 80 120 The odds ratio is calculated as ad/bc, or (32 x 60) / (8 x 20), which equals 1,920 / 160 or 12.0. 25. C. The crude mortality rate reflects the mortality experience and the age distribution of a community, whereas the age-adjusted mortality rate eliminates any differences in the age distribution. So if Community A's age-adjusted mortality rate was lower than its crude rate, that indicates that its population is older.

6. In a state that did not require varicella (chickenpox) vaccination, a boarding school experienced a prolonged outbreak of varicella among its students that began in September and continued through December. To calculate the probability or risk of illness among the students, which denominator would you use? A. Number of susceptible students at the ending of the period (i.e., June) B. Number of susceptible students at the midpoint of the period (late October/early November) C. Number of susceptible students at the beginning of the period (i.e., September) D. Average number of susceptible students during outbreak

C. Probability or risk are estimated by the incidence proportion, calculated as the number of new cases during a specified period divided by the size of the population at the start of that period.

8. Investigators enrolled 100 diabetics without eye disease in a cohort (follow-up) study. The results of the first 3 years were as follows: Year 1: 0 cases of eye disease detected out of 92; 8 lost to follow-up Year 2: 2 new cases of eye disease detected out of 80; 2 had died; 10 lost to follow-up Year 3: 3 new cases of eye disease detected out of 63; 2 more had died; 13 more lost to follow-up The person-time incidence rate is calculated as: A. 5 / 100 B. 5 / 63 C. 5 / 235 D. 5 / 250

D. During year 1, 92 returning patients contributed 92 person-years; 8 patients lost to follow-up contributed 8 x ½ or 4 years, for a total of 96. During the second year, 78 disease-free patients contributed 78 person-years, plus ½ years for the 2 with newly diagnosed eye disease, the 2 who had died, and the 10 lost to follow-up (all events are assumed to have occurred randomly during the year, or an average, at the half-year point), for a total of 78 + 14 x ½ years, for another 85 years. During the third year, returning healthy patients contributed 60 years; the 3 with eye disease, the 4 who died, and the 11 lost to follow-up contributed 18 x ½ years or 9 years, for a total of 69 years during the 3rd year. The total person-years is therefore 96 + 85 + 69 = 250 person-years.

A commonly used epidemiologic ratio: death-to-case ratio

Death-to-case ratio is the number of deaths attributed to a particular disease during a specified period divided by the number of new cases of that disease identified during the same period. It is used as a measure of the severity of illness: the death-to-case ratio for rabies is close to 1 (that is, almost everyone who develops rabies dies from it), whereas the death-to-case ratio for the common cold is close to 0. For example, in the United States in 2002, a total of 15,075 new cases of tuberculosis were reported.3 During the same year, 802 deaths were attributed to tuberculosis. The tuberculosis death-tocase ratio for 2002 can be calculated as 802 / 15,075. Dividing both numerator and denominator by the numerator yields 1 death per 18.8 new cases. Dividing both numerator and denominator by the denominator (and multiplying by 10n = 100) yields 5.3 deaths per 100 new cases. Both expressions are correct. Note that, presumably, many of those who died had initially contracted tuberculosis years earlier. Thus many of the 802 in the numerator are not among the 15,075 in the denominator. Therefore, the death-to-case ratio is a ratio, but not a proportion

A serological test is being devised to detect a hypothetical chronic disease. Three hundred individuals were referred to a laboratory for testing. One hundred diagnosed cases were among the 300. A serological test yielded 200 positives, of which one-fourth were true positives (20pts)

Disease Positive Disease Negative Total Test Positive 50 150 200 Test Negative 50 50 100 Total 100 200 300 Sensitivity= a/a+c 50/50+50=50/100=0.5 or 50% Specificity= d/b+d 50/150+50=50/200=0.25 or 25% Positive Predicative (+) Value= PPV=/+b 50/50=150=0.25 or 25% Negative Predictive (-) Value= NPV=50/50+50= 0.5 or 50% Prevalence=? Accuracy=?

Exercise 3.6 Use the HIV data in Table 3.9 to answer the following questions: 1. What is the HIV-related mortality rate, all ages? 2. What is the HIV-related mortality rate for persons under 65 years? 3. What is the HIV-related YPLL before age 65? 4. What is the HIV-related YPLL65 rate? 5. Create a table comparing the mortality rates and YPLL for leukemia and HIV. Which measure(s) might you prefer if you were trying to support increased funding for leukemia research? For HIV research?

