Cheatsheet for exam PDF

Preview

Summary

two page cheatsheet front and back...

Description

Epidemiology Def: Distribution & determinants Of frequency in human pop & appli. To control health prob Selected sub-disciplines Disease, Exposure & pop Hypothesis Testable statement, Set of observation Reject/not r Two Types of Epi Descriptive (Describes disease patterns) PERSON, PLACE,TIME,OUTBREAKS 1. Monitor public’s health 2. evaluate success of intervention programs 3. generate hypo about cause of disease Identify and count cases of disease in pop And conduct simple studies - Case report/case series/cross-s/ecologic Analytic/scientific (search for cause/prevention) 1.Evaluate hypo about causes of disease 2.Evaluate success of intervention programs Compare groups & systematically determine: is there an association? - Clinical trial/experimental/CC/cohort *descrip – cannot identify the cause of disease main use – identify prob/trends/high-risk, public health planning: Spending, Generate hypo Cancer is the main cause of death 4 components for measuring disease frequency 1. population 2. Cases of disease (numerator) 3.size of population (denominator) 4. Time *catchment area – service pop. of a medical facility Fixed pop – defined on the basis of some event membership is permanent (veterans of Vietnam war, people born in 1982, enrolled in study) Dynamic – defined by being in a state or condition Membership is transient (resident of Ottawa, Parents of a teen) 3 generic types of measurement RATIO - # don’t have to be related PROPORTION – division of 2 related #’s Usually % RATES – division of 1 # by another Time is Intrinsic part of denominator Measures of Disease Frequency 1. Prevalence 2. Cumulative Incidence 3. Incidence rate Prevalence Measures presence of Existing cases of Disease. Involves being in a state… denominator Is total population. Proportion of pop. who have Disease POINT PREVALENCE = proportion of pop. That has the disease at a single point in time Prevalence = # existing cases at time point/ # in TOTAL POP. at time point Incidence Measure occurrence of new cases, involves Transition from 1 state to another… dem includes Pop. at risk Cumulative Incidence – time (during a specified time is described in words that go along w/# denomi=(pop. at risk) Incidence rate –time is an intrinsic part of the denomi *no loss to follow up * Cum I = # NEW cases during time period/ # in POP AT RISK at start of time period Attack rate – proportion of individuals exposed to an Infectious agent who become infected over a certain time ----Same formula as CI Case fatality rate – proportion of individuals w/ a disease who die of the disease # death due to disease X/ #people w/ disease X *CI not good for dynamic population , doesn’t consider time of occurence* Incidence rate Measures Speed at which new cases occur involve

transition Dem = pop. at risk – Do NOT assume group observed has been followed IR=true rate *IR = # NEW cases during time period / total person-time of observation in pop. at risk *Relationship between prevalence and incidence -have been an incident case at some point -Still have the disease so duration (time living w/disease is also involved) D=P/IR D= average duration of disease ends w/death or Recovery Crude mortality (or death) rate - # of deaths from all causes Per 100,000 -- crude=raw Cause-specific MR - # of deaths from a specific cause per 100,000 **same goes with age Years of potential life lost - # of years an individuals was Expected to live beyond his/her death Livebirth rate – livebirths/1000, must show evidence of life Infant mortality rate - # of deaths of infants less than 1 yr/1000 livebirths per yr Birth defect rate per 10,000 include live and still births Measures of Association Absolute measure  diff b/w 2 measures of disease freq Ex. travel 30 mph over the speed limit [0 = no asso] Relative measure  ratio of 2 measures of = Ex. you were traveling twice the speed limit [1 = no asso] CI = Risk Difference IR = Rate difference Excess Relative Risk = (RR-1) x 100% RD=ABSOLUTE MEASURE  Represents measure Of public impact. On people and prevention (impactful?) RR=RELATIVE MEASURE  strength/magni of asso Population Attributable Proportion (PAR%) Assess how much disease attributes to exposure if Caused by disease Difference Measures = Rt-Ru/Rt x 100 Ratio Measures = [Pe (RR-1)]/[Pe (RR-1)+1] x 100 To get Pe = #exposed/total pop. Prevented fraction PF = (Ru –Re)/Ru x 100 Population rate difference (PRD) Rt-Ru or RD x Pe Disease Cluster  aggregation of relat uncommon Events/disease in space/time in amount greater Expected by chance CASE REPORT  one patient; usually new or Unusual symptom or problem CASE SERIES  group of patients w/ same symptom Or problem; usually new or unusual ** useful –reco & des of new disease, drug side effects, insight into disease mech, info for hypo **No explicit comparison group ECOLOGICAL STUDIES  examines rates of disease to Pop.lev. factor (exposure and outcome data..group lev) **bias – due to pop. No link b/w expo/dis, need individ effects, can’t adjust for CF **+ -- inexpensive/fast on available data, for early knowle, wide range on expo (internat), easy for correl coeffi/line regress CROSS-SECTIONAL/SURVEYS  study/survey exam relati Onship b/w expo&dis at single point in time Takes snapshot, measures expo pre/dis prev, are many Government surveys **lim – can be confused with which came first ex. physical inactiveCHD (interchange) **good – immutable charac, measure of long-term expo, historical exposure, relatively quick & inexpen, generalizable *Prevalence is not ideal for etiologic research b/c it combine incidence and duration CRUDE RATE = Sum of (% of pop in age group) x (age-specific rate) Age-adjusted rates needs:1. Age specific rate 2. weights from standard population **depends on standard being used not age no real, only good for comparison ** direct standardization – removes unwanted usually differences b/w pop. Components of a Study

