Last updated: June 28, 2023

Summarytoggle arrow icon

Classical epidemiology is the study of the distribution and determinants of disease in populations. Clinical epidemiology applies the principles of classical epidemiology to the prevention, detection, and treatment of disease in a clinical setting. The two main types of epidemiological studies are observational and experimental. Descriptive observational studies (e.g. case series, ecological studies) characterize factors related to individuals with a particular outcome. Analytical studies (e.g., cohort studies, case-control studies) seek to assess the association between exposures and outcomes. In experimental studies (e.g., randomized controlled trials), an intervention is performed to study its impact on a particular outcome. Measures such as proportions, rates, and ratios can be calculated for the data collected. An association between two variables (i.e., exposure and outcome) does not necessarily imply a causal relationship. Other reasons for observed associations in epidemiological studies may include errors (e.g., random error, systematic errors), confounding, and reverse causality.

The following concepts are discussed separately: measures of disease frequency (e.g., incidence, prevalence, mortality rates), measures of association (e.g., relative risk, absolute risk reduction), measures used in the evaluation of diagnostic research studies (e.g., sensitivity, specificity), precision and validity, critical appraisal, the practice of evidence-based medicine, and foundational statistical concepts (e.g., measures of central tendency, measures of dispersion, normal distribution, confidence intervals).

See also “Interpreting medical evidence,” “Statistical analysis of data,” and “Population health.”

Fundamental concepts in epidemiologytoggle arrow icon


  • Classical epidemiology: the study of the determinants and distribution of disease in populations [1][2]
  • Clinical epidemiology: the application of the principles of classical epidemiology to patients in a clinical setting

Goals of research [3]

  • Descriptive research: to summarize characteristics of a group
  • Predictive research: to forecast outcomes (e.g., for prevention, diagnosis, or management)
  • Explanatory research or causal inference: to establish causal mechanisms

Elements of epidemiological studies

See also “Types of epidemiological studies” and “Statistical analysis of data.”

  • Population (epidemiology): the total number of people in the group being studied [4]
  • Sample (epidemiology): a group of people selected from a larger population; meant to be representative of the larger population
  • Data (epidemiology): information collected during observation and/or experimentation that is used as a basis for analysis and discussion [5]
  • Exposure: a factor that is potentially associated with a particular outcome
  • Intervention: a treatment, drug, or management step that is being studied in an experimental study
  • Outcome: an endpoint (e.g., a disease or health-related event) that may occur after exposure to a risk factor or intervention.
  • Latency period
    • A period of apparent inactivity between stimulation (e.g., infection with a pathogen, exposure to a toxin, onset of a disease process, administration of a drug) and response (e.g., signs and symptoms, pharmacological effects). [6]
    • In clinical studies, the latency period must be considered to ensure adequately long follow-up duration. [7]

Research questions can be formulated using the PICO criteria: Population, Intervention, Comparison (or Control group), and Outcome

Study participants

  • Study group: a group of participants with a common feature that distinguishes them from other participants in the study.
    • In a clinical trial, participants are divided into a group that receives a particular intervention and a group that does not receive the intervention.
    • In a cohort study, participants are divided into a group that has had a particular exposure and a group that has not had the exposure.
    • In a case-control study, participants are divided into a group that has a particular outcome and a group that does not have the outcome.
  • Control group (definition varies by study type)
    • In a clinical trial, the control group is the people in the sample that do not receive the intervention (e.g., a drug), while the treatment group is the people in the sample that do receive the intervention.
    • In a case-control study, the control group is the people who do not have the outcome being studied (e.g., a disease); the control group is compared to the cases, i.e. the people who have the outcome.

Basic data parameters [2]

These basic measures are often used to describe or compare findings from epidemiological studies. See also “Measures of disease frequency”, “Measures of association”, and “Evaluation of diagnostic research studies.”

  • Proportion
    • Comparison of one part of the population to the whole
    • Proportions are usually expressed as percentages.
  • Rate (epidemiology)
    • A measure of the frequency of an event in a population over a specific period of time
    • Rates are usually reported as numbers of cases per 1,000 or 100,000 in a given time unit.
    • Types
      • Crude rate: applies to the entire population (specific characteristics are not taken into account)
      • Specific rate: applied to a population group with specific characteristics (e.g., sex-specific, age-specific)
      • Standardized rate (adjusted rate): crude rates that have been adjusted for potential confounding characteristics to allow for comparison between different populations (e.g., age-standardized rates are commonly used for death rates)
  • Ratios
    • Comparison of two values or the magnitude of two quantities
    • Ratios are usually expressed as X:Y or X per Y.
    • The numerator and denominator do not necessarily need to be related (e.g., a ratio comparing the number of hospitals in a city and the size of the population living in that city).

Elements describing population health

See “Population health” for the following:

Epidemiological studiestoggle arrow icon

Principles of study design

  • Study designs may be:
    • Observational: Participants are not assigned to any intervention.
    • OR experimental: Some participants are assigned to receive an intervention.
  • The choice of study design should be tailored to the research question.
  • Certain study designs typically produce results with a higher level of evidence than others.

Choose study designs and methods that maximize the strength of study results while minimizing random errors, bias, and confounding, within the limits of available resources and ethical standards.

Overview of types of epidemiological studies [8]

Observational studies Experimental studies
Descriptive studies Analytical studies
  • An intervention is applied
  • Usually involve 3 elements:
    • Study participants
    • Treatment (i.e., the procedure applied to the study participants)
    • Response (i.e., the effect of the intervention applied to the study participants)
  • The independent variable is manipulated to determine its effect on the dependent variable.
  • To identify individual characteristics (age, sex, occupation), location (e.g., residence, hospital), and/or time of events (e.g., during diagnosis, reporting) in relation to an outcome (e.g., disease)
  • To determine the relationship between an exposure and an outcome
  • To determine the effect of an intervention on outcomes, e.g., diseases
  • Create hypothesis
  • No comparison group
  • Test hypothesis
  • Always involve a comparison group.
  • Test hypothesis
  • Always involve a comparison group.
  • Informed consent is usually required.