Ex ercise 3.6 1. HIV-related mortality rate, all ages = (# deaths from HIV / estimated population, 2002) x 100,000 = (14,095 / 288,357,000) x 100,000 = 4.9 HIV deaths per 100,000 population 2. HIV-related mortality rate for persons under 65 years = (# deaths from HIV among <65 years year-olds / estimated population < 65 years, 2002) x 100,000 = (12 + 25 + 178 + 1,839 + 5,707 + 4,474 + 1,347 / 19,597 + 41,037 + 40,590 +39,928 + 44,917 + 40,084 + 26,602) x 100,000 = 13,582 / 252,755,000 x 100,000 Measures of Risk Page 3-54 = 5.4 HIV deaths per 100,000 persons under age 65 years 3. HIV-related YPLL before age 65 Deaths and years of potential life lost attributed to HIV by age group — United States, 2002 Column 1 Age Group (years) Column 2 Deaths Column 3 Age Midpoint Column 4 Years to 65 Column 5 YPLL 0-4 12 2.5 62.5 750 5-10 25 10 55 1,375 15-24 178 20 45 8,010 25-34 1,839 30 35 64,365 35-44 5,707 40 25 142,675 45-54 4,474 50 15 67,110 55-64 1,347 60 5 6,735 65+ 509 - - - Not stated 4 - - - Total 14,095 291,020 4. HIV-related YPLL65 rate YPLL65 rate = (291,020 / 252,755,000) x 1,000 = 1.2 YPLL per 1,000 population under age 65. 5. Compare mortality rates and YPLL for leukemia and HIV Leukemia HIV # cause-specific deaths, all ages 21,498 14,095 cause-specific death rate, all ages (per 100,000 pop) 7.5 4.9 # deaths, < 65 years 6,221 13,582 death rate, < 65 years 2.5 5.4 YPLL65 117,033 291,020 YPLL65 rate 0.5 1.2 An advocate for increased leukemia research funding might use the first two measures, which shows that leukemia is a larger problem in the entire population. An advocate for HIV funding might use the last four measures, since they highlight HIV deaths among younger persons.

Exercise 3.3 In 2001, a total of 15,555 homicide deaths occurred among males and 4,753 homicide deaths occurred among females. The estimated 2001 midyear populations for males and females were 139,813,000 and 144,984,000, respectively. 1. Calculate the homicide-related death rates for males and for females. 2. What type(s) of mortality rates did you calculate in Question 1? 3. Calculate the ratio of homicide-mortality rates for males compared to females. 4. Interpret the rate you calculated in Question 3 as if you were presenting information to a policymaker.

Exercise 3.3 CDC 1. Homicide-related death rate (males) = (# homicide deaths among males ⁄ male population) × 100,000 = 15,555 ⁄ 139,813,000 × 100,000 = 11.1 homicide deaths ⁄ 100,000 population among males Homicide-related death rate (females) = (# homicide deaths among females ⁄ female population) × 100,000 = 4,753 ⁄ 144,984,000 × 100,000 = 3.3 homicide deaths ⁄ 100,000 population among females 2. These are cause- and sex-specific mortality rates. 3. Homicide-mortality rate ratio = homicide death rate (males) ⁄ homicide death rate (females) = 11.1 ⁄ 3.3 = 3.4 to 1 = (see below, which is the answer to question 4). 4. Because the homicide rate among males is higher than the homicide rate among females, specific intervention programs need to target males and females differently.

Exercise 3.4 Table 3.7 provides the number of reported cases of diphtheria and the number of diphtheria-associated deaths in the United States by decade. Calculate the death-to-case ratio by decade. Describe the data in Table 3.7, including your results. Table 3.7 Number of Cases and Deaths from Diphtheria by Decade — United States, 1940-1949 143,497 11,228 7.82 1950-1959 23,750 1,710 7.20 1960-1969 3,679 390 10.60 1970-1979 1,956 90 1980-1989 27 3 1990-1999 22 5 Data Sources: Centers for Disease Control and Prevention. Summary of notifiable diseases,

Exercise 3.4 Decade Number ofNew Cases Number ofDeaths Death-to-CaseRatio Decade #Cases #Death #dth to case 1940-1949 143,497 11,228 7.82 (Given) 1950-1959 23,750 1,710 7.20 1960-1969 3,679 390 10.60 1970-1979 1,956 90 4.60 1980-1989 27 3 11.11 1990-1999 22 5 22.72 The number of new cases and deaths from diphtheria declined dramatically from the 1940s through the 1980s, but remained roughly level at very low levels in the 1990s. The death-to-case ratio was actually higher in the 1980s and 1990s than in 1940s and 1950s. From these data one might conclude that the decline in deaths is a result of the decline in cases, that is, from prevention, rather than from any improvement in the treatment of cases that do occur.