1. Population 2. Exposure 3. Outcome 4. Potential Confounders 5. Analysis 6. Communication of Findings POPULATION -- Source pop.  Interested in Knowing more about / Study pop. Enroll in ur Study to represent the source pop. OUTCOME – descript of event you studying EXPOSURE – determinant of interest upon outcome CONFOUNDER – extra determinant of an outcome Study Designs – Experimental OR Observational COHORT STUDY – subjects selected according to expo Lev.& follows them for disease occurrence ** not ethical, too expensive, large effect expected, expo rare, (exposed group vs. comparison) **Prospective – exposed/ue at start of study to future (more expensive, time consuming, not efficient for disease w/ long latent periods, less vulnerable data) Retrospec – e/eu in past and compared incidence of (cheaper, faster, efficient w/ = exposure and confounder data may be inadequate, vul ===) Disease at start of study CASE-CONTROL – have disease and compared to Source pop. That gave rise to disease and exposure Is compared **not ethical ==** Types of experimental studies Parallel – each group receives 1 treatment admin concur Crossover – = all treatment 1 after another, differs each Group SIMPLE – 1 TREATMENT FACTORIAL 2/MORE TREAT HALO EFFECT & HAWTHRONE EFFECT – participants Consciously/unc change their behavior just b/c they are Being studied *counterfactual refers to what would have happened to an individual in absence of exposure Sources of comparison group INTERNAL COMPARE, COMPARE COHORT GEN POP COMP Cohort – clear temporal sequence, efficient for rare expo Good info on expo, cf, can study multi outcomes BUT may need large # of subjects to follow for long Periods, expensive, time consuming, not good for rare Disease CASE-CONTROL  Selection 2 necessary requirements 1. controls must come from the same source pop. As cases 2. controls must be selected independently of exposure ODDS RARIO = (ad/bc) BOX e-ue c|nc **less people, time and money, useful for disease w.. little info ** lim – limited to studying a single outcome, inefficient for rare exposures, more opportunity for systematic bias maybe be unsure about temporal sequence, can’t calc absolute measure of association Bias  leads to incorrect est. of association *creates/masks association when there is none/exists direct of bias - + assoc is biased TOWARDS (1.4/0.7) = - + assoc is biased AWAY from null (2.6 pass T2.0/0.3 T0.5) S Bias – ppl who agree to take part in study are diff from the Source pop. (at time of recruit/during process of retaining ppl) TYPES  CC – 1. inappropriate control selection 2. Differential Participation (CC, Co) 3. Diff loss to follow up (CO, experi) Control –select  occurs if controls are more/less likely to be Selected if they are exposed (use identical select criteria) Diff Part  occur if willingness or ability to participate Is Related to both exposure and disease status (obtain high Participation rate from all groups) Diff loss to fu  study ppl exit study for reasons both related to exposure and disease (main high part rates) Assessment of s-bias 1) describe possible sources 2) des likely impact on results (direct/magnitude) 3.des pos solu *Measurement (misclass) error is the most common form of bias non-diff misclass – missclass is the same (not diff) for both groups Recall bias – away from null / Poor recall – towards null

