Prev Med Flashcards
evidence based clinical practice vs medicine
clinician eval strength of evidence & use it in best clinical practice for pt vs incorporating clinical expertise, best external research evidence, & pt preferences
steps in evidence based clinical practice
1) ask focused question
2) find best evidence
3) appraise evidence validity & applicability
4) apply evidence to pt
focused clinical questions: PICOT
pt/problem = describe pt demographics, primary problem, coexisting conditions. intervention/prognostic factor/exposure = what do you want to do for pt?. comparison = are there alternatives to compare intervention to?. outcome = what do you hope to improve or accomplish?. time = what time interval is considered?
type of PICOT questions: dx/prevention vs therapy vs prognosis vs harm/etio w/ ideal study type
how to pick & interpret dx tests –> prospective, blind = gold standard, randomized trial, cross-sectional vs how to pick tx that does good > harm –> randomized ctrl trial, cohort vs how to estimate pt’s clinical course based on other interventions & dz complications –> cohort, case ctrl vs how to ID causes of dz –> cohort, case ctrl
what to think when appraise for evidence validity vs applicability
what were the results & were they valid/truthful? lg tx effect? bias? lg loss to f/u? equal & comparable groups? vs were study pts similar to my pop of interest? were all clinical outcomes considered? are likely txs worth potential harm/costs?
how to apply evidence to your pt
shared decision making approach –> integrate evidence w/ clinical expertise, consider pt’s goals & values, consider costs & AE, communicate, monitor results & outcomes
dz vs pt oriented evidence
pathophysiologic markers, labs, PE vs directly related to pts’ experience of their illness (mortality, morbidity, QOL)
statistical vs clinical significance
observed w/ trivial outcomes vs matter of judgment & depends on mag of effect being studied
narrative vs scoping vs systematic review vs meta analysis
comprehensive objective analysis of current knowledge on topic vs looks at emerging evidence & research while still unclear what other specific questions can be addressed vs high lvl overview of primary research on specific research question that identifies & eval all research evidence; method to combine individual studies vs quantitatively measure combined effect of all similar studies
study designs for systematic review
analytic studies (RCT, cohort, case ctrl), need to have measures of effect like RR, OR to compare groups, strong design & few flaws; don’t do descriptive studies (case report/series, cross sectional)
attributable risk vs relative risk. absolute risk reduction vs relative risk reduction
incidence in exposed minus incidence in unexposed vs compares incidence in exposed to incidence unexposed. excess risk b/w pts undergoing intervention & pts who aren’t (ie. how many pts = spared adverse outcome when taking tx) vs proportion of risk reduction by doing intervention. Lec 1, slides 23-24
number need to tx vs harm
of pts dr needs to tx to prevent 1 pt from having adverse outcome in predefined period; 1/ARR vs # of pts who need to be txed for 1 pt to experience adverse outcome in predefined period; 1/invARR. Lec 1, slide 25
what’s clinical research? device research?
determines safety & efficacy of meds, devices, diag products, txs intended for human use. determine mechanical fxn, safety, effectiveness & clinical utility of investigational device including biocompatibility, biodegradability, malfxn
class 1 vs 2 vs 3 medical devices. rigors of research required for device depends on?
cardiovasc, GI –> stethoscope, enema kit vs cardiovasc, neuro –> EKG, EEG vs ENT, OBGYN –> implantable prosthetics, IUD. class, risk, predicate of device
phase 1 vs 2 vs 3 vs 4 clinical trials
exptl drug/tx in sm group to eval tx safety, dose range & side effects vs exptl drug/tx in lg group to eval safety & efficacy vs exptl drug/tx given to lger group for safety, efficacy, side effects & compare to other txs vs FDA approved, post-marketing studies to find additional info on exptl drug/tx’s risks, benefits, best use
why do clinical trials? how to protect ourselves & pts during trials?