Observational studies can be either descriptive or analytical, while experimental studies are always analytical in nature (i.e., they are used to test a hypothesis).

Observational studiestoggle arrow icon

Descriptive studies

  • No intervention involved
  • Patients are observed and the clinical course of the disease is studied.
  • The observations are used to form a hypothesis.
Overview of descriptive studies
Case report Case series report [9] Ecological study (correlation study)[10] Cross-sectional study (prevalence study) [11]
  • A report of a disease presentation, treatment, and outcome in a single subject or event
  • A report of a disease course or response to treatment that is compiled by aggregating several similar patient cases
  • A study that assesses links between an exposure and an outcome
  • Typically used if the outcome being studied is rare
  • A study that determines the prevalence of exposure and disease at a specific point in time
  • Can be either descriptive or analytical
Study method
  • An unusual or unique finding in a single subject is described in detail.
  • Researchers assess aggregated data of similar patient cases.
  • Typically, all of the patients have received the same intervention.
  • There is no control group.
  • Researchers assess aggregated data, where at least one variable (e.g., an outcome) is at a group level and not at an individual level.
  • The unit of observation is a large population (e.g., an entire country)
  • Ecological fallacy: making inferences about an individual in a group based on the characteristics of that group
  • Cannot control for confounding variables
  • Cannot assess causality
  • Examining a single case of cervical cancer in a 25-year-old female individual
  • Collecting and examining several cases of pericarditis at a local hospital
  • Determining the incidence of cholera deaths in different parts of a city to identify the source of exposure
  • Assessing the association between gross domestic product and cancer incidence across multiple countries

Analytical studies

Cross-sectional study

  • Can be analytical, e.g., when used to assess the association between an exposure and an outcome
  • See “Cross-sectional study.”

Case-control study [2]

  • Aim: : to study if an exposure (i.e., a risk factor) is associated with an outcome (i.e., disease)
  • Study method
    • Researchers begin by selecting patients with the disease (cases) and individuals without the disease (controls).
    • Controls are selected from the same source population and ideally have similar characteristics (e.g., gender, age) to the cases to reduce potential confounding.
    • The odds ratio is then determined between these groups.
  • Advantages
    • Helps determine whether individuals with a disease are more likely to have been exposed to a risk factor than patients without that disease.
    • Cost-efficient (exposure and outcome are measured only once)
    • Can be used to study rare diseases
    • Can be used to study diseases with long latency periods
  • Disadvantages
    • Recall and/or survivorship bias occurs in retrospective studies.
    • Cannot be used to determine prevalence or incidence
  • Example: A group of patients with histologically confirmed cervical cancer (cases) is compared to otherwise similar patients without histologically confirmed cervical cancer (controls) for the presence of human papillomavirus (exposure).

A case-control study generally examines a small population group over a short period of time (less cost-intensive) and evaluates the association between multiple exposures and one outcome. A cohort study generally examines a large population over a long period of time (more cost-intensive) and determines how one exposure is associated with multiple outcomes.

Cohort study [2]

  • Aim: to study the incidence rate and whether a given exposure is associated with the outcome of interest
  • Study method
    • The researchers gather a group of study participants who have common characteristics.
    • Participants are then classified into two groups: exposed and unexposed.
    • The incidence of the outcome of interest is compared between the two groups.
Types of cohort studies
Prospective cohort study Retrospective cohort study
Description Study begins before the groups develop an outcome of interest Study begins after the exposure and outcome of interest have already occurred
Exposure Study participants are categorized into an exposed group and an unexposed group. Study participants are categorized into a group that was previously exposed to a given risk factor (exposed; e.g., smoking) and a group that was not (unexposed).
Exposure of interest has to be present prior to outcome of interest.

The participants are followed prospectively for a period of time to see whether there is a difference in the rate at which the exposed and unexposed groups develop the outcome of interest.

Data previously collected about the participants is compared to see whether there was a difference in the rate at which the exposed and unexposed groups developed the outcome of interest (e.g., lung cancer) over a period of time.

Individuals with a smoking history of ≥ 1 pack of cigarettes a day (exposed group) are compared to individuals who are nonsmokers to see if there is a difference in the proportion of patients in each group that develop lung cancer (e.g., the outcome) within a specific follow-up period.

Individuals with a smoking history of ≥ 1 pack of cigarettes a day (exposed group) 5 years ago are compared to individuals who were nonsmokers 5 years ago to see if there is a difference in the proportion of patients in each group that eventually developed lung cancer (outcome) within a specific follow-up period.
Measurements must be taken at a minimum of two points in time.

In cohort studies, the study sample is selected based on exposure to a risk factor. In case-control studies, the study sample is selected according to having a disease or not, and then it is determined which participants were exposed to a risk factor.

Community survey [12][13]

  • Description: a study design in which self-reported data is collected from a large cohort
  • Method
    • Researchers identify a study question and compose a study form to gather relevant data
    • A sample of study participants is selected from the general population (can be random or nonrandom).
    • Data are gathered from study participants either via interviewing or filling in a pre-designed form.
    • The data are then analyzed with the appropriate statistical methods.
  • Advantages
    • Requires fewer resources than other study designs
    • Requires a relatively small amount of time to complete
    • Can cover a large number of study participants (easy to get a representative sample)
  • Disadvantages
    • Response rate is hard to control.
    • All data are subjectively reported by study participants and therefore may be imprecise.

