Prev Med Flashcards
Epidemiology
Studying distribution and determinants of dz freq in human pop
Key components of epi
Dz distribution (dz patterns), dz determinants (preventive or casual factors changing person’s health), dz control (surveillance)
Pop at risk
Specific pop of individuals truly capable of acquiring condition or event of interest. Pops can be classified as open/dynamic or closed/fixed
Closed pop/fixed cohort vs open pop/dynamic pop
Members of pop = constant vs members of pop changes (adding or losing members)
What is a sample?
Subset/group of pop to represent the entire pop
Rate vs proportion vs ratio
How fast; freq of event in a defined time; incidence, mortality rate vs how much; freq of event in defined pop; prevalence vs relationship b/w 2 groups and health income; odds ratio
diff b/w Rate vs risk
Estimates risk if event occurs once per person, proportion of event <5%, short time interval vs based on chance; the probability that an event will occur
how to find Measures of dz: counts vs proportions/prevalence vs rates/incidence
Measure of dz freq, number of cases or health outcomes being studied, no denominator vs # of dz cases in pop in specified time/ # of ppl in pop in specified time vs # of new dz cases in specified time/total # of ppl at risk in specified time
How to measure mortality rate vs infant mortality rate
of deaths in 1yr/# of persons in pop at mid year vs # of deaths under 1yo in specified time/# of live births in specified time
number of deaths in a year/# of ppl mid year vs # of deaths <1yo in a specified time/# of live births during that specified time
Attack rate
Similar to incidence, used when dz = observed for short time period like food outbreaks. # of new cases during specified time/total # of ppl at risk during specified time
Case fatality rate
Number of deaths d/t dz —> measures lethality. # of ppl dying of dz during specified time/total # of ppl w/ dz
Relationship b/w prevalence, incidence, duration
P = I x D
Primary vs secondary data sources
Orig research/findings/project, gathers new data not collected before (researchers collect data themselves) vs docs analyzing primary sources, gathers existing data (researchers don’t collect data)
Common sources for data collection
Surveys (pop based regionally or nat’lly), healthcare provider based (PE, EHR, clinical pop), registries (pt org, health ministry, public health), administrative data (enrollment/eligibility, claims)
2 examples of data collection methods
Counts (individuals tally dz or target info) or sampling (subset of reference pop)
Impt considerations of health data
Objective, collecting procedures, data completeness, timing, type of data, size of sample, primary or secondary data
Strengths vs weaknesses of health data
Better surveillance, assess dz trends, allocate resources, develop health policy and scientific inquiry vs not all data = freely avail, time frame of data, incomplete data, different data collecting methods
Surveillance definition
Ongoing systematic collection, analysis, interpretation, dissemination of health data to plan, implement, and eval public health
4 types of surveillance: active vs passive vs syndromic vs sentinel
Health depts collect info from labs, drs, healthcare; complete and accurate reports vs labs, drs, healthcare report to health depts; case reports based on standard case definition; deaths reported on standard certificate vs using health data to ID dz clusters early before dx and report to health depts; uses “real time” vs collecting and analyzing data by designating institutions for their location or specialty —> dx and report high quality data
Types of research
Bench science, clinical, primary care, pharmaceutical, public health
Purpose of research
Advance sci knowledge using scientific method: lit review, follow protocol, have research question, test/analyze, share results
Community Health Assessments (CHA)
ongoing process of ID health status, needs, assets of a community —> find opportunities and priorities to improve; requirement for tax-exempt hospitals and gov’t health; don’t follow scientific method but does follow standardized protocol and have goals
Main challenge of health data overload
No quantity but quality: which data is useful?