discover new txs for dzs; detect, dx, reduce chance of developing dz. document & train; clinical data management generates reliable statistically sound data; informed consent, Human Subjects Protection, IRB, FDA
adverse events vs serious AE/ADR vs unexpected ADR
unintended & unfavorable s/s torwards pt/subject when using exptl drug vs any dose resulting in death, life-threatening, requiring hosp, disability, congenital defect vs adverse rxn not consistent w/ applicable production formation
common rule of HHS vs IRB vs FDA
requires researchers get informed consent from volunteers –> give info about the study, risks, benefits vs ensure researchers follow HHS rules & ethical guidelines vs protects public health by ensuring safety & efficacy of drugs, bio products, med devices
corrective actions to resolve common issues: src documentation
orig place where data was recorded; hosp/clinic, EHR/charts/img/labs; memoranda, subjects’ diaries, evals, pharm dispensing records
corrective actions to resolve common issues: product accountability
investigational product = preventive, therapeutic, diagnostic or palliative product in clinical trial; can be used in un/approved forms. record investigation product that was used only in clinical trial for participants and adheres to protocol; chain of custody: PI = responsible for accurate accountability log
corrective actions to resolve common issues: consent. regulatory docs.
“concise & focused” informed consent document to give subjects info they need to make a decision to volunteer. signature logs, study personnel edu, monitor reports, financial disclosure
Good Clinical Practice
int’l ethical & scientific quality standard for designing, conducting, reporting trials that involve human subjects; compliance gives public assurance that rights & safety of pts = protected
how is GCP demonstrated?
study registration, GCP training for those involved in conducting trial, Human Subjects Research Training, IRB submission & approval
Good Laboratory Practice
guidelines governing process, organization, conditions of lab studies
community health research per NIH vs community based participatory research
addresses dzs & conditions disproportionately affecting community’s health; promotes collab b/w healthcare, scientists, & community leaders to facilitate research advances vs equitable, partnered approach to research involving community members, representatives, & researchers
health program vs community interventions
planned activities to accomplish specific health related goals vs small pervasive changes applied to subpop (more specific community)
health belief model vs social cog theory vs stages of change model vs theory of reasoned action vs social eco model
individuals’ perception of threat, health problem, benefit of avoiding threat, decision to act vs personal & environ factors & human behavior influence e/o vs individuals’ motivation to change problem behavior vs individuals toward behavior of ease or difficulty to change vs change behavior guided by social environ (mult lvls: individual, interpersonal, org, community, policy)
constructs of social cog theory
behavior, self-efficacy, expectations, observational learning, modeling, self-reg, rewards/reinforcement
stages of change: cog vs behavioral stage
focus on beliefs, attitudes, motivation, thoughts; pre/contemplation vs assist pt to make & implement plan; prep, action, maintenance
asmptns of stages of change model
pt behavior change happen over time thru seq of stages; most pts not prepared to act –> won’t be served effectively by traditional advice; specific counseling strats = emphasized at specific stages; nudge pt from stage to stage; relapse = nml part of change process
theory of reasoned action
build intention to inc likelihood of behavior by influencing their attitude, perception, dis/approval, action
how to promote healthy communities?
health edu to implement health promotion & dz prevention programs tailored to target pop
program planning in community research
- social, epi, edu & eco, admin & policy assessment
- implementation
- process, impact, outcome eval
Precede-Proceed Model: 1-4 = get info for planning decisions; 5-8 = detail program development, implementation, eval
program eval: formative vs process vs outcome vs economic vs impact eval
use during developing new program or modifying existing program; see if proposed program = needed & accepted by target pop; useful for any needed mods vs use when starting program or existing program; see how well program is working, accessible, acceptable; gives early warning for problems vs use when made contact w/ 1+ target; see if there’s effect on target vs use when starting program or existing program; see what resrcs = used & there costs/outcomes vs use during existing program or end; see if program reached its main goal; for future funding & policy
grounded theory vs ethnography vs phenomenology vs story telling vs focus group vs survey development vs community assessment
analysis & making theories of own data collection vs study ppl in their environ –> obs, face/face interview; classic ethnographic research vs reflecting one’s one XP vs shape life events into stories vs 6-8ppl talking abt topics guided by moderator vs match survey design w/ survey objectives; concept should reflect nominal, ordinal, interval or ratio-appropriate type of analysis; give info abt what’s impt to community –> narrows scope of public health project vs ID needs & priorities w/in community, can utilize above methods
validity (accuracy) vs reliability (precision). know graph on Lec 4, slide 5
systematic error; degree to which data measures what it intends to measure vs nonsystematic/random error; when rpted measurements by diff ppl at diff times get similar results
srcs of systematic error
selection bias, information bias, confounding
what is bias?