Twin concordance study

Adoption study

  • Aim: to determine the influence of genetic and environmental risk factors on the development of a disease
  • Study method
    • The frequency of disease is compared between adopted children whose biological parents have the disease and adopted children whose biological parents do not have the disease [14]
    • Among adoptees with a disease, the frequency of the disease is compared among birth and adoptive parents.
  • Example: Two groups of adults that were adopted are studied. Individuals in the first group have a biological parent with schizophrenia (exposure), and individuals in the second group have biological parents without schizophrenia. The prevalence of schizophrenia (outcome) is compared between the two groups; if the prevalence is higher in children of parents with schizophrenia, genetic factors are likely involved.

Registry study

  • Aim: a retrospective study to analyze data obtained from patient registries
  • Study method
    • A registry contains data on patients who have something in common (e.g., a lung cancer registry contains demographic and clinical information about patients diagnosed with lung cancer).
    • A good quality registry has complete data on the population, including data on potential confounding factors.

Experimental studiestoggle arrow icon

An experimental study is one in which the researcher manipulates one or more independent variables to see how the changes affect one or more outcomes (dependent variables). Experimental studies can be divided into those that occur in clinical settings (e.g., randomized controlled trials) and those that occur in community settings (e.g., field trials and community trials).

Randomized controlled trials (RCTs) [15]

  • Aim: to determine the possible effect of a specific intervention on a given population
  • Advantages
    • Minimizes bias
    • Can demonstrate causality
  • Disadvantages
    • Cannot be used to evaluate rare diseases
    • Cannot be used when treatments have well-known adverse side effects
    • Expensive and time-consuming
    • Results may not be applicable to “real world” settings, and may not be generalizable to other populations.

Randomized controlled trials are considered the gold standard for testing interventions.

RCT design elements

  • Enrollment: Study participants should be selected according to predefined inclusion and exclusion criteria.
  • Intervention group: a group of study participants that receive the studied intervention.
  • Control group: a group of study participants that are compared to the intervention group and do not receive the studied intervention. [16]
    • Placebo-controlled trial: Control group participants receive a placebo.
    • Study designs in which control subjects receive a treatment used in standard practice:
      • Superiority trials: aim to prove that the studied intervention is better than an existing intervention
      • Equivalence trials: aim to prove that the studied intervention is as effective or has nearly the same effectiveness as an existing intervention
      • Noninferiority trials: aim to prove that the studied intervention is not inferior to an existing intervention [17][18]
  • Randomization
  • Blinding: the practice of not informing an individual or group about which study participants are part of the control group and which are part of the treatment group (used to reduce bias)
    • Single-blind study: Study participants do not know whether they are part of the control or treatment group.
    • Double-blind study: Neither the researchers nor the study participants know which study participants are part of the control group and which are part of the treatment group. Double blinding is the gold standard when studying treatment outcomes. [19]
    • Triple-blind study: Neither the researchers, the study participants, nor the data analysts know which study participants are part of the control group and which are part of the treatment group.
  • Unblinding
    • In blinded clinical trials, unblinding is the intentional revelation to investigators and/or research subjects of participant allocation into either the intervention or control group.
    • May take place after completion of the trial or because of specific circumstances (e.g., pregnancy, adverse events that affect safety)

RCT variants based on the study design [20][21]

Parallel group design

  • Definition: an experimental study design in which participants remain within their assigned groups for the entire duration of the study
  • Parallel group design is the most commonly used RCT variant.
  • Advantages
    • Relatively simple and easy to implement
    • Suitable for studying irreversible or long-lasting effects

Crossover design

  • Definition: an experimental study design in which each participant switches from the intervention group to the control group or vice versa with a washout period in between
  • The order in which the participant is assigned to the intervention group or control group is typically randomized.
  • Advantages
    • Low risk of confounding bias because participants serve as their own controls
    • High statistical power because the sample size is doubled with the same number of participants
    • High recruitment rates because each participant knows they will receive some treatment during the trial
  • Disadvantages
    • Increased dropout rates because of the long duration of the study
    • Cannot be performed if the intervention has a long-lasting effect because the effects of one phase can persist into the next (carryover effect)

Factorial design

  • Definition: an experimental study design in which multiple interventions are studied simultaneously by assigning participants to various combinations of interventions and placebo
  • Advantages
    • Can investigate a high number of interventions simultaneously
    • Can identify potential interactions between interventions (e.g., synergistic effects of drug combinations)
  • Disadvantages
    • Interim data analysis and safety monitoring can be complicated.
    • Participant recruitment and protocol adherence are more demanding.

Cluster design

  • Definition: an experimental study design in which the unit of randomization is a group rather than an individual participant
  • Advantages
    • Lower implementation cost and easier enrollment
    • Decreases risk of contamination of intervention
  • Disadvantages
    • Results measure groups, which may have a similar response, rather than reflecting individual variations.
    • Low statistical power because the units of analysis are groups, not individual participants.
    • Increases risk of recruitment bias.

RCT variants based on the type of intervention

Clinical drug trials [22]

Surgical and medical device trials [23][24]

  • Aim: to assess the safety and efficacy of surgical intervention (surgical trial) or a device used for treatment or diagnostics in human subjects (medical device trial)
  • Study methods
    • Surgical trials can compare a surgical with a nonsurgical intervention or compare surgical techniques.
    • Medical device trials are required before FDA approval for devices that are considered high-risk (e.g., pacemakers, stents, infusion pumps).
  • Challenges
    • Implementing a placebo (e.g., sham surgery, sham device) and double-blinding may not be feasible or ethical in many circumstances.
    • Standardization of a surgical intervention is difficult to achieve because of individual variations in surgical technique.
    • Participant recruitment is more difficult than with clinical drug trials.

