Epi 712: General definitions Flashcards
Bias
Prejudice in favor of or against one thing, person, or group compared with another, usually in a way considered to be unfair.
case fatality risk (CFR)
Proportion of deaths within a designated population of “cases” over the course of the disease.
of deaths from specific disease/ # with disease
aka. or case fatality rate, case fatality ratio or just fatality rate
Cohort
Group of subjects who have shared a particular event together during a particular time span.
disability-adjusted life year (DALY)
Measure of overall disease burden, expressed as the number of years lost due to ill-health, disability or early death. Include equivalent years of ‘healthy’ life lost by virtue of being in states of poor health or disability. In so doing, mortality and morbidity are combined into a single, common metric.
Years LOST to premature death or disability) = YLLs (death) + YLDs (disabilty)
Estimate of morbidity that accounts for the burden of disease due to specific cause in a population
Efficacy
Ability to produce a desired or intended result. Intervention or drug in medicine.
Endemic
Regularly found among particular people or in a certain area.
(from Greek ἐν en “in, within” and δῆμος demos “people”
Epidemic
Higher than normal (baseline) rates of disease in a community at a particular time.
Incidence
- New Cases
- Measures change in disease occurance during a given period of time.
- Often for infectious/acure disease studies
- Used for investigating causes of disease (etiolog)
- I = # of NEW cases of disease occurring in one time period/ Total # AT RISK in time period
Morbidity
Diseased state, disability, or poor health
Mortality
Death.
Crude Mortality Rate (CMR)
Measure of the number of deaths in a population, scaled to the size of that population, per unit of time.
deaths from all causes/ # of persons in total pop.
units: x per 1000
Prevalence
- Existing cases; how much disease is in population
- Measure of disease that allows us to determine a person’s likelihood of having a disease.
- Often used in chronic disease studies
- Used for description and planning health care needs
- P= # of EXISTING cases of disease at specified time/ total population at specified time
Proportionate mortality rate (PMR)
of deaths from specific disease/
of all deaths
= % of deaths in a population caused by disease
Validity
Degree to which the results of a study are likely to be true, believable and free of bias. Results are true to target population
Descriptive Epidemiology
Distribution “Person - Place - Time”
WHO participated in study or event
WHERE did study or event occur?
WHEN did study or event occur?
Analytic Epidemiology
Determinants “Agent - Host - Environment”
Look for causes/ Test Hypothesis.
Why did disease outcome occur in certain people or population groups?
Epidemiology
“Study of the distribution and determinants of health-related states in human populations” Framework for IDing public health problems, causes, and solutions.
Histoty: Hippocrates
460-390 BCE 1st to recognize environments role on disease etiology wrote descriptions of clinical diseases
History: Use of Vital Statistics
14th & 15th centuries Italy: track death rates and causes England: issue death certificates
History: Describe Physical Universe
17th Century Rammazzini: publish enviro hazards in occupations Graunt: use population-based mortality data to study disease occurrence
History: Scientific Reasoning
18th century Figured out disease is caused by exposures Lind: studies etiology of scurvy (lemon guy) Pott: IDed chimney soot as cause for scrotal cancer for chimney sweeps Jenner: developed smallpox inoculation
History: Miasma Theory of Disease
Early 19th Century Epidemics caused by “spontaneous generation” & “Miasma Atmospheres” - enviro conditions/poor sanitation
History: William Farr
Late 19th Century Founding Father of modern Epi (Nonmechanistic) Established national registration system in England
History: John Snow
Mid-19th Century Founding Father of Modern Epi (Mechanistic) Determined cholera as waterborne disease
Non-Mechanistic view of disease
- Environmental based Disease
- Theory= Miasma Health
- Intervention = Hygiene, Enviro/behavioral
- No emphasis on cause
Mechanistic view of disease
- biological based Disease
- Theory = Microbiologic Health
- Intervention = avoid contagion, infectious agent/gene
- Emphasis on cause
History: Germ Theory
19th Century Henle: sick person pass contagious substances to healthy people Pasteur: isolates bacteria & kills by boiling Koch: used microscope to see TB and cholera microbes
Henle-Koch Postulates
- Agent must be present in every cause of the disease 2. One agent = One disease 3. Exposure of healthy subjects to agents results in disease
ID of arthropod vector and asymptomatic carriers
Late 19th Century
History: Modern Epi
20th Century new methods: outbreak investigation, biostat, clinical trials Cigarette smoking causes lung cancer (1960s) Eradication of smallpox (1978) Focus on chronic disease
“The Epidemiologic Transition”
Population shift in disease and mortality patters in high-income countries: - low-income: deaths by infection/malnutrition, high infant mortality, short life expectancy - high-income: deaths by chronic disease, low infant mortality rate, long life expectancy
Health
Complete state of physical, mental and social well-being, not merely absence of disease (WHO definition)
Research
Process of systematically, carefully investigating a single well-defined subject to learn or discover.