any systematic error in design, conduct, analysis of study resuting in incorrect/invalid estimate of measure; pos –> overestimate findings, neg –> underestimate findings
selection bias. can you fix it & how?
from procedures that have diff results from chosen subjects v from eligible ppl in pop; retrospective cohort, case ctrl. nope not during study –> use same criteria for selecting cases & ctrls, get all participants’ records, get high participation rates, take diag & referral patterns of dz
selection bias: ctrl selection bias vs self selection bias vs differential selection bias vs loss to f/u bias
selecting inappropriate ctrl group vs refusal, nonresponse by pts to participate related to exposure & dz vs tendency to dx, hosp, refer pts differently based on exposure/dz status vs loss to f/u –> systematically diff pop from orig/target pop
observation bias. can you fix it & how?
during collection of data or after participants entered study. blind interviewers
observation bias: recall bias vs interview bias vs procedure bias vs observer expectancy bias
ppl who had AE = more likely to recall previous risk factors than ppl who never experienced it vs systematic diff in soliciting, recording, interpreting info vs subjects in diff groups = txed differently based on exposure and/or dz status vs researcher’s belief in efficacy of tx changes outcome of tx
overall methods to minimize bias
- select dzed ctrl group
- design structured or self-administered questionnaire
- use mult biological measurements/markers
- mask interviewers & participants to study’s hypothesis
- use standardized methods of outcome
criteria for confounding. pos vs neg confounding?
X = assoc w/ dz B, X = assoc w/ risk factor A, X = NOT a result of factor A. exaggerates true association vs hides true association
design vs analysis phase for ctrlling confounding
group or individual matching on suspected confounding factor; restriction –> limit study subjects to those w/ certain characteristics (all young ppl, all men) vs stratification, standardization, adjustment
effect modification
when 1 group = affected by a 3rd variable (kinda like confounding but that’s across all groups)
association vs causation
statistical relationship b/w cause & effect; observed but doesn’t mean causation vs change in exposure –> corresponding change in outcome; inferred (ie. difficult to prove causal relationship in epidem)
characteristics of cause
association, time/temporal relationship (must precede effect ie. event must precede dz or else dz wouldn’t have occurred), direction (asymm relationship b/w cause & effect; pos = presence induce dz, neg = absence induce dz)
causal inference. 2 steps?
process of how epidemiologists determine causative & preventive dz factors.
1) is result valid & true?
2) assess whether exposure has actually caused dz
what’s risk factor?
