Question 


Answer 


Narrative Review 
Selective review of the literature that broadly covers a specific topic Does not follow strict systematic methods to locate and synthesize articles 

Systematic Review 
Utilizes exacting search strategies to make certain that the maximum extent of relevant research has been considered Original articles are methodologically appraised and synthesized 

Metaanalysis 
Quantitatively combines the results of studies that are the result of a systematic literature review Capable of performing a statistical analysis of the pooled results of relevant studies 

Systematic reviews 
Overview of primary studies which used explicit and reproducible results Can be performed on group or single subject studies Can include both quantitative and qualitative analyses 

Advantages of systematic reviews 
Increased sample sizes Can control for betweenstudy variation 

Disadvantages of systematic reviews 
• Not a primary study Limited by validity of individual studies Subject to bias 

Metaanalysis 
Statistical analysis of results of several similar studies (aka quantitative synthesis) Type of quantitative systematic review, or included as part of systematic review 

Hierarchy of Evidence 
1 Systematic review
2 Random Controlled Trials
3 Cohort
4 Casecontrol series
5 Case series
6 Case reports
7 Editorial and opinions
8 Animal research and lab studies 

Stages and phases involved in a systematic review 
Planning the review (ID of the need for a review, preparation of a proposal for review, and development of a review protocol) Conducting the review (ID of research, selection of studies, study quality assessment, data extraction and synthesis) 

Forest plots 
Compares different studies Statistically different when 95% confidence intervals cross effect line 

Narrative reviews advantages 
Present a general overview covering a specific topic that provides primary information or an update, or both Fairly easy for novice authors to prepare 

Narrative reviews disadvantages 
May not provide the best available answers Findings are less reliable 

Systematic reviews advantages 
Present a comprehensive review of the literature based on all available research with regard to a focused research question Provide an estimate of the "true" answer to the research question 

Systematic reviews disadvantages 
Specialized expertise of reviewers is required Involve a formal research protocol Findings are only relevant to a single question 

Publication bias 
File Drawer Problem = only positive data presented, the rest are put in the file drawer In situ = some parts of the studies published 

Study homogeneity 
Similarity between studies, increases their ability to be compared 

Study heterogeneity 
Differences between studies Hinders comparison of subjects Study design Observed treatment effects more dissimilar than due to chance Statistical test can estimate and account for this Treatments that work across, elevate confidence 

Study heterogeneity 
Subgroup analysis (older v younger)= may be more valid, reduce stat power Meta regression analysis = analysis of hetero. between subgroups Sensitivity analysis = considers variation between factors other than subjects Cumulative metaanalysis 

IMRaD 
Introduction (What was asked)
Methods (How was it studied)
Results (What was found)
and
Discussion 

Introduction 
Objective = the exact questions asked 

Methods 
Design
Setting
Patients
Interventions
Main outcome


Results 
Key findings 

Discussion 
Conclusion = key conclusions including direct clinical application 

Incidence 
# of new cases in time period/population x 100,000 

Risk 
estimate of proportion of unaffected person who will develop the disease of interest over a specified period of time 

Odds Ratio (OR) 
odds of developing disease in exposed group / odds of developing disease in unexposed group 

Prevalence 
# of existing cases in a time period / population x 100,000 

Point prevalence 
proportion of population with disease at a given time (can miss episodic conditions) 

Period prevalence 
proportion of population that has disease within a defined period of time 

Causation in epidemiology 
3 key criteria:
Temporality (temporal precedence)
Consistency
Doseresponse 

Temporality 
A causes B
or
B causes A
or
X causes A + B
A not related to B; occurrence is a mere chance 

Consistency 
Reproduction of study results in different populations 

Dose response 
Greater exposure to risk factor leads to greater effect on health 

Bradford Hill's Criteria of Causation 
Strength of Association
Consistency
Specificity in the case
Temporality
Doseresponse relationship = increase dose > increase occurrence
Plausibility
Coherence
Experimental evidence
Analogy 

Cohort studies 
Longitudinal
Prospective = know patient's exposure, observing ahead for disease
Retrospective = know patient's exposure, looking back in time for disease 

Casecontrol study 
Similar to retrospective cohort study Disease or condition is known, looking back in time for risk factors 2 Types: Prevalent case (includes all persons) and cumulative incidense (only new cases) 

For casecontrol studies you must calculate ________ and not __________ 
Odds Ratio (OR), risk 

OR formula 
ad/bc a=exposed cases b=exposed controls c=nonexposed cases d=nonexposed controls 

Casecontrol studies advantages 
Good for investigating rare diseases Can be performed quickly and inexpensively Useful for studing disease with long latency periods Facilitate study of multiple potential cases at once Existing records can often be used 

Casecontrol studies disadvantages 
Typically rely on patients recall of past exposure Do not permit calulation of true disease rates in the population Difficult to validate information on exposure Other variables that may be associated with disease are not controlled 

Recall bias 
systematic differences between cases and controls in ability to recall past exposures 

Berkson's bias 
(Admission rate bias) type of selection bias where hospitalized cases are different than hospitalized controls 

Stratified analysis 
Cnsiders confounding variable (e.g. alcohol consumption and lung cancer) Looks at effect each independent variable has on outcome separately 

Cohort studies 
Follow a disease free group of subjects forward in time Some subjects are exposed to a risk factor and some are not Purpose is to see if there is a greater proportion of disease among those who are exposed to the risk factor 