Health Research
Examines factors that contribute to health
What is Epi
- basic science - way of thinking - tool for testing/evaluating interventions - study of distribution & determinants of health related states in human population
Etiology
Cause of disease
Research Process
- Identify study question 2. Select study approach 3. Design study & collect data 4. Analyze data 5. Report findings
Criteria for Risk Factor
- Frequency of disease varies by category or value of exposure - Risk factor precedes disease onset observed assoc. is not due to any source of error - Assoc. does not = Causation
Age-Adjusted Rates
Ussed to compare two populations when crude mortality rates are inaccurate due to different population age structures.
rates can be compared if they have been adjusted to teh same STANDARD population.
Age Standardization
Removes effect of age among population being compared and STANDARDIZES the population.
Direct Age Adjustment
Defines what the expected number of events in the standard pop given the rate in the study pop
Use when comparing large, well-defined study pop.
Method:
- Measure age-specific rates in pop. being compared
- Choose a STANDARD population (i.e large established pop, such as US census data)
- Apply age-specific disease/death rate in one of the study populations to teh age distribution of the STANDARD pop to calculate adjusted rate in STUDY pop.
Indirect Age Adjustment
Define what is the expected number of events in the STUDY pop. given the rate in the STANDARD pop.
- Use when you have limited info on the study pop (observed deaths not known) and the study pop is small
Method:
- Select appropriate STANDARD population for the STUDY pop.
- Use the age distribution of the STUDY pop and the age-specific rates from the STANDARD pop to calculate SMR.
Standardized Mortality Ratio (SMR)
RATIO calculated in In-direct Age Adjustment
= total # Observed deaths/ total # Expected deaths
- SMR = 1: no diffierence b/w stydy pop and standard pop
- SMR> 1: study pop is experiencing higher than expected death (EXCESS Disease)
- SMR < 1: Study pop is experiencing lower than expected death (REDUCED Disease)
Conditional Probability of Survival
Probability of surviving 2 years GIVEN the person survived the year before. Relies on previous event! =P(survive 2 yrs | survive 1 yr)
Cumulative Probability of Survival
Probability of surviving 2 years from diagnosis is: P (survive 1 yr)* P(survive 2 yrs | Survive 1 yr)
Life Tables
Survival analysis method.
- based on regular interval check-ins (i.e. Yearly)
- Primary statistic is SURVIVAL
- Cases lost to follow-up are accounted
- Cases may be FIXED or DYNAMIC
Kaplan-Meier Method
Survival analysis method
- Event triggered by DEATH
- Primary statistic is SURVIVAL
- Cases lost to follow-up are accounted
- Cases may be FIXED or DYNAMIC
Primary Study Design
Collect and analyze new data.
Individual level data
Study Types; Case Series, Cross sectional, case-control, cohort, experimental, Quantitative
Secondary Study Design
Analyze existing data at indivudal or population level
Study Type; Ecological, case series, cross-sectional, case-control, cohort, experimental
Tertiary Study Design
Uses existing data to review and synthesis of literature
Individual level data
Study Types: Review, meta-analysis
Descriptive Stidues
- Case series - cross-sectional
Analytic Studies
- Case control - cohort - experimental
Study Type: Review/ Meta Analysis
Gathers all prior publications on topic and summarizes into big-picture analysis.