Use of surveillance
Estimate magnitude of problem; determine geo dz distribution, dz hx, epidemics, changes in health practices; make hypotheses to do research; eval control measures; facilitate planning
observational studies
study wider range of exposure than experimental studies; studies causes, tx, prevention
2 types of observational studies: descriptive vs analytical
when little info is known about dz –> find potential associations (hypothesis); objective = estimate dz freq; variables examined = person, place, time (who/where/when = affected?) vs when enough info of dz = done –> answers “why”/cause of associations and additional questions about dz –> new data (why/how pop = affected?)
types of descriptive studies from least evident to most evident
case report/series, cross sectional studies, ecological studies
case report vs series
single occurrence of incident, reports new/unique findings (new dz, adverse rxn, possible link b/w exposure, tx outcome) vs report on characteristics of a group of ppl w/ certain dz, no comparison group –> just look at 1 group; can also be based on 1 person
advantages vs disadvantages of case report/series
ID new clinical issues that lead to hypothesis, educational vs depends on avail and accuracy of data; cases = not generalizable; not based on systemic studies; no comparison group; can’t make tx decisions
cross sectional study
exposure and dz status w/in well defined pop = measured at one point in time –> one time assessment (surveys/questionnaire, lab tests, physicals); compare dz prevalence b/w those exposed and not exposed
advantages vs disadvantages of cross sectional study
cheap and quick to conduct, outcomes and risk factors = assessed, good generalizability vs difficult to determine temporal relationship b/w exposure and dz, may have high prevalence from long duration cases
how to ID cross sectional study?
look for one time frame, no f/u, no intervention, only analyze prevalence, can ID risk factors and dz they studied
ecological study
studying freq of characteristic and outcome in a pop, not individual or location; often uses scatter plots or Pearson correlation coefficients
advantages vs disadvantages of ecological study
easy to do at global lvl –> give quick answers to group phenomena vs ecological fallacy –> apply findings of pop to individual
types of analytical studies from least evident to most evident
case control studies, cohort studies
case control studies vs cohort studies
examine groups based on dz status (dz’ed vs control group) –> follow those 2 groups –> past exposures determined in each group; uses indirect measure of risk –> no incidence measured –> use odds ratio vs examine groups w/ or w/o specific exposure in pop w/o disease at beginning of study
advantages vs disadvantages of case control studies
good for rare dzs or that have long indux and latent periods; can eval mult exposures, quick and cheap, small sample size = okay vs investigates only 1 dz outcome, not good for rare exposures, can’t directly determine incidence rates or temporal relationship b/w exposure and dz, vulnerable to bias b/c retrospective (recall bias)
case vs control in case control studies
sample should reflect all cases in one pop, should not be selected b/c of exposure vs comparable to cases in every way except they don’t have dz –> must be chosen independent of exposure
retrospective vs prospective cohort studies
examine cohorts from the past to present; cheaper, faster, good for dzs w/ latent period, past exposure data = inadequate vs examine cohorts from the present to future; more expensive, slower, not good for dzs w/ latent period, better exposure data, less bias
how to find good sample size for cohort?