Using the data in Table 3.8, calculate the missing proportionate mortalities for persons ages 25—44 years for diseases of the heart and assaults (homicide). Diseases of heart Malignant neoplasms Cerebrovascular disease Chronic lower respiratory diseases Accidents (unintentional injuries) Diabetes mellitus Influenza & pneumonia Alzheimer's disease Nephritis, nephrotic syndrome, nephrosis Septicemia Intentional self-harm (suicide) Chronic liver disease and cirrhosis Assault (homicide) HIV disease All other

Exercise 3.5 Proportionate mortality for diseases of heart, 25-44 years = (# deaths from diseases of heart ⁄ # deaths from all causes) × 100 = 16,283 ⁄ 128,294 × 100 = 12.6% Proportionate mortality for assault (homicide), 25-44 years = (# deaths from assault (homicide) ⁄ # deaths from all causes) × 100 = 7,367 ⁄ 128,924 × 100 = 5.7%

"Exercise 3.7 Table 3.14 illustrates lung cancer mortality rates for persons who conti to smoke and for smokers who had quit at the time of follow-up in one" "the classic studies of smoking and lung cancer conducted in Great Britain. Using the data in Table 3.14, calculate the following: 1. Rate ratio comparing current smokers with nonsmokers 2. Rate ratio comparing ex-smokers who quit at least 20 years ago with nonsmokers 3. What are the public health implications of these findings?" "nued of" "Table 3.14 Number and Rate (Per 1,000 Person-years) of Lung Cancer Deaths for Current Smokers and Ex-smokers by Years Since Quitting, Physician Cohort Study — Great Britain, 1951-1961" "Cigarette smoking status" "Lung cancer deaths" "Rate per 1000 person-years" "Rate Ratio" "Current smokers" 133 1.30 " " "For ex-smokers, years since quitting: <5 years" 5 0.67 9.6 "5-9 years" 7 0.49 7.0 "10-19 years" 3 0.18 2.6 "20+ years" 2 0.19 " " "Nonsmokers" 3 0.07 "1.0 (reference group)"

Exercise 3.7 1. Rate ratio comparing current smokers with nonsmokers= rate among current smokers ⁄ rate among non-smokers= 1.30 ⁄ 0.07= 18.6 2. Rate ratio comparing ex-smokers who quit at least 20 years ago with nonsmokers= rate among ex-smokers ⁄ rate among non-smokers= 0.19 ⁄ 0.07= 2.7 3. The lung cancer rate among smokers is 18 times as high as the rate among non-smokers. Smokers who quit can lower their rate considerably, but it never gets back to the low level seen in never-smokers. So the public health message might be, "If you smoke, quit. But better yet, don't start."

Exercise 3.8 Calculate the odds ratio for the tuberculosis data in Table 3.12. Would you say that your odds ratio is an accurate approximation of the risk ratio? (Hint: The more common the disease, the further the odds ratio is from the risk ratio.)

Exercise 3.8 Odds ratio = ad ⁄ bc = (28 × 133) ⁄ (129 × 4)= 7.2 The odds ratio of 7.2 is somewhat larger (18% larger, to be precise) than the risk ratio of 6.1. Whether that difference is "reasonable" or not is a judgment call. The odds ratio of 7.2 and the risk ratio of 6.1 both reflect a very strong association between prison wing and risk of developing tuberculosis.

Component-cause model

Factors that influence whether or not this combo is sufficient - Susceptibility - Genetics - Social determinants - Clinical factors - Cultural influence - Psychological factors

Systematic random sampling

First subject drawn using formal random process. Sampling interval (source population size / required sample size) used to choose every nth subject after the first one. (Sampling interval = 5, then every 5th person after the first is chosen).