Interim analysis and safety monitoring [25][26][27]

  • Every experimental study involving human subjects requires a monitoring plan to ensure participant safety and scientific merit.
  • Data and safety monitoring board
    • An independent panel of experts who periodically review the conduct and data of ongoing trials to ensure participant safety and scientific merit.
    • A data and safety monitoring board is required for nearly all phase 3 trials and for phase 1 and phase 2 trials that are multicenter, blinded, or involve a high-risk intervention.
  • Interim analysis
    • Data analysis that occurs in an experimental study before all data has been collected
    • The frequency and methodology of interim analysis should be specified in the study protocol before the start of the study.
  • Serious adverse event (clinical trial): an adverse event during a clinical trial that is life-threatening, causes or prolongs hospitalization, or results in congenital anomalies, significant morbidity, or death
  • Unexpected adverse event (clinical trial)
    • An adverse event during a clinical trial whose nature, severity, or frequency is unknown at the start of the trial (e.g., not described in protocol-related documentation of the intervention or product labeling by the manufacturer) or unexpected given the participant's underlying diseases or risk factors
    • Can also occur after market approval (i.e., during phase 4 trials)
    • An unexpected adverse event, especially if it is serious, should result in temporary discontinuation of the clinical trial until the reasons for the event are determined.
    • If the unexpected adverse event is determined to be related to the trial, it may be necessary to modify the research protocol to minimize the risk to participants (e.g., by altering the exclusion or inclusion criteria, changing the dosing, increasing the monitoring frequency).
    • If it is not feasible to minimize risk to an acceptable level, the trial should be terminated.
  • Discontinuing experimental studies
    • Ongoing trials are discontinued if any of the following become evident on interim analysis:
      • Evidence of operational futility (e.g., poor enrollment rate, higher than expected dropout rate, clearly inferior treatment efficacy)
      • Safety concerns
      • Early evidence of clear benefit
    • Stopping rules
      • A set of statistical criteria established before a clinical trial that specifies when the intervention should be stopped because of safety concerns from expected adverse events, operational futility, or early evidence of clear benefit.
      • Should be specified in the study protocol

Analysis upon RCT completion

Intention-to-treat and per-protocol analyses

Methods of analysis for randomized controlled trials
Intention-to-treat analysis [28] Per-protocol analysis [29][30]
  • Study participants are analyzed according to the group to which they were originally randomized, regardless of whether they actually received the intervention or dropped out of the study.
  • Evaluates the question of what happens when a particular treatment is prescribed
  • Only participants who adhered to the study protocol are analyzed.
  • Evaluates the question of what happens when a particular treatment is used
  • As-treated analysis: Participants are evaluated based on the ultimate treatment they received, regardless of the initial group assignment. [31]
  • More reflective of the “real world,” where nonadherence and other protocol deviations occur
  • Reduces the probability of saying the intervention has an effect when it actually does not (i.e., type I error)
  • Helps to reduce selection bias
  • Preserves sample size and thus preserves statistical power
  • Improves the estimate of the effect of treatment under optimal conditions
  • Nonadherence can lead to a conservative estimate of the treatment effect.
  • Analyzing nonadherent, dropout, and adherent study participants together may introduce heterogeneity.
  • Susceptible to type II error: Including patients who did not adhere to treatment can weaken the effect size and make the novel treatment appear less effective than it is.

Intention-to-treat analyzes as-randomized (once randomized, always analyzed).

Confidence in the study results increases when intention-to-treat analysis and per-protocol analysis produce the same results. [28]

Subgroup analysis [20][32]

  • Aim: to investigate the heterogeneity of results in a study or recognize significant discrepancies or similarities in the treatment outcome among different subgroups of patients
  • Study methods
  • Advantages: may show stable treatment effects over similar subgroups of patients
  • Disadvantages

Other types of experimental studies

Field trials

  • Aim: to determine the effect of disease prevention interventions in individuals who do not already have a disease
  • Example: observation of children who did and did not receive the Salk vaccine for prevention of poliomyelitis to see whether they developed paralysis or death

Community trials

  • Aim: similar to field trials, but follow communities rather than individuals
  • Example: studying the incidence of myocardial infarction and stroke in communities who implement lifestyle changes to prevent cardiovascular disease compared to communities who do not implement such changes

Other types of studiestoggle arrow icon

Systematic reviews and meta-analyses are considered secondary research study designs as they analyze data from multiple primary research studies. Survival analyses can be performed on data from prospective cohort studies or RCTs.

Systematic review [2][33]

  • Aim: to answer a defined research question
  • Study method
    • Researchers collect and summarize evidence from the existing literature that fits established criteria.
    • Quality assessment of the study is achieved by methods such as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (see “Tips and Links”).
    • A systematic review may include a statistical analysis of the data (i.e. metaanalysis).
  • Advantages: can improve evidence-based clinical decision-making
  • Disadvantages
    • Cannot correct issues with the quality of the individual studies included in the review
    • Susceptible to publication bias

Metaanalysis [34]

  • Aim: : to increase statistical power and achieve more precise results
  • Study method
    • Data from multiple studies are systematically assessed, combined, and processed with statistical methods.
    • Risk-of-bias assessment (See also “Systematic errors”).
    • The results are often reported graphically using forest plots.
  • Advantages: can identify similarities and/or differences between individual studies
  • Disadvantages
    • Unable to eliminate limiting factors particular to the study types included (variability among studies is referred to as statistical heterogeneity)
    • Only as good as the individual studies used
    • Susceptible to several forms of bias (e.g., selection bias, publication bias)