Good for identifying consensus and confussion
Tertiary analysis study design
Study Type: Ecological
- Secondary Analysis study deisgn
- Uses population-level to examine the relationship b/w E and D in P
- Exposure often “environmental”
- Pop characteristics in aggregate formed
- Scatter plot best analysis for correlation
- Beware of Ecological Fallacy
PROS: cheap, convenient, simple
CONS: “ecological fallacy,” pop level analysis only
Ecological Fallacy
Correlation studies compare groups rather than individuals. Incorrect attribution of pop level associations to indivudals
Study Type: Cross-Sectional
- Descriptive, Observational study
- Measures proportion of pop with a particular E or D at ONE POINT IN TIME.
- “Snap shot” aka. prevalence study.
- Uses: describe communities, assess op needs, evaluate programs, establish baseline
- Participants “representative” of larger pop
- Analysis: Prevalence, comparative statistics, ASSOCIATION or relation b/w E and D, but NOT CAUSE.
PRO: Cheap, simple, rapid collection of data
LIMIT: needs representative pop.; Can’t study causality
Sampling
- Defining a representative study population of a target population
- Bigger sample = narrower confidence intervals = more likely to have “statistically significant” result
- Power: ability of statistical test to detect significance in pop when differences exist
Study Type: Case Series
- Descriptive, Observational study
- Describes 2 or more patients who have same DISEASE condition or same procedure.
- No controls; no comparison group
- Uses: describe characteristics among group, identify UNUSUAL syndromes or refine case definitions, develop hypothesis for future research
- Case report describes ONE patient
Study Type: Case-Control
- Analytic, observational study
- Participants selected based on DISEASE status and looks back at EXPOSURE history.
- Casese: w/ disease;
- Case definition: all cases have same disease, inclusion/exclusion criteria must be defined.
- Control: w/o disease Use: observational or analytic;
- Control definition: reasonably similar to cases, same inclusion criteria as cases expect for having disease
PRO: Good for studying RARE DISEASES
CONS: prone to misclassificaiton and recall bias
Nested case-control study: use records to assess exposure histories. Minimizes recall bias.
Misclassification bias
- IN accuracies in methods od data aquistion.
- Caused when definitions b/w cases and controls are fuzzy.
- Happens in Case-Control Studies
Recall bias
Occurs when cases or controls have different memories of the past.
Can be avoided by using a nested case-control that uses old records to assess histories.
Happens in Case-Control Studies
i.e. cases have more vivid memories than controls b/c they are searching for illness explanations.
Frequency Matching
Recruit a control population similar to the case population
Individuals are not tied to a indivudal case
Used in case-control
EX: controls are from hospital registration with same admission time, sex, and similar age.
Matched-pair matching
Each case is linked to a particular individual control.
Good for genetic studies.
Matching criteria CANNOT be considered as exposure during analysis, such as age.
Ex: sibling or close friend.
Study Type: Cohort
- Analytic, observational study
- Recruit based on EXPOSURES to watch foward in time for DISEASE or risk factors.
- Group of similar people w/o disease followed through time together
- Calculates disease incidence over period of time.
- Needs at least 2 measurement times: baseline and follow-up
Three cohort types: retrospective, prospective, and longitudinal
PRO: Good for RARE EXPOSURES
CON: Prone to information bias, which is when exposed are more throurghly assessed than unexposed.; Lost to follow-up is a major concern (prospective and longitudinal)
information bias
Biasin the way that infmaotoin is collected from study participants
- Recall Bias:
- interviewer bias: Results when exposed participants in a cohort study are more thoroughly examined for disease than unexposed.
- non-response bias
Study Type: Retrospective cohort
Recruit based on EXPOSURE status: one group w/ exposure, one group w/o exposure.
Use baseline info collected in the PAST and follow cohort to a FORWARD point in time.
BEWARE: Must be able to demonstrate the outcome of interest was not present in any members of the cohort baseline.