*True rates have

ITC

Steps 9 & 10 (outbreak)

Implement measures - Control measures implemented early as long as they are evidence-based - Directed against agent, source, mode of transmission, portal of entry, host - Monitor impact of these measures • Is number of cases decreasing? • Is epidemic curve changing? • What are the tests reporting? • Are people complying with measures? Disseminate - Written report of the findings of the investigation (actions, performance, medical/legal issues, events) - Important so investigators learn from outbreak to minimize and prevent future outbreak

R ate Definition of rate

In epidemiology, a rate is a measure of the frequency with which an event occurs in a defined population over a specified period of time. Because rates put disease frequency in the perspective of the size of the population, rates are particularly useful for comparing disease frequency in different locations, at different times, or among different groups of persons with potentially different sized populations; that is, a rate is a measure of risk. To a non-epidemiologist, rate means how fast something is happening or going.

number of women in Framingham Study who have died through last year from heart disease/ number of women initially enrolled in Framingham Study

Incidence Proportion: A; denominator is size of population at start of study, numerator is number of deaths among that population

number of women in Framingham Study who have died through last year from heart disease/ number of person-years contributed through last year by women initially enrolled in Framingham Study

Incidence Rate: B; denominator is person-years contributed by participants, numerator is number of death among that population.

MORBIDITY Point prevalence

Number of current cases (new and preexisting) at a specified point in time/Population at the same specified point in time

MORBIDITY Period prevalence

Number of current cases (new and preexisting) over a specified period of time/Average or mid-interval population

MOrbidity Secondary attack rate

Number of new cases among contacts./Total number of contacts

MOrbidity Incidence proportion (or attack rate or risk)

Number of new cases of disease during specified time interval/Population at start of time interval

MORBIDITY Incidence rate (or person-time rate)

Number of new cases of disease during specified time interval/Summed person-years of observation or average population during time interval

Method for calculating a proportion

Number of persons or events with a particular characteristic /Total number of persons or events, of which the numerator is a subset x 10^n For a proportion, 10^n is usually 100 (or n = 2) and is often expressed as a percentage. Example A: Calculate the proportion of men in the NHANES follow-up study who were diabetics. Numerator = 189 diabetic men Denominator = Total number of men = 189 + 3,151 = 3,340 Proportion = (189 / 3,340) x 100 = 5.66% Example B: Calculate the proportion of deaths among men. Numerator = deaths in men = 100 deaths in diabetic men + 811 deaths in nondiabetic men = 911 deaths in men Notice that the numerator (911 deaths in men) is a subset of the denominator. Denominator = all deaths = 911 deaths in men + 72 deaths in diabetic women + 511 deaths in nondiabetic women = 1,494 deaths Proportion = 911 / 1,494 = 60.98% = 61% Your Turn: What proportion of all study participants were men? (Answer = 45.25%)

Prevalence

PAR = (a+c)/n - (c/c+d) Rate: PAR = ((a1+a0) / (t1+t0)) - a0/t0 Population attributable fraction - Proportion of entire population outcome that of caused by exposure Prevalence: PAF = PAR / (a+c)/n Rate: PAF = PAR / ((a1+a0)/(t1+t0)) Measures of association: RR, OR, IRR Measures of effect: RD, APe, PAR, PAF

number of women in town of Framingham who reported having heart disease in recent health survey / estimated number of women residents of Framingham during same period

Prevelance: C; numerator is all existing cases.

number of women in State A who died from lung cancer in 2004/number of women in State A who died from cancer (all types) in 2004

Proportion

number of women in State A who died from heart disease in 2004/estimated number of women living in State A on July 1, 2004

RATE

number of women in State A who died from heart disease in 2004/number of women in State A who died from cancer in 2004

RATIO

number of women in State A who died from lung cancer in 2004/estimated revenue (in dollars) in State A from cigarette sales in 2004

RATIO

Reporting

REFLECT: reporting guidelines for randomized controlled trials for livestock and food safety

Chapter 5

Screening and diagnostic tests • Must distinguish between those with disease and those without disease to understand the disease and provide proper care • Tests must be able to accurately do this!

Calculate the Standard Deviation for following incubation period of Hepatitis C: 30, 55, 75, 35, and 85 days was investigated

Standard Deviation=?

When the serum cholesterol levels of 4,462 men were measured, the mean cholesterol level was 213, with a standard deviation of 42. Calculate the standard error of the mean for the serum cholesterol level of the mean Standard Error of the Mean

Standard error of the mean = 42 divided by the square root of 4,462 = 0.629

5. All proportions are ratios, but not all ratios are proportions. A. True B. False

TRUE 5. A. All fractions, including proportions, are ratios. But only ratios in which the numerator is included in the denominator is a proportions.