Forest plot

  • Definition: a graph that combines the estimated results from multiple studies investigating the same question and visualizes each study's effect size, confidence interval, and calculated overall effect
  • Elements
    • X-axis: primary outcome measure (e.g., relative risk)
    • Central vertical line on x-axis: null value (e.g., RR=1, ARR=0)
    • First column: a list of the individual studies
    • Squares: the results (i.e., point estimate) of each study
      • Vertical lines: the exact point estimate
      • Horizontal lines: the confidence interval of the result
    • Area of the squares: proportional to the weight that the individual study is given in the overall metaanalysis (e.g., larger square due to larger sample size and/or statistical power)
    • Diamond at the bottom of the plot: summary combining the results of all studies
      • Midpoint: the point estimate of the combined result
      • Width: the confidence interval of the combined result

Survival analysis (prognosis study) [2][35]

  • Aim
    • To determine the average time to a given outcome identified on follow-up
    • Often used to measure disease prognosis
  • Analysis
    • Always prospective in nature (i.e. using data from cohort studies or RCTs)
      • Time-to-event analysis: Individual follow‑ups are performed from the onset of a disease to a chosen endpoint (e.g., death, development of a particular complication), or after exposure to a risk factor until the onset of a disease.
      • Five-year survival rate: the percentage of patients with a particular disease who have survived for 5 years after the initial diagnosis
    • In survival analysis, a hazard ratio is often calculated to compare outcomes among two groups.
  • Disadvantages
    • Censored cases: Not all participants will have the study endpoint during the period of observation; no prediction can be made for these participants.
    • Censoring can be caused by the following:
      • The study period ending before participants experience the outcome of interest
      • Participant drop out
      • Competing events (e.g., death from a different cause)
  • Kaplan-Meier analysis [35][36]
    • Used to analyze incomplete time-to-event data and to estimate the survival of subjects over a set period of time following a certain treatment
    • Ideal for estimating the survival of a cohort composed of a small number of cases
    • Events of interest include death, treatment effectiveness, and recovery.
    • Allows for an estimation of survival over time even if participants are studied over different time intervals (e.g., some participants drop out or are lost to follow-up)
    • Data on survival analysis can be graphically displayed as a Kaplan-Meier curve.
  • Kaplan-Meier curve: graphical representation of Kaplan-Meier analysis in a step-shaped diagram
    • Allows for an estimation of the probability that the subject will survive up to a point in time
    • The horizontal axis represents time.
    • The vertical axis represents the estimated probability of survival.
    • Time intervals are defined by specific events: a time interval ends when an outcome of interest occurs.
    • Probability of survival for each time interval = number of patients for whom the event has not occurred/number of patients who are at risk

Qualitative studies [37][38][39]

  • Aim
    • To understand the subjective perspective of individuals on their experience of reality and the implications of these perspectives and experiences through the analysis of qualitative data
    • Hypothesis generation
  • Analysis
    • Structural or thematic coding of the gathered data is used to group data and draw insights.
    • No statistical analysis is performed.
  • Data collection methods
    • Observation
    • Interviews
    • Focus groups
    • Surveys with open-ended questions
    • Archive work and review of recorded sources (e.g., text, audio, video)
  • Qualitative research methodologies [40]
    • Grounded theory: the induction of hypotheses from systematically collected and analyzed data
    • Ethnography: the study of culture and social interactions of a particular group of people
    • Narrative inquiry: the study of narratives as the basic unit of describing and recording human experience
    • Phenomenological hermeneutics: a study of the lived experiences and perceptions of individuals within specifc contexts (e.g., cultural context, historical context)
    • Action research: a research method that involves collaboration between researchers and individuals in a particular setting to solve problems
    • Case study: the systematic investigation of a complex phenomenon within a specific context to gather insight into the phenomenon of interest

Random errors, bias, and confoundingtoggle arrow icon

An observed association in a study does not necessarily imply causation. Other underlying causes such as random error, bias, confounding, and/or effect modification should first be excluded.

High degrees of random errors, bias (e.g., selection bias), and confounding limit the validity (i.e., accuracy) of a study.

Random error [8]

Systematic error (bias)

  • Definition: An error in the study design or the way in which the study is conducted that causes systematic deviation of findings from the true value
  • Risk of bias

Selection bias

  • Description: The individuals in the sample group are not representative of the population from which the sample is drawn because the sampling or the treatment allocation is not random.
  • Types of selection bias include:
    • Sampling bias (ascertainment bias)
      • Occurs when certain individuals are more likely to be selected for a study group, resulting in a nonrandom sample
      • This can lead to incorrect conclusions being drawn about the relationship between exposures and outcomes.
      • Limits generalizability
      • Types of sampling bias
        • Nonresponse bias: Participants who do not return information during a study (i.e., nonresponders) have systematically different characteristics from those who do (e.g., some study participants do not return a call because they are feeling unwell).
        • Healthy worker effect: The working population is healthier on average than the general population.
        • Volunteer bias: Individuals who volunteer to participate in a study have different characteristics than the general population.
    • Attrition bias
      • A type of nonresponse bias
      • Selective loss of participants to follow up
      • Most commonly seen in prospective studies
      • Risk that the remaining participants differ systematically from those lost to follow up
    • Berkson bias: Individuals in sample groups drawn from a hospital population are more likely to be ill than individuals in the general population.
    • Susceptibility bias: One disease predisposes affected individuals to another disease, and the treatment for the first disease is mistakenly interpreted as a predisposing factor for the second disease.
    • Survival bias
      • Also known as prevalence-incidence bias and Neyman bias
      • When observed subjects have more or less severe manifestations than the standard exposed individual
        • If individuals with severe disease die before the moment of observation, those with less severe disease are more likely to be observed.
        • If individuals with less severe disease have a resolution of their disease before the moment of observation, those with more severe disease are more likely to be observed.
        • Most commonly occurs in case-control and cross-sectional studies.
  • Solutions
    • In clinical trials, randomize to control for known and unknown confounders.
    • In case-control studies, ensure the sample is representative of the population of interest.
    • Ensure the correct reference group is chosen for comparison.
    • Collect as much data on the characteristics of the participants as possible.
    • Nonresponder characteristics should not be assumed. Instead, undisclosed characteristics of nonresponders should be recorded as unknown.