Study Type: Prospective Cohort
Recruit based on EXPOSURE status: one group w/ exposure, one group w/o exposure.
Collect baseline data in PRESENT and follow cohort to a FORWARD point in time.
BEWARE: Must be able to demonstrate the outcome of interest was not present in any members of the cohort baseline.
Study Type: Longitudinal Cohort
Recruits based on membership in well defined population REGARDLESS of exposure
Usually last for years to decades
Assesses at baseline for SEVERAL exposures and disease, then followed FORWARD to determine incidence for one OR MORE outcomes of interest.
Works with fixed or dynamic population
PRO: examine multiple effects of single exposure
Target Population
General pop that study seeks to understand
Source Population
Specific individuals from with a representative sample is drawn
Sample Populiation
Individual asked to participate in study
Study Population
Eligible participants
Systematic Sampling
Sampling after a random start point,, every nth person is selected
Stratified sampling
simple random samples selected from each of several strata.
ex: divide into gender groups and pick same # from each
Cluster Sampling
Area is divided into geographic clusters and some clusters are selected for inclusion.
ex: include select streets in nieghborhood
Simple Random Sampling
Sampling method were each person has equal chance of being selected.
Measures of association (For Case-Control Study)
Asks if exposure and outcome (disease) are statistically related.
- Cohort: RR (rate ratio)
- Case Control : OR (odds ratio)
Rate Ratio (RR)
(For Cohort Study)
RR is the number of times greater the risk of disease in the exposed compared to unexposed. *denomimator if always reference popoulation.
aka. Incidence Rate Ratio (IRR)
Formula: **RR = Ie/ Iu **
Ie = a/ (a +b)
Iu = c/(c+d)
- RR > 1: POSITIVE association = RISKY effect
- RR = 1: NO association - NULL association
- RR < 1: NEGATIVE association = PROTECTIVE EFFECT
Interpretation: Subjects with exposure were (RR) times more likely than subjects w/o exposure to have disease/outcome.
Attributable Risk (AR)
(For Case-Control Study)
Excess rate of disease in the exposed - that is, the incidence of disease among exposed that is due to exposure.
aka. Risk difference
Formula: **AR = Ie - Iu **
EX: What is the incidence of lung cancer in smokers due to smoking?
Attributable Risk Percent (AR%)
Proportion of disease among the exposed that is due to the exposure.
aka. Etiological fraction
Formula: AR% = (Ie- Iu) / Ie
EX: What percent of lung cancer in smokers is due to smoking?
Population Attributable Risk (PAR)
Excess rate of disease in the total population that is due to the exposure.
Formulat: **PAR = It - Iu = AR x Pe **
It= (a+c)/ (a+b+c+d) = incidence in total pop
Pe= (a+b)/ (a+b+c+d) = Prevalence of exposure
EX: What is the incidence of lung cancer in the tial US pop due to smoking.
Population Attributable Risk (PAR%)
Proportion of disease in the total population that is due to the exposure.
PAR% = (PAR / It) x 100
EX: What percent of lung cancer in the total US pop is due to smoking.
Odds Ratio (OR)
OR is the odds of people likely to have been exposed to the disease. Odds of exposure among diseased (cases)/ odds of exposure among non-diseased (controls)
Formula: OR= (a/c) / (b/d) or = ad/bc
- OR > 1: POSITIVE association = RISKY effect
- OR = 1: NO association - NULL association
- OR < 1: NEGATIVE association = PROTECTIVE EFFECT
Interpretation: “Cases have (OR rate) times greater/lesser odds of being exposed than controls.”
95% Confidence Interval Interpretation
RR: (Upper bound, lower bound)
- Both < 1 = exposure is PROTECTIVE
- Lower < 1, Upper > 1 = NO association
- Both > 1 = Exposure is RISKY
OR: (Upper bound, lower bound)
- Both < 1 = Cases sign. Lower odds of exposure than controls
- Lower < 1, Upper > 1 = NO association
- Both > 1 = Cases have sign. Higher odds of exposure than controls
Matched Case-Control Studies (For Case-Control Study)
Individually Matched aka matched-pairs Matched -Pair OR:
- Concordant pairs: (a + d): same exposure experience for case and control
- Discordant pairs: (b + c): different exposure experience for cases and controls
- Matched pairs “OR”: = ratio of discordant pairs = b/c
Experimental Study
Assigns participants to intervention and control groups in order to examine whether intervention causes an intended outcome.