exposure, behavior, attribute that inc probability of dz
sufficient cause vs component cause vs necessary cause
set of conditions w/o any 1 of which the dz would not have occurred (whole pie) vs any 1 set of conditions that’s necessary for completion of sufficient cause (piece of pie) vs a component cause that’s a member of q sufficient cause
types of causal relationships: directly causal association vs indirectly causal association vs noncausal association
factor exert its effect in absence of intermediary factors vs factor exerts its effect thru intermediary factors vs relationship b/w 2 variables = stat sig but no causal relationship b/c temporal relationship = incorrect or another factor responsible for cause/effect
causal “guidelines” for est causation per Sir Austin Bradford Hill
1) temporal relationship
2) strength of assoc
3) dose-response bio gradient
4) replication of findings
5) biologic plausibility
6) analogy
7) reversibility
8) consistency of findings
9) specificity of assoc
which checklists are used to guide critique? describe abstract vs intro
CONSORT, PRISMA. succinct, accurate summary; context, obj, methods, results, conclusion vs sets up the question, context (researcher’s motivation, study’s contribution, clinical imptance), objective (goal, hypothesis, funding)
describe methods 1: study type vs population
limitations of study type vs sample size (how it was determined), sampling procedure (recruitment & any selection bias), in/exclusion criteria, case v ctrl for comparability
describe methods 2: data collection vs pt safety
primary exposure (minimize bias? when/where collected?), outcome (was it appropriately measured?), reduce confounding (randomization, restriction, matching) vs IRB, safety eval (pt safety monitored? when?), power analysis (ability to finding sig diff when real diff exists), statistical stability (p value, 95% CI), associations (rate/risk, RR, OR)
describe results vs discussion vs conclusion
major results, strength of relationship; do results match what authors said? any new/unanticipated results? bias or confounding? drop out rates? vs address limitations, extrapolation, generalizability/external validity (does other lit dis/agree?), statistical v clinical sig vs implications supported by evidence in data, future research & policy recs
cochrane library vs medline vs uptodate
systematic reviews, meta analyses vs national lib of med vs by physicians/for physicians, cont updated, peer-reviewed
internal/comparability vs external validity/generalizability. effects of each?
extent to which study results = true for target pop; reflected by selection/randomization vs extent to which study results = applied to pop not involved. loss to f/u, lack of randomization vs loss to f/u, low response rate, narrow inclusion criteria
sensitivity vs specificity. pos predictive value vs neg predictive value
rule out; true pos/(true pos + false neg) vs rule in: true neg/(false pos + true neg). probability of pos pt actually have dz; high = pt really has dz vs probability of neg pt actually doesn’t have dz; high = pt really doesn’t have dz
what are ROC curves? what does the diagonal line mean?
explains relationship b/w sensitivity & specificity; X = 1-sens, Y = sens. relationship b/w true pos & false pos rates for useless test –> gives on additional info than what’s known before test was performed
likelihood ratio. advantages? know the strength
describes performance of diag test –> probability of test result in ppl w/ dz/ w/o dz. for mult lvls & compared to other data, go beyond test as ab/nml, gather info for range of values, help choose diag test, help calculate post-test probabilities. Study Guide pg 24
asmptn of independence
validity of final results depends on info in each test & independent from info of preceding test
what other tests can we use for validity? –> parallel vs serial/sequential
perform several tests all at once and any pos = evidence for dz –> maximizes sensitivity & NPV vs perform several tests consecutively based on preceding results, ALL tests must be pos to rule in dx –> maximizes specificity & PPV
define screening. indicate characteristics of good screening test
presumptive ID of unrecog dz thru rapid exams/tests/procedures –> ID asx ppl, early detection & tx; NOT a diag test. suitable dz (highly prevalent, progressive, effective tx at early stage), suitable test (simple, fast, cheap, safe), suitable screening program (using specific test on specific pop for specific dz –> reduce morbidity/mortality
lead time vs length time vs overdx vs referral bias
screened pts live longer b/w dxed earlier in asx phase but time of true onset to death = same (ie. early detection does not mean inc survival) vs screening catches long preclinical periods but dz can have short or long periods vs when excitement of screening leads to more tests & positives –> false pos pts inappropriately impact survival rates vs pts seeking preventive care = healthier than pts who need attn for acute probs
ethical considerations for screening
cross-cultural differences in health literacy & practices; geographic disparities (insurances, minorities); genetics screen (good for fetus/children; bad if costly/ins, stress, DNA ownership)
o Upstream:
o Midstream:
o Downstream:
of community health research
-structural/societal determinations (redlining causing neighborhood SES inequality)
-intermediary determinants (need for parks, grocery stores, etc)
-medical intervention & providing care
Rules of P’s & N’s
● Increased Prevalence = increased PPV , decreased NPV
● Increased specificity = increased PPV
● Decreased prevalence = increased NPV, decreased PPV
● Increased sensitivity = increased NPV