Information bias

  • Description: incorrect data collection, measurement, or interpretation that leads to misclassification of groups or exposure
    • Information is gathered differently between the treatment and control groups.
    • Insufficient information about exposure and disease frequency among subjects
  • Types of information bias
    • Measurement bias: any systematic error that occurs when measuring the exposure or outcome
    • Reporting bias: a distortion of the information from research due to the selective disclosure or suppression of information by the individuals involved in the study
      • Can involve the study, design, analysis, and/or findings
      • Results in underreporting or overreporting of exposure or outcome
    • Interviewer bias: The interview approach distorts the responses provided by study participants, which results in researchers finding differences between groups when there are none.
    • Surveillance bias: An outcome is diagnosed more frequently in a sample group than in the general population because of increased testing and monitoring.
    • Performance bias: differences between study groups that are related to group assignment
      • Hawthorne effect
        • Subjects change their behavior once they are aware that they are being observed.
        • Especially relevant in psychiatric research
        • This type of bias is difficult to eliminate.
      • Procedure bias: This bias occurs when patients or investigators decide on the assignment of treatment and this affects the findings. The investigator may consciously or subconsciously assign particular treatments to specific types of patients (e.g., one group receives a higher quality of treatment).
      • May be reduced by blinding , or use of a cluster randomization design
    • Recall bias: awareness of a condition by subjects changes their recall of related risk factors (recall a certain exposure)
      • Common in retrospective studies
      • Example: After claims that the MMR vaccine caused autism became public, parents of children diagnosed with autism were more likely to recall the start of autism being soon after their child was vaccinated, as compared with parents of children who were diagnosed with autism prior to these claims becoming public.
      • Can be reduced by decreasing time to follow-up in retrospective studies (e.g., retrospective cohort studies or case-control studies), or using data on risk factors that was collected prior to the occurrence of the outcome (if available)
    • Misclassification bias: an error in which research subjects are classified into the wrong exposure or outcome groups, thereby distorting the observed association
      • Nondifferential misclassification bias: a misclassification bias in which the frequency of classification errors is similar in the groups under comparison
        • Occurs when misclassification is unrelated to other variables in the study
        • If the misclassified variable is dichotomous, the odds ratio is skewed toward the null value.
        • If the misclassified variable is not dichotomous (i.e., more than two exposure or outcome categories), the odds ratio can be skewed either towards or away from the null value.
          • Example: In a study on risk factors for lifetime cancer diagnosis, participants are asked if they have ever been exposed to radiation but are not asked for further clarification (e.g., what type, how recently, for how long). This leads to an incorrect conclusion about the association between radiation exposure and lifetime cancer diagnosis. Because all participants were asked the same vague question, the frequency of classification errors is similar across exposed and unexposed groups.
      • Differential misclassification bias: a misclassification bias in which the frequency of classification errors differs between the groups under comparison
        • Occurs when misclassification is related to other variables in the study
        • Skews the odds ratio either toward or away from the null value
        • Example: Subjects who develop a rash after being exposed to a chemical are more likely to remember the exposure than those who do not develop a rash (recall bias). This results in unequal misclassification of study participants into exposed and unexposed groups.

Allocation bias

  • Definition: a systematic difference in the way that participants are assigned to treatment and control groups
  • Example: assigning patients with better baseline blood pressure to the treatment group for a new antihypertensive treatment
  • Solution: randomization

Cognitive bias

  • Description: : The personal beliefs of the study participants and/or investigators influence the results of the study.
  • Types of cognitive bias
    • Response bias: Study participants do not respond truthfully or accurately because of the manner in which questions are phrased (e.g., leading questions) and/or because subjects interpret certain answer options to be more socially acceptable than others.
    • Observer bias (experimenter-expectancy effect or Pygmalion effect): The measurement of a variable or classification of subjects is influenced by the researcher's knowledge or expectations.
    • Confirmation bias: The researcher includes only those results that support their hypothesis and ignores other results.
    • Placebo and nocebo effects: A placebo or nocebo affects study participants' preconceptions/beliefs about the outcome.
  • Solutions
    • Use of a placebo
    • Blinding
    • Prolong the time of observation to monitor long-term effects.

Publication bias

  • Definition: type of bias that occurs when the findings of a study influence the decision to publish it
  • Solutions
    • Registration of clinical trials before any participants are enrolled
    • Publishing study protocol before commencing the study

Biases specific to screening tests

See “Evaluation of diagnostic research studies” for details.


  • Definition
    • A confounder is any third variable that is associated with the exposure and the outcome but is not on the causal pathway between exposure and outcome.
    • A confounder can be responsible for the observed relationship between the dependent and independent variables.
  • Example: : Exposure to coal can cause lung cancer in mine workers. Many miners also smoke cigarettes, which can lead to lung cancer as well.