- exact timing, dose, duration, and frequency of exposure is known
PROs: Gold standard for CAUSALITY, researcher has control over exposure
LIMITS: ethics considerations, costly, time consuming, liability; issues of noncompliance
Randomized Controlled Trial (RCT)
(for Experimental Study)
- Participants are randomly assigned to an ACTIVE intervention group; remaining assigned to CONTROL group.
- Participants are both followed forward in time to see outcome,
- Reasons to NOT randomize and treat all: unethical to not provide treatment to all, Use of “before” population as controls for the “after” population becasue it would be unethical to not treat group.
Sample Randomization approaches (for Experimental Study)
- simple
- Block
- Stratified
Superiority Trials
New intervention is “better” than the control. * most common*
Equivalence trials (for Experimental Study)
New intervention as good as comparision
Non-Inferiority trials (for Experimental Study)
New interventions no worse than the comparision
Placebo (for Experimental Study)
Inactive comparison group (control) that is similar to the therapy tested.
Hawthorne Effect
Participants change their behavior for the better simply b/c the know they are being observed. May cause interference with accurate measurement of impact.
- Ex: Controls in weight loss experiment will try to loos weight anyway in anticipation of being weighed.
Blinding
Participants in experimental study do not know whether they are in intervention or control group. Minimizes information bias
aka. masking Types:
- Single-blind: participants are unaware
- Double-blind: Participants and persons assessing (researchers, PI, interviewers) are unaware -
Equipose
Experimental research should be conducted only when there is GENUINE UNCERTAINTY about which treatment will work better.
Distributive justice
Source population must be an appropriate and non-exploitative one.
ex. giving advanced treatment to diseases people in poor country, where after conclusion of the study the treatment is not available.
Beneficence vs. nonmaleficence
Go good vs. Do not harm
Researchers must balance the two likely risks and benefits
Respect for persons
- Participants must volunteer for study on their own w/o unduly influence for being compensated
- Participants understand what is means to participants randomization of being in active or control group
- Give informed consent
- Researchers protect safety and privacy of participants
Efficacy (For Experimental study)
Proportion on subjects in control group who experience unfavorable outcome who were expected to have a favorable outcome had they been in the active group.
- High efficacy is an indicator that intervention is successful
Interpretation: “E % of people in control group could have been prevented their bas outcome if they had been in the intervention group instead.”
Formula: E = (rc- ri)/rc = Rate of unfavorable outcome in control group.
rc= c/(c+d) = rate of unfavorable outcome expected in the control group
ri= a/(a+b) = rate of unfavorable oucome in intervention group.
Interpretation: “X% of the people in the control group could have prevented their bad outcome if they had been in the intervention group.”
Number Needed to Treat (NNT) (For Experimental study)
Expected number of people who would have to receive treatment to prevent an unfavorable outcome in one person.
- small NNT indicates a more effective intervention
Formula: NNT = 1/ (rc - ri)
Interpretation: NNT of x means that x people have to take the drug to prevent one of the x from having the disease.
Treatment-received approach (For Experimental study)
Limit analysis to the participants who were fully compliant with their assigned intervention
(efficacy = laboratory setting)
Treatment-assigned approach (For Experimental study)
Includes all participants even if they were not fully compliant with assigned intervention
(effectiveness = real-word setting)
Experimental Study Analysis concerns
Non-compliance: not all participants follow protocol.