The association between job-related exposure to welding fumes and chronic obstructive pulmonary disease (COPD) was explored in a case controlled study. The following data were reported for 399 COPD patients: 37 currently employed as welders, the remainder had no occupational exposure. Among the 800 controls, 48 were employed as welders (20 pts).

Table of COPD by Welding Exposure Welding No Welding Yes Total COPD No 752 362 1114 COPD Yes 48 37 85 Total 800 399 1199 Chronic obstructive lung disease and occupational exposure: Case control study was carried out to see the association of the exposure of welding fumes and chronic obstructive lung disease. The data lists that among 399 people with chronic obstructive lung disease 37 have exposure to welding fumes. Whereas, among 800 controls 48 employees are welders. Number of patients with Chronic obstructive lung disease=399 Number of individuals with Chronic obstructive lung disease working as welders=37 Number of controls: 800 Number of controls working as welders: 48 Number of patients without exposure to welding fumes=399-37=362 Number of controls without exposure to welding fumes=800-48=752 Odds Ratio= Odds exposure of cases/ Odds exposure of controls Odds Ratio: odds exposure of cases/ods exposure of controls OR=(OR)=(AD)/(BC) A=total # of incidence with exposure=37 B= total number of unexposure non cases=48 C=total# uneposed incidences 362 D= #unexposed non cases=752 OR=37x752/362x48=27824/17376=1.60 1.60

Morbidity and mortality Morbidity

a diseased state, considers the burden of the disease, such as sickness or impairment

Exclusion criteria

a set of characteristics that, if possessed, will prevent a potential subject from participating in the study

Counts

a simple representation of a number of events/outcomes (ex. 1, 3, 75, 2000)

Illness

a subjective state of the person who feels aware of not being well

Power

ability of a study to detect differences between groups when a real difference exists (typically 80%). 1 - power = probability of not detecting a difference when there is a difference. Zβ = -0.84

Specificity

ability of test to correctly identify those who do not have the disease

Sensitivity

ability of test to correctly identify those who have the disease

Cutoff level

above this = disease, below this = no disease

Primary prevention

action taken to prevent disease in healthy person (immunization, reduce exposure)

Sampling unit

actual unit of measurement (person, household, farm)

Non-differential

all groups equally likely to be accurately/inaccurately assigned, more findings bias to null hypothesis (harder to find difference)

Test

any critical examination, observation, or evaluation used to determine health and disease outcomes and potential exposures

Morbidity

any departure, subjective or objective, from a state of physiological or psychological wellbeing. In practice, morbidity encompasses disease, injury, and disability. In addition, although for this lesson the term refers to the number of persons who are ill, it can also be used to describe the periods of illness that these persons experienced, or the duration of these illnesses.

Bias

any systematic error in the design, conduct, or analysis of a study that results in a mistaken estimate of an exposure's effect on the disease risk

Healthy worker effect

applies when studying a population of actively employed individuals certain level of health required to work, those who can join are of better health than those who cannot

Misclassification bias

assigning participants to wrong exposure or outcome status - Tests for diagnosis are inaccurate - Unclear questions - Unaware participants - Old/incomplete records

Chapter 14

association →causation • Is there an association between exposure to a factor and disease outcome? • Is the observed association casual?

- Approximate denominator

average between initial NAR and final NAR multiplied by an internal time component (ITC, equal to or less than the time between initial and final measurements)

On the basis if the RR and etiologic fractions associated with smoking from lung cancer and coronary thrombosis, which one of the following statements is most likely to be correct? a) Smoking seems much more likely to be causally related to coronary thrombosis than to lung cancer. b) Smoking seems much more likely to be causally related to lung cancer than to coronary thrombosis c) Smoking seems to be equally causally related to both lung cancer and coronary thrombosis d) Smoking does not seem to be equally causally related to either lung cancer or coronary thrombosis e) No comparative statement is possible between smoking and lung cancer or coronary thrombosis

b) Smoking seems much more likely to be causally related to lung cancer than to coronary thrombosis

Chapter 15

bias, confounding, and interaction

Stratified random sampling

breaking down sampling frame into sub-categories based on some factor likely to influence the level of characteristics being measures (like sampling different age ranges). Percentage of each sub-group sampled doesn't have to be the same. *These are easy, provides representative sample, can be used for many types of populations, but can be time consuming and difficult and often require large samples

Etiology

cause of disease

2. Development of disease Pre-symptomatic disease

changes in host pathology but no signs/symptoms

Confounding (non-causal)

characteristic + another factor →disease

Non-probability sampling

choosing subjects in a non-random way, every member has a different probability of being chosen

evidence-based minimum set of items for trials reporting production, health and food-safety outcomes CONSORT

consolidated standards of reporting trials, for randomized trials with human subjects

Mortality

death - Measures of these can be crude (capture all causes/types of disease/death) or cause-specific (express level of disease/death caused by a particular factor) - Similar formulas to incidence but with different numerators

External validity

degree to which inferences drawn from a study can be generalized, or extrapolated, to the broader population of interest. 2. How will you choose these subjects?