Minimizing confounding during study design

Methods to reduce potential confounding during the study design phase include:

  • Randomization
    • Random allocation of study participants to treatment and control groups (e.g., in a randomized controlled trial)
    • Helps to distribute potential known and unknown confounders among study groups
  • Crossover study design: Each study participant serves as his or her own control, thereby reducing the influence of confounding.
  • Restriction (epidemiology):
    • Definition: a study design in which only individuals who meet certain criteria are included in the study sample (e.g., only male individuals with a particular disease are included in a study to avoid the influence of biological sex on the exposure and outcome)
    • Disadvantages
  • Matching (epidemiology)
    • Definition: selection of study participants so that the distribution of variables is similar between study groups
    • Can be done in two ways:
      • Study participants are matched individually to participants with similar attributes (pairwise or individual matching).
      • Study participants are matched in groups such that the groups have similar frequency of variables (frequency matching).
    • Commonly used in case-control studies to minimize confounding
    • The matching variable should meet the criteria for a confounder.
    • Disadvantages
      • Does not completely eliminate confounding
      • The matching factor cannot be studied as a predictor of the outcome.
      • Can introduce bias if the variables that are matched are not actually confounders
    • Example: In a study on the association between hypertension and end-stage renal disease, obesity is a potential confounder because it is associated with both diseases. By matching each participant with hypertension to a participant with a similar BMI who does not have hypertension, the potential for confounding by obesity is reduced.

Minimizing confounding during data analysis

Methods to reduce potential confounding during the data analysis phase include:

Causal relationshipstoggle arrow icon

Causal inference

Once other reasons for an association between two variables have been excluded, it can be evaluated whether the association is likely to be causal. The Bradford-Hill criteria are a list of criteria that, if met, help to establish causality in epidemiological studies.

Association is not the same as causation. Remember to exclude random errors, bias, and confounding, and evaluate for effect modification before evaluating causal criteria.

Causal criteria (Bradford-Hill criteria)
Criteria Description Example
  • The outcome occurs after the exposure within an expected amount of time.
Strength of association (effect size)
  • A quantitative measure of the degree of relationship between two variables
  • The stronger the association between an exposure and its observed outcome, the more likely there is to be a causal relationship between them.
  • The risk of lung cancer is severalfold higher in cigarette smokers than nonsmokers.
Dose-response relationship (biological gradient)
  • Greater exposure is associated with a higher occurrence of the outcome.
Reproducibility (consistency)
  • Similar findings are observed in different studies (e.g., in different places, with different sample sizes).
  • An association between a specific exposure and specific disease occurs in a specific population at a specific time, or an exposure leads to only one outcome.
Biologic plausibility
  • The relationship between an exposure and an outcome is consistent with current biological and medical knowledge.
Coherence (epidemiology)
  • New evidence is consistent with previously established evidence.
Experimental evidence
  • Data drawn from experimental studies support the presumed causal relationship between exposure and outcome.
  • An empirical observation of high incidence of lung disease among coal mine workers is supported by experimental data linking chronic coal exposure and the development of anthracosis.
Analogy (epidemiology)
  • When there is strong evidence of a causal relationship between an exposure and an outcome, there is a greater likelihood of a causal relationship between another similar exposure and outcome
  • When one class of medication is known to produce an effect, it is likely that another agent of that class produces a similar effect.

To establish causation, the cause must precede the effect. A temporal relationship may be difficult to establish using case-control studies or cross-sectional studies because the possible effect (i.e., outcome) and cause (i.e., exposure) are measured at the same time.

Reverse causality

  • Definition: an association between exposure and outcome that is different than common presumption
  • Example: people assume that low socioeconomic status causes schizophrenia, but in fact, schizophrenia causes a decline in socioeconomic status over time

Complex multicausal relationships [42]

Occur when a third variable influences the effect of an exposure on an outcome, leading to a different effect on the control and treatment groups. These are not considered types of bias, but rather biological phenomena.

  • Effect modification: occurs when the level of the effect of an exposure on an outcome is different across different strata of a third variable (i.e., the exposure has a different impact in different circumstances)
    • Example: A certain drug works in children but does not have any effect on adults.
    • Stratified analysis can allow better identification and understanding of effect modification.
    • Effect modification can be inferred if stratifying participants into subgroups according to the third variable results in a stronger relationship in one subgroup.
  • Interaction: occurs when two or more interdependent exposures influence the outcome measured in a variety of ways

Effect modification is not a type of error. Causal relationships can exist even if the effect of the exposure on the outcome changes across strata for another variable (e.g., age groups).

Conducting research projectstoggle arrow icon

Preparation phase [15][43]


  1. Perform a literature review to establish what is already known on the topic.
  2. Select a mentor or mentorship team.
  3. Develop the research question using PICO criteria and FINER criteria.
  4. Choose an appropriate study design.
  5. Choose the study population.
    • Inclusion and exclusion criteria
    • Sites of recruitment
  6. Select relevant variables on which to collect data.
  7. Establish an analysis plan.
    • Develop a hypothesis.
    • Meet with a statistician to discuss study methods and planned analyses.
    • Perform a power/sample size calculation to identify the number of participants required to see a difference in effect between groups.
  8. Obtain approval to do research on human subjects through an institutional review board (IRB).
  9. Determine the process for obtaining informed consent from study participants.
  10. Explore potential funding opportunities, if necessary.

Meeting with a statistician before starting a research project can help in determining the best study design and analysis plan.

Research question

The following criteria can help formulate an appropriate research question:

  • PICO criteria [44]
    • Patients
    • Intervention (or exposure)
    • Comparison group
    • Outcome
  • FINER criteria [15]
    • Feasible
    • Interesting
    • Novel
    • Ethical
    • Relevant

Choosing a study design

Considerations in choosing a study design
Study feature Appropriate study design(s)
Rare outcome
Rare exposure
Cost/resource limitations
Time limitations
Assessment of causality

Multiple study designs may be appropriate for answering the same clinical question.