Placebo effect: even a placebo will make many participants fee better
Loss to follow-up: even if the same % of ppl in each treatment group are lost to follow-up, there may be bias (treatment group may quit b/c they get healthier and the placebo group may quit b/c they get sicker)
Conditional Probability
Probability of a surviving two years GIVEN that the person survived the one year
Formula: P(survive 2 years | survive 1 year) = .5
Cumulative Probability
Probability of a surviving two years
Formula: P(survive 2 years from start) = P( survive 1 year) * P(survive 2 years | survive 1 year)
Counfounding
When two exposures are realted and that makes it look like exposure A causese the disease when really exposure B causes the disease.
Ex. Heavy alcohol consumption is associated with higher rate of lung caner, but really heavy drinkers are more likely to smoke, and tabacco is what causes the cancer.
- 3rd variable evidence: 1) stratified ORs are equal (OR1 = OR2) AND 2) Stratified ORs DO NOT equal Crude OR;
- BD Test: P > .05, meaning strata are SAME.
- Measure reported: Mantel-Haenszel (MH) - summary measure existing somewhere b/w OR1 and OR2
Effect Modification
When different subpopulations with different biological responese are incorrectly grouped, OR or RR is not accurate to individual groupings
ex. Males and females have different biological risks; when grouping them together may hide the true difference that exists.
- 3rd variable evidence: if: OR1 not = OR2 not = Crude OR;
- BD Test: P < .05, meaning strata are different.
- Measure reported: stratum specific measuse (both ORs)
Interaction
Presence of two risk factors at the same time causes different outcomes than presence of either one separately. Can be additive or multiplicative.
Four Casual Relationships
Necessary = disease cannot occur without factor
Sufficient = disease always develops when factor is present
- Necessary and sufficient: FACTOR –> DISEASE
- Rare; infectious agent causes disease in every exposure instance
- Necessary, but NOT sufficient: FACTORS 1 +2 +3 –> DISEASE
- INfectious agent must be contracte dto get disease, but not everyone gets sick (stages of cancer)
- Sufficient, but NOT necessary: FACTORS 1 or 2 or 3 –> DISEASE
- Disease can be caused by various factor independently.
- NEITHER necessary nor sufficient: FACTOR 1 +2 or 3 +4 –> DISEASE
- complex factors contribute to disease
Types of Prevention
- Primary: Prevention, ppl w/o disease; ex. vaccines, han-washing, etc.
- Secondary: Ealry diagnisis, ppl with early, non-symptomatic disease, ex. screening
- Teritary: Treastment and rehabilitation; ppl with sympyomatic disease, ex. antibiotics, physical therapy, rehab, etc.
*
Screening
Application of a test to persons w/o known disase for purpose of determining likelihood they have disease.
Criteria:
- Disease is of public health inportance, SEVERE, has high FREQUENCY in pop, and has LATENT stage.
- TREATMENT is AVAILABLE to screened pop.
- Screening test is ACCEPTABLE to pop (in terms of cost, risk, comfort and info) and VALID
Screening test validity
How closely the test observes actual presense or absense of disease?
Does test really measure what is it meant to?
Are the disease classificaitons accurate?
Two Types:
Sensitivity and Specificity
Cut-off points:
- When cut-off is lowered, SE increases/ SP decreases (more FPs)
- When cut-off is raised, SE idecreases/ SP increases (more FNs)
Sensitivity (screenign test validity)
Probability a person with disease is classified at having disease.
Interpretation: SE% of people with disease will test positive for disease
Formula: TP/ (TP + FN) * denominator = total w/ disease
Specificity (screening test validity)
Probability a person with w/o disease is classified as NOT havingt the disease.
Interpretation: SP% of people w/o disease will test negative for disease
Formula: TN/ (TN + FP) * denominator = total w/o disease
Positive Predictive Value (Screening test validity)
PPV or PV+
Proportion of ppl who test postive that actually have disease
Forumla: TP/ (TP + FP)
- PV + is trongly influenced by disease prevalence. Low PRev = low PV+ and many FPs
- Higher PV+ indicates more efficient use of screening resources
Negative Predictive Value (Screening test validity)
NPV or PV-
Proportion of ppl who test negative that DO NOT have disease
Forumla: TN/ (TN + FN)
Kappa
Measure of the degree of nonrandom agreeent bewteen observers or measurements of some categorical value.