Internal validity

degree to which the observed findings in a study lead to correct inferences about the outcome of interest in the source population. Improved with an unbiased study. "How well do the results relate to the source population?"

Disease

derangement in the function of the whole body or any of its parts, physiological or psychological dysfunction

Epidemic curve

distribution of incubation periods

Representativeness

establish inclusion/exclusion criteria before you start sampling

Estimated variance

estimate variance before you begin (a priori decision) Variance for proportions = p*q P = those who have it, q = those who don't have it Variance for means = σ2 = Σ(x-x)̄ 2 / n x = individual value x̄ = mean value n = total number/population Equal to standard deviation squared

Rates

expresses the occurrence of an event or outcome in a population in a specified period of time can be considered a measure of "speed" (1 infection per 100 person- years, 5 deaths per 1000 cow-decades) ...

Causal relationships Sufficient but not necessary

factor alone can produce disease but so can other factors

Causal relationships Direct

factor directly causes disease (no intermediate step)

- Cutoffs chosen based on the consequences of

false positives and false negatives for a given disease

Simple random sampling

fixed percentage of the course population being chosen using formal random process, all have same probability of being chosen, sample should be representative of done correctly.

Other Cluster sampling

focuses on sampling at a group level (households, herds, regions). Unit of concern is still the individual - once a cluster is selected, all individuals are tested. Can be through probability or non-probability.

Foodborne disease

from contaminated food

• Potential for organism to spread depends on

growth rate and transmission rate

Endemic

habitual presence of a disease within a given geographical area

Bimodal distribution

has 2 peaks, allows clear separation of population (those with and those without disease) • Both distributions have a "gray zone" in which people could belong to eithergroup• Both allow clear separation from overall "normal" from "abnormal"

Infected

host invaded my microorganisms, they multiply, host immune system responds

Secondary prevention

identify people in preclinical phase of disease (not showing symptoms) to reduce severity, early detection (ex. screening for cancer)

Randomization

if each person has an equal chance of being placed in any one group, it's likely that CFV's will also be equally distributed

Diseased

infection that causes symptoms

Vector

insect/living carrier that transports agent from infected individual (reservoir) to susceptible individual (final host)

Causal relationships Indirect

intermediate step(s), more common in human biology

Judgment sampling

investigator chooses what he deems to be units that are representative of the population

Exclusion bias

investigators apply different eligibility criteria to cases vs. controls in regards to which past clinical condition would exclude them from the study

secondary attack rate

is sometimes calculated to document the difference between community transmission of illness versus transmission of illness in a household, barracks, or other closed population. It is calculated as: Number of cases among contacts of primary cases / Total number of contacts x 10^n ***Often, the total number of contacts in the denominator is calculated as the total population in the households of the primary cases, minus the number of primary cases. For a secondary attack rate, 10n usually is 100%.

food-specific attack rate

is the number of persons who ate a specified food and became ill /divided/ by the total number of persons who ate that food, as illustrated in the previous potato salad example.

an attack rate

is the proportion of the population that develops illness during an outbreak. For example, 20 of 130 persons developed diarrhea after attending a picnic. (An alternative and more accurate phrase for attack rate is incidence proportion.)

A prevalence rate

is the proportion of the population that has a health condition at a point in time. For example, 70 influenza case-patients in March 2005 reported in County A.

Overall attack rate

is the total number of new cases divided by the total population.