Confounding  systematic difference b/w the groups bring Compared that distorts the true association b/w an expo/dis Experi/Cohort – occurs when expo/unexpo groups differ by More than just the expo… differ by some other variable CC- when C&C have diff charac (occurs in other epi studies) Unlike bias, inherent characteristic of the population ** thought of as a mixing of effects (third variable) Counterfactual ideal (read above) epiologist select Different sets of people who are similar as possible 1. w/respect to other factors that can influence outcome 2.w/respect to collection of comparable & accur info **CF can be though as failure to come close to Confact occurs when risk in both e/ue does not equal CF MUST… independent predictor of outcome (risk Factor for ue, associated w/ expo, can’t be an intermediate On the casual pathway between expo/diesase CF can be controlled or adjusted through Design phase (RRM)/analysis phase(SSM) RANDOM – no limit on the # of CF, do not need info on CFs BUT limited to experi studies, less effective w/small sample RESTRICT – limit study to people who are w/in on catrgory **simple, effective control of charac BUT only good for known CFs, incomplete control of CF, cannot evaluate restricted variable, limits sample size & generalizability MATCH – select subjects so that CFs are distributed ident Ically among e/ue (cohort) or C & C (CC) **simple &effective, useful for variables thay are complex/ difficult to capture BUT good for known CF, difficult/ expensive/time-consuming to find matches, cant evaluate matched variable STRATIF – separate and calculate measure of associ **straightforward and easy to perform, effective control of charac BUT diff to control for many CFs simultaneously, diff presentat, continuo vari not easy 1. calc crude measure/2.divide subjects into strata of CF 3. calc strat-specific measure of associ/4. Calc adjusted measure of associ/ 5. Determine magnitufe MAGNITUDE OF CF (RRcrude – RRadjusted)/ RRadjusted x100 MULTIVARIATE REGRESSION – statistical model Describes association b/w exposure/dies/CF **simultaneously adjust for several variables BUT difficult to conceptualize RNDOM ERROR – Caused by measurement error, Sampling variability Non-systematic measurements can lead to error Statistical inference: process of making statements about A pop. Based on info from a single sample (accuracy Depend on validity of the study) 2 approaches for random error – hypo testing (p-value) & CI hypo testing 1. Specify a null and alternative hypo 2. use statistical test to determine compatibility of study results w/null 3. Decision to reject null based on p-va P-value – observing by chance alone only evaluate random error, calculate assuming Null is true interpreting P-values: P 0.05  large p-va, high degree of compat b/w ==, do not reject null, likely due to chance, not statistically significant **LIM: p-va mixed magnitude and sample size, and do not rule our bias/CF, over reliance on 0.05 CI are not CFd, separates magnitude and sample size Effect Measure Modification Not an error, but a biological process worthy of investigation OCCURS when stratum-specific estimates are meaningfully Different from one another called an effect modifier Epiologist use 2 methods to determine if the Stratumspecific results are different from 1 another: Visual inspection or a statistical test SYNERGY  + interaction is said to occur When the excess relative risk among individuals w/ Both factors is greater than the sum of the excess RR Of each factor considered alone