based on exposure status and follow them, pick a well defined pop –> do questionnaires, lab tests, physicals, procedures, existing records to gather data
advantages vs disadvantages of cohort studies
directly determines incidence and risk, relationship b/w exposure and dz (better if exposure = rare), follows logic of clinical question, vs needs large sample size, long time to complete, not good for rare dzs, expensive, validity affected by bias d/t subject attrition and loss to f/u
nonrandomized experimental study: quasi experimental study
study pop –> current tx (control) or new tx (experimental) –> improve or not improve for both
randomized study: random controlled trials
researcher takes large sample size and uses random assignment to divide them into smaller groups, subjects = blinded to their group membership –> 1 group w/ real med, other group w/ placebo
equipoise
randomization = ethical when no compelling reason to believe either of randomly allocated txs = better than the other
types of experimental studies: preventive vs therapeutic vs individual vs community vs cross over vs parallel vs simple vs factorial
investigating a measure that prevents dz vs investigating a measure that tx dz vs tx given to individual vs tx given to group vs planned reversal of experimental and control group throughout trial –> each person has own control vs everyone has same study tx vs each group has tx consisting of one component vs each group has tx consisting of 2+ components –> answers more research questions
overall conduct of experimental studies
Hypothesis → participant recruitment → participants allocated to groups → participants’ outcome mentored → rates of outcomes are compared → conclusions and implications
absolute risk
risk of developing dz over time; same as incidence
measures of association/effect
measures b/w exposure and dz that are compared to risk –> represent diff concepts of risk or diff magnitude of risk
attributable risk. =0 vs <0 vs >0
diff in risk b/w exposed and unexposed group: incidence in exposed minus incidence in unexposed; also use 2x2 table. exposure = NOT assoc w/ dz vs dec in dz risk –> protective vs inc in dz risk –> harmful
relative risk. =1 vs >1 vs <1
compares incidence in exposed to unexposed –> incidence in exposed/incidence in unexposed; also use 2x2 table. exposure NOT assoc w/ dz vs exposure assoc w/ inc risk of dz vs exposure assoc w/ dec risk of dz
odds ratio. =1 vs >1 vs <1
compares odds of case being exposed to odds of control being exposed –> odds case exposed/odds control exposed. dz = NOT assoc w/ exposure vs exposure –> inc odds of dz vs exposure –> dec odds of dz
absolute risk reduction vs relative risk reduction
excess risk assoc w/ exposure compared to control vs how much risk = reduced in experimental group to control group
allocation concealment vs triple blind
investigators won’t know which tx groups will receive vs participants, investigators, analysts don’t know GROUP MEMBERSHIP
intention to tx vs per protocol
how well the intervention worked between assigned groups, irrespective of participant adherence to control/intervention vs compares outcomes of those who did or did not receive treatment, regardless of assigned group (cross-over); loss of randomization
comparison groups: control vs usual vs placebo/sham
no intervention; Hawthorne effect - pts have tendency to change behavior when they are target of interest vs compares expt’l tx to known tx vs compares expt’l tx to placebo
noncompliance in random controlled trials: loss to f/u or overt noncompliance vs covert noncompliance vs contamination
officially notifies investigators and drops out of study vs participant stops or modifies assigned treatment without notifying investigators –> do compliance checks to elim vs control group does intervention
Number needed to treat (NNT) vs Number needed to harm (NNH)
number of patients needed to treat to prevent one patient from having an adverse event over a predefined period of time vs number of patients needed treat for one patient to experience an adverse event.
systematic review vs meta analyses
qualitatively answer a single, focused question by summarizing original research vs use statistical methods to combine results of individual studies and provide a quantitative answer within the context of a systematic review; includes insignificant data
strengths vs weaknesses for systematic review
answers single focused question, qualitative vs publication bias –> excludes nonsignificant findings, unpublished data, grey literature/ fugitive literature, disagreement on findings
strengths vs weaknesses for meta analyses
adeq sample size/power, detect tx complications and effectiveness, quantitative vs publication bias, few studies –> high risk of false neg; doesn’t include biology of disease or clinical experience
fixed effects vs random effects model
studies answer the same question –> answers may differ by only chance vs studies answer diff questions but form a fam of studies answer similar question; studies = considered a random sample
Forest plot
Shows the point estimate of effectiveness and confidence interval for each study in the review; values to left of null line –> favor expt’l group, values to right of null line –> favor control group
PRISMA checklist
minimum set of items for reporting in systematic reviews and meta-analyses
How to fix publication bias?
Encourage releasing all results, registration for all clinical trials, Cochran collaborators. Funnel plots can assess the bias
PICO
Pt, intervention, comparison (other tx, placebo), outcome
which measure of association would you use for cohort vs case control vs expt’l design?
relative risk, attributable risk vs odds ratio vs NNT, NNH, ARR, RRR
what is CONSORT criteria used for?
reporting results: consolidated standards of reporting trials