Prognosis

likely outcome

Kappa statistic

measurement of agreement beyond chance alone

Acquired immunity

memory, specific, increased response with each exposure, natural or artificial - Natural active: exposed to live antigen - Natural passive: antibodies from mom to fetus - Artificial active: vaccine - Artificial passive: injection of antibodies

Co-morbid infections

more than one disease

Causal relationships Neither sufficient nor necessary

most common

Unimodal distribution

most common, has 1 peak, must choose cutoff to separate people (not always easy!) • Both distributions have a "gray zone" in which people could belong to eithergroup• Both allow clear separation from overall "normal" from "abnormal"

Net sensitivity

must be identified as positive by test A, test B, or both 1. Determine positive by A and B (don't want to count twice) a. Sensitivity of test A x total positive people b. Multiply this by sensitivity of test B 2. For only A a. (Sensitivity of A)(Total with disease) - (positive by both) 3. For only B a. (Sensitivity of B)(Total with disease) - (positive by both)

Innate immunity

natural, unspecific, no memory, same response with each exposure (ex. stomach acid, skin, inflammation)

Causal relationships Necessary but not sufficient

need factor to develop disease but can't do so by itself

Latent

no active multiplication of agent (ex. when virus incorporates its genome into that of host), only genetic material present in host, not actual organism!

Immunity and susceptibility Immune

not at risk due to immunization or previous infection

- Too high a cutoff

not enough true diseased people included, low sensitivity

Case fatality rate

not true rates! - Measures mortality of disease (often acute)

Pre-clinical

not yet apparent but will progress to clinical

- NAR =

number at risk - Assume withdrawals left halfway through time span

proportion is = number of women in State A who died from heart disease in 2004/number of women in State A who died in 2004

number of women in State A who died from heart disease in 2004/number of women in State A who died in 2004 =B. Proportion

• Host susceptibility determined by

nutrition, immune system, genetics

Epidemic

occurrence in a community/region of a group of illnesses similar in nature, in excess of normal expectancy, derived from a common source of propagated source

Recall bias

one group more likely to remember info

Differential

one group of participants more likely than the other to be inaccurately assigned to exposure or outcome (ex. those with disease more likely to remember exposure than those without), can make associations more or less likely

Net sensitivity =

only A + only B + A and B / total with disease Net specificity: must be identified negative with both tests

Reporting bias

participant is reluctant to disclose info because of beliefs/perceptions

Unplanned/natural experiments

participants have been exposed for non-study purposes to compare to non-exposed individuals 1. Clinical observations 2. Identify routinely available data 3. Case-control study (ex. compare cancerous smokers to non-cancerous smokers) 4. Cohort study (ex. comparing lung cancer in smokers to non-smokers) 5. Randomized trial (only for beneficial outcomes, not for cancer)

Social desirability

participants tend to report answers they believe to be more acceptable

Loss of follow-up

participants who withdraw are systematically different than those who remain

Non-responders

people who don't reply when contacted for study participation can create bias (*do not confuse with "selection of subjects")

Screening test

population approach on apparently healthy people to detect subclinical disease, provides early indication

Source population

population from which the study subjects are drawn (each member has a non-zero probability of being selected)

- Screening will only spit out two results

positive and negative

Sample size for estimating a mean Sample size for estimating a proportion (talking about percentages) Analytical studies Goal is to test differences between two groups with respect to a factor First state null hypothesis (no difference exists between A and B), and alternative hypothesis that predicts a difference. One-sided hypothesis

predicts one specific trend (A higher than B) Two-sided hypothesis: more common, predicts trend in either direction (A higher or lower than B) Expected difference: based on previous work

Tertiary prevention

preventing complications in a diseased individual already showing symptoms (clinical phase), lessen impact of disease

Risks

proportions that express the probability that something will occur

Probability sampling

random process to select subjects, all with non-zero probability of being chosen.

Proportions

ratio where number is included in the denominator (1/100, 27%, 0.55)

Persistent/chronic

re-manifestation of symptoms many years after infection was "resolved"

Errors Type 1 error

rejecting the null hypothesis when it is true (saying there is a difference when there isn't), this probability is equal to α, results in a false claim Type 2 error: saying there is no difference when there is one (accepting the null hypothesis when it is false), this is equal to β, results in a missed opportunity

Multistage sampling

sampling takes place at group and individual level (sample a few of the individuals from the group). 3. How to determine sample size Descriptive studies

Convenience sampling

sampling units are chosen because they are easy to get

Purposive sampling

sampling units are chosen on purpose because of their exposure of disease status. Common in analytical studies (looking at rare outcomes). Used because there are very few of these subjects in a population. *These are cheap, easy, and good for similar populations, but can lead to biased results!