**happens when 2 factorswork in concert to produce more disease than 1 would expect “departure from additivity (low) or multiplicity (high)” ANTAGONISM  - interaction occurs when the excess RR among individuals w/ both factors is lee than the sum **mean that one factor reduces the impact or even cancels out the effect of the other factor Chapter 14 – Critical Review A. Collection of data 1. context of study 2. objectives (hypo being tested) 3.primary expo of interest ? was it accurately measured (misclss) 4. primary outcome == 5.type of study 6. Describe study base, process of subject selection, sample, and ratio of propositi to comparison subjects **Propositi are exposed subjects in an experi/cohort study, and case in a CC study 7. bias in the selection of study subjects? How likey? s.bias results from refusal, non-response, loss to fu to both exposure & disease 8.bias in the collection of information? 9. what provisions were made to minimize influence of confounding factors prior to analysis of data RRM + collect data on CF + ensure compared groups From same source pop. B. Analysis of Data 1. methods used to control CF bias during analusis SSM? 2.what measures of association was reported rate/risk ratio/difference, RR/OR 3. what measures of statistical stability was reported (p-value, CI) C. Interpretation of Data 1. major results (###) 2. Interpretation of these results affected bu info bias, s.bias and CF (use a7-a9 &b1) 3. interpt of results affected by any non-diff misclss both in magnitude and direction of misclss…. Very common and towards null 4. address limitation 5.authors main conclusions 6 can the results of the study be generalized CHAPTER 15 – CAUSATION 3 essential attributes of a cause 1. association (causal factor must occur toget w/its effect 2 time order (cause must precede effect 3. direction (asymmetrical relationship b/w cause & effect sufficient casue model multi-factorial causal model , pie pieces, all of the pie pieces need to fall into place for a disease to occur whole pie is called the sufficient cause, pie pieces are called component causes HILLS GUIDELINES 1. strength of association 2. Consistency 3. Specificity 4. temporality 5. Biological gradient 6. Plausibility 7. coherence 8. Experiment 9. Analogy 1) bias is less likely to explain observed association 3) falls apart for non-infectious expo (holdover form Koch) 4) must precede occurrence of disease, factors that are coincident w/ the disease and factors that are a result of the disease can’t be causes of disease 5)strength of the association increases as the expo level increases, the associ is more likely causal BUT ignores threshold effects, and experi lab studies 6/7) bio/social models should exist to explain an associ, associ should not conflict w/ current knowledge of natural history and bio of disease Screening  presumptive identification of an unrecognized disease/ Defect by the applications of test examinations or other procedures Classifies asymptomatic ppl as likely/unlikely *seems simple but complex, hidden cost and risk YOU NEED  SUITABLE DISEASE/=TEST/=SCREENING PROGRAM Suitable disease – serious consequences, is progressive, treatment Must be effective at earlier stage, detectable at pre-clinical phase NATURAL HISTORY OF DISEASE A-Pathological onset B -clinical symptoms, diagnosis/treatment C – remission D – relapse E –Death A-C =pre-clincial phase(30-60), B-C = detectable pre-clincial phase Lead Time: duration of time by which the diagnosis is advanced As a result of screening – b-c= 45-60

SUITABLE TEST Inexpensive, easy to administer, minimal discomfort has level Of validity and reliability Valid Test: does what it’s supposed to do, correctly classify ppl w/pre-clincial disease as +and neg Reliable Test: gives you same results on repetition Sensitivity – enables you to pick up cakes of disease a/a+c Specificity – enables you to pick out no diseased ppl d/b+d Vaalid test has high sensitivity and specificity *If criterion is low (point a), then sensitivity is good, but specificity suffers. If you criterion is high (point c, then specificity is good, but sensitivity suffers You want to determine if screening program is successful. Does it reduce morbidity and mortality? How to evaluate? 1. feasibility measures – acceptability, cost, predictive value of a positive test (PV+) (sens) and neg (spec) PV increases when sensitivity, specificity and disease Prevalence increases 2. Efficiacy measure of evaluation ** want to reduce morbidity and mortality bias when evaluating a screening program 1. volunteer bias, 2. lead-time bias, 3. length bias screening programs are evaluated by examining predictive value and outcome measures such as stage distribution and cause-specific mortality ETHICS 1974 Belmont report “cornerstone statement” 1. respect for individual autonomy 2. Beneficience 3. Justice informed consent is not a form..signature but 4 considerations 1. information exchange 2. Comprehension 3. Violuntariness 4. documentation 1)all info to potential research partcipants 1.states the study involved research, purpose,duration, descript 2.descriptoion of any foreseeable risk/discomforts 3. description of any benefits to subject. Others from research 4. disclosure of appropriate alternative procedures or course of treat 5. statement describing extent confidentiality records will be given 6. research involving more than minimal risk 7. contact for answers 8. statement that participation is voluntary 2) understandable info by seeking community input that fit culture, providing info in individual & group setting, minimize reading level, declarative statements, glossary, audio tape recording, use educators not doctor, time to process and ask questions, training o monitor informed consent comprehension by open ended questions, quizzes 3) setting in free of coercion a undue influence -benefits of research must not be overstated, cannot say no finances should not be offered, no rushing 4)approved by IRB, good clinical practise provison of informed consent be documented in participants study records...