Discreet vs Continuous variables 
Discreet> smokers vs. nonsmokers Continuous> cholesterol levels 

Inception cohort study 
Tracking cohort with early stage of chronic condition 

Cohort studies 
Less subject to bias than casecontrol studies bc exposure levels evaluated before disease develops Best design to determine risk level Better for studying relatively common diseases Most expensive type of epidemiological study, but cheaper than RCTs 

When an outcome in a research study is common (occurs in more than 10% of the unexposed group)... 
The odds ratio will tend to OVERESTIMATE the risk ratio 

Relative Risk (RR) in cohort studies 
RR = (a/(a+b)) // (c/(c+d))..........................RR > 1 = increase in incidence of disease in exposed group...................RR < 1 = protective effect of exposure (can be used to calculate placebo effect)


Attributable risk (AR) in cohort studies AND Absolute risk reduction (ARR) in cohort studies 
AR or ARR= a/(a+b)  c/(c+d) 

Relative risk reduction (RRR) in cohort studies 
Comparative reduction in rates of bad outcomes between experimental and control groups............RRR = ARR // c/(c+d) 

Number needed to treat (NNT) in cohort studies 
# of patients would would need to be treated in order to prevent one additional bad outcome......NNT = 1/ARR........related term = NNH (number needed to harm 

Cohort studies advantages 
Portray the natural history of disease Don not rely on patient recall Better for establishing a cause and effect relationship than casecontrol studies Less vulnerability to bias or chance Permit calculation of true disease rates in the population 

Cohort studies disadvantages 
Typically very expensive Many people must be followed to obtain enough with the disease Very timeconsuming Subjects frequently drop out of study over time Difficult to generate a control group to study very common conditions 

Case studies 
Usually retrospective, can be prospective, low validity but high clinical relevance, similar to IMRaD format but with case description rather than M & R (storied case report and evidence based case report) 

4 types of case studies 
1) Unique case 2) Unexpected association 3) Unexpected development 4) Unusual presentation 

Is a case study the same thing as a case report? 
NO, a case study is more in depth 

Purposes of case reports 
Detect rare conditions Educational value Learn how other doctors manage certain cases 

Limitations of case reports 
Susceptible to many biases Unable to test hypotheses Does not determine the effectiveness of an intervention  Unable to generalize results to other patients or practices 

Case series 
Are especially prone to: Selection, observation, and publication bias (can be subjected to meta analysis) 

SSTSDs 
Single subject time series design (aka N=1 design)............AB, ABA (withdrawal design), ABAB variants, ABAC (comparing 2 treatments 

Suitable candidates for SSTSDs 
Condition is chronic Condition is stable Spontaneous remission is not likely Previous treatment has had limited success No concurrent treatment is involved 

Trendline examples 
Stable downward and accelerating downward trend 

Multiple baseline design 
Variant: Simultaneous replication design........Patients begin studies at same time Treatment administered sequentially to patients only after clear treatment effect is observed for prior patient 

Descriptive statistics 
Usually retrospective, can be prospective, low validity but high clinical relevance 

4 types of descriptive statistics 
1) Unique case 2) Unexpected association 3) Unexpected development 4) Unusual presentation 

Frequency distribution 
Usually in a histogram 

Measure of central tendency in descriptive statistics 
Mean of a sample or mean of a population 

Levels of measurement 
Nominal, ordinal, interval, ratio 

Nominal 
Counting, central tendency=MODE, example=NUMBERS 

Ordinal 
Greater or less than operations, central tendency=MEDIAN, example=MILITARY RANK 

Interval 
Addition or subtraction, central tendency=(symmetrical) MEAN / (skewed) MEDIAN, example=Fahrenheit 

Ratio 
Addition, subtraction, multiplication, and division, central tendency=(symmetrical) MEAN / (skewed) MEDIAN, example=Kelvin, R.O.M. 

Normal distribution 
Symmetrical, unimodal histogram where the Mean=Median=Mode (they are in the same position, in the center of the Bell curve) 

Modal Division 
Unimodal = 1 peak, Bimodal = 2 peaks, Multimodal = < 2 peaks 

Standard error of the mean 
SE(n) = S/(square root of n) 

1 standard deviation from the mean 
68.3% confidence interval 

2 standard deviations from the mean 
95.5% confidence interval 

3 standard deviations from the mean 
99.7% confidence interval 

The wider the bars on the histogram are distributed... 
the higher the standard deviation (larger spread) 

The narrower the bars on the histogram are distributed... 
the lower the standard deviation (smaller spread) 

Skewed distributions 
Positive skews have tails extending to the right, where negative skews have tails to the left Mean is drawn towards the tail Mode is at the peak Median is between 

Know how to calculate zscore 
Zscore is the percentage to the left of the point in question, to find the other side, subject that percentage from 100% 

True H(o) > Reject H(o) 
Type 1 error, the odds of saying the hypothesis is true when it is actually false 

False H(o) > Reject H(o) 
CORRECT DECISION, the odds of saying the hypothesis is false when it actually is false 

True H(o) > Fail to reject H(o) 
CORRECT DECISION, the odds of saying the hypothesis is true when it cannot be proven to be false 

False H(o) > Fail to reject H(o) 
Type 2 error, the odds of saying the hypothesis is false, when it cannot be proven to be false 