Sample

selecting a portion of the population and acquiring and recording relevant information from those selected members (like randomly selecting 10% of Ontarians) - Allows researchers to infer information about a population without investigating every individual - Describe characteristics about a population - Less time consuming and more cost effective - Must have a large enough sample size to infer a true association

Inclusion criteria

set of characteristics the subject must have to be included in the study

3. Development of disease Clinical disease

signs/symptoms

Prepare the field for outbreak (response) - Briefing and background research, all members involved must know everything about disease and suspected agents - Review literature to assess potential sources of pathogen, modes of transmission, risk factors, diagnostics, treatment... Assemble the team - May be regional or national in scope depending on severity of disease/pathogen - Lab scientists, epidemiologists, clinicians, food regulators, communications, managers and media - Must be a level of politics and jurisdiction for expertise and authority Assign responsibilities - Each member needs role - Collection and analysis of data, implementation of control measures, communication, overall coordination Control measures - Immediate control measures should be put into place to contain outbreak - Must be evidence-based and balance public health risk of causing unnecessary stress and negative impacts on businesses Step 5 - Very important to establish a case definition

standard set of criteria for deciding whether an individual should be classified as having the condition of interest (CDC) - Simple and most objective criteria are best, guided by usual presentation of disease - May need to change over time as the investigation unfolds

Health

state of complete physical, mental, social wellbeing, not just absence of disease

Epidemiology

study of the frequency, distribution and determinants of health and disease and application of this to control health problems, aims to improve control of disease through prevention and treatment • Disease, illness, health are not randomly distributed • Based on individual's predisposition (genetic or acquired)

Outbreak

sudden increase in cases of disease above what is expected in a population (very similar to epidemic, but in a more localized location)

Census

systematically acquiring and recording relevant information about every member of a given population (like the Canadian census) - Good way to get detailed information, but impractical! - Time consuming - Costly (research team and resources) - Tedious - Difficult if participation is voluntary (not mandatory)

2. High-risk approach

target high-risk groups (can be more expensive and invasive)

Time of outbreak

temporal patterns Common source outbreak - Individuals exposed to a common noxious agent Point source - Individuals exposed over short period of time - Everyone becomes ill at the end of one incubation period - Epidemic curve (graph of number of cases by day) has steep upslope and gradual downslope • Example: contaminated food eaten at a wedding Intermittent source - Individuals exposed over days, weeks... - Epidemic curve has pattern of intermittent exposure • Contaminated cheese sold every weekend at the market Continuous source - Individuals exposed over days, weeks... - Epidemic curve has smaller and wider peaks, continuous exposure results in few gaps in curve • Example: contaminated dog food with a long shelf like

Net sensitivity =

tested true positive for both tests / total with disease

Study/sampled population

the group of sampling units that was selected from the sampling frame) *This order is the way subjects are selected

Chance agreement

the probability that two examiners agree due to chance alone rather than actual agreement

case-fatality rate

the proportion of persons with the disease who die from it. For example, one death due to meningitis among County A's population. All of these measures are proportions, and none is expressed per units of time. Therefore, these measures are not considered "true" rates by some, although use of the terminology is widespread.

Precision

the width of your confidence intervals around your mean, L, the allowable error in a given study.

- Too low a cutoff

too many healthy people included, low specificity

4. Development of disease Disability/recovery

treatment leading to one of these 2

Net specificity =

true negative for test 1 + true negative for test 2 / total without disease

Level of confidence

typically 95%, Zα Level of significance (α) = 1 - level of confidence (0.05) - 95% of the time the range of values around the estimate we calculate will include the true value. Z value is 1.96

Diagnostic test

used on sick people to confirm, classify, inform treatment, or provide prognosis for a particular disease Biologic variation of human populations

Zoonotic disease

vertebrate animal →human (75% of emerging diseases, ex. West Nile virus)

Subclinical

will not become clinically apparent, detected by antibody response

Causal relationships Necessary and sufficient

without factor, disease never develops, and with factor, it always does (rare)

Dichotomous variables

yield a positive or negative result

- 5 main criteria

• Agent • Clinical criteria • Person • Place • Time - Questionnaire or survey used to collect information from cases • Identifiers (name, address, phone number) • Demographics (age, sex, occupation, ethnicity) • Clinical: verify case definition is met, determine date of onset, if hospitalized • Risk factors: specifically tailored to disease being investigated

- Reduce selection bias by

• Careful design study (inclusion and exclusion criteria) • Reduce non-response: good recruitment strategies

Confounding bias

• Confounding factors can cause one to see a true association and assume it is causal when it isn't

Gold standard "true" results

• Must know who actually has disease to do this...which never really happens

- Reduce information bias by

• Use best available collection tools/tests and understand how potential bias affect these • Reduce misinformation reported (decrease recall period, ensure participants feel safe)