Formula: Kappa = (Observed agreement - Exepcted Agreement)/ (1- Expected Agr)
Kapp value range: -1 to 1 where:
- negative value = measurement disagree more than expected
- positive value (0-1) = measurements agee more than expected
- closer to 0 = agreeent is chance
- closer to 1 or -1 = complete agreement
Surveillance
Ongoing systematic collection of health data
- monitoring health events (need base-line data)
- priority setting, planning, implmenting, and evaluating disease
Common Problems:
- only captures those who sought formal care
- healthcare workers have little incentive to report
- case defiinition is unclear
- incidence rates imposisble to calculate if pop # is unknown.
Biases
- Attendence patterns
- diagnostic methods (more testing = more disease)
- Screeing (“seek and ye shall find”)
- Reporting propensity (notification delays, failure of agency to report back)
Active Surveillance
Public health agencey reacheds out to local healthcarea providers to request info
- When a disease is thought to be occuring and help is sought.
Passive Surveillance
Local heatlthcare providers provide reports to public health orgniazation (either voluntary reports or notifications mandated by law).
- Continuus, as disease cases come up.
Sentinal Surveillance
Public health orgnaization selects a sample of healthcare providers and receives reports from them; active surveillance of a larger # of providers can be initaitied if an outbreak appears to be occuring
- random monitoring of small sample size.
Syndromic Surveillance
Asking for a report to be submitted when a patient has a particular set of symptoms, rather than reqiring a formal lab diagnostic
- Good when no cheap tests are available; just watch and report
Community-Based Surveillance
Community health volunteers assist with data collection and reporting of a limited number of syndromes
- select ppl to help with repoting in community.
Crowd-sourced Surveillance
An emerging technique that scans reports from Teitter and other social media to detect outbreaks early
- monotring google seraches for key words associated with disease by region
Resaerch Approaches for public health policy
- Implementation (process) research: what policies would improve health services
- Evaluation (outcome) resaerch: Do implemented policies work?
Types of resaerch Outcomes
- Efficacy: Does program work in lab setting (ideal, controlled); evaluate treatment-received.
- Effectiveness: Does program work in real life (uncontrolled; evaluate treatment-assigned.
- Efficiency: Is cost-benefit ratio favorable?
Type 1 error
Finding an association (outcome) in a test when there is NO real association. Occurs by chance.
- False positive.represented by proporton ALPHA
- CI = 1-alpha (i.e 95%) (power)
Type 2 Error
Finding NO association (outcome) in a test when there is a real association. Occurs by chance.
- False positive.represented by proporton BETA
- CI = 1-BTA (should be no less than 8-%)
Modes of Disease transmission
Direct (person to person)
Indirect
- airborne
- vetor-borne
- vehicle-borne
Agent Reservior
- Soil
- Water
- humans (anthroponosis)
- animal (zoonosis)
*
Infection
Agent reproduces inside a person.
Infectivity
infected/ # exposed (and suspectiptible)
Pathogenicity
ill/ # infected
Virulence
with severe illness or death / # with symptoms
* High virulence is associated with high CFR
Disease Control types: Control
Limit infection within area
Disease Control types: Elimination
NO new cases of the infection in region
Disease Control types: Eradication
No new cases anywhere in world
Extinction
Not disease in world, even stored samples
Epi Curve types
- Point source epidemic: everyone exposed at same time (one peak)
- Continuous common source epidemic: constant exposure (plateu shape)
- Propogated outbreak: person-to-person spread (eries of peaks
Outbreak Steps
- Case Investigation - define cases
- Cause investigation - verify diagnosis
- Control measures (implmented ASAP)
- Conduct analytical study
- conclusions
- continue surveillance
- Communicate findings
Steps for Primary Resaerch Study
- Generate Ideas (EDPs)
- Choose study design
- Choose source population & sampling method
- Draft protocol
- Design/refine questionaire
- get IRB approval
- Recruit participants
- Collect data
- Clean/enter data for analysis
- Analyize data
- Write and report