Implementation phase [15][43]

  • Collection of primary data
    • Subjective, e.g., surveys (paper or online), interviews
    • Objective, e.g., electronic health records, laboratory testing, imaging studies or other diagnostic testing
  • Analysis: Select appropriate analytic techniques in consultation with mentors and/or statisticians.
  • Reporting results
    • Report results according to guidelines for main study types.
    • See “Equator network reporting guidelines” in “Tips and Links.”
  • Dissemination of results
    • Submission of manuscripts to scientific journals.
    • Submission of abstracts to local, national, or international meetings.

Participant data should be entered and stored in a secure database.

Referencestoggle arrow icon

  1. Raymond S. Greenberg. Medical Epidemiology: Population Health and Effective Health Care, 5th Edition. McGraw-Hill ; 2015
  2. Smith V, Devane D, Begley CM, Clarke M. Methodology in conducting a systematic review of systematic reviews of healthcare interventions. BMC Med Res Methodol. 2011; 11 (1).doi: 10.1186/1471-2288-11-15 . | Open in Read by QxMD
  3. Haidich AB. Meta-analysis in medical research.. Hippokratia. 2010; 14 (Suppl 1): p.29-37.
  4. Altman DG, Bland JM. Statistics Notes: Time to event (survival) data. BMJ. 1998; 317 (7156): p.468-469.doi: 10.1136/bmj.317.7156.468 . | Open in Read by QxMD
  5. Rich JT, Neely JG, Paniello RC, Voelker CCJ, Nussenbaum B, Wang EW. A practical guide to understanding Kaplan-Meier curves. Otolaryngol Head Neck Surg. 2010; 143 (3): p.331-336.doi: 10.1016/j.otohns.2010.05.007 . | Open in Read by QxMD
  6. Appendices. . Accessed: August 9, 2010.
  7. Renjith V, Yesodharan R, Noronha JA, Ladd E, George A. Qualitative Methods in Health Care Research.. International journal of preventive medicine. 2021; 12: p.20.doi: 10.4103/ijpvm.IJPVM_321_19 . | Open in Read by QxMD
  8. Patricia M. Qualitative research in medical education. Oxford University Press ; 2013: p. 638-647
  9. Cristancho S, Goldszmidt M, Lingard L, Watling C. Qualitative research essentials for medical education. Singapore Med J. 2018; 59 (12): p.622-627.doi: 10.11622/smedj.2018093 . | Open in Read by QxMD
  10. R. Beaglehole. Basic Epidemiology. WHO ; 2006
  11. Hulley SB, Cummings SR, Browner WS, Grady D, Newman TB. Designing Clinical Research. Lippincott Williams & Wilkins ; 2013
  12. Hamann KL, Fancher TL, Saint S, Henderson MC. Clinical Research During Internal Medicine Residency: A Practical Guide. Am J Med. 2006; 119 (3): p.277-283.doi: 10.1016/j.amjmed.2005.12.001 . | Open in Read by QxMD
  13. Brian Haynes R. Forming research questions. J Clin Epidemiol. 2006; 59 (9): p.881-886.doi: 10.1016/j.jclinepi.2006.06.006 . | Open in Read by QxMD
  14. Jekel JF, Katz DL, Elmore JG. Epidemiology, Biostatistics, and Preventive Medicine. Saunders ; 2007: p. 3-4
  15. Hernán MA, Hsu J, Healy B. A Second Chance to Get Causal Inference Right: A Classification of Data Science Tasks. CHANCE. 2019; 32 (1): p.42-49.doi: 10.1080/09332480.2019.1579578 . | Open in Read by QxMD
  16. Pacific Northwest Ecosystem Research Consortium. Willamette River Basin Atlas: Trajectories of Environmental and Ecological Change (2nd edition). Pacific Northwest Ecosystem Research Consortium
  17. Measuring the Non-Observed Economy A Handbook. Updated: January 1, 2002. Accessed: July 9, 2020.
  18. Mann CJ. Observational research methods. Research design II: cohort, cross sectional, and case-control studies. Emerg Med J. 2003; 20 (1): p.54-60.doi: 10.1136/emj.20.1.54 . | Open in Read by QxMD
  19. Song JW, Chung KC. Observational studies: cohort and case-control studies.. Plast Reconstr Surg. 2010; 126 (6): p.2234-2242.doi: 10.1097/PRS.0b013e3181f44abc . | Open in Read by QxMD
  20. Gupta U, Verma M. Placebo in clinical trials. Perspectives in Clinical Research. 2013; 4 (1): p.49.doi: 10.4103/2229-3485.106383 . | Open in Read by QxMD
  21. Steven M Snapinn. Noninferiority trials. Curr Control Trials Cardiovasc Med. 2000; 1 (1): p.19.doi: 10.1186/cvm-1-1-019 . | Open in Read by QxMD
  22. Piaggio G, Elbourne DR, Pocock SJ, Evans SJW, Altman DG, CONSORT Group for the. Reporting of Noninferiority and Equivalence Randomized Trials. JAMA. 2012; 308 (24): p.2594.doi: 10.1001/jama.2012.87802 . | Open in Read by QxMD
  23. Kabisch M, Ruckes C, Seibert-Grafe M, Blettner M. Randomized Controlled Trials. Deutsches Aerzteblatt Online. 2011.doi: 10.3238/arztebl.2011.0663 . | Open in Read by QxMD
  24. Higgins JPT, Thomas J, Chandler J, et al.. Cochrane Handbook for Systematic Reviews of Interventions, 2nd edition. John Wiley & Sons ; 2019
  25. Evans SR. Clinical trial structures. Journal of experimental stroke & translational medicine. 2010; 3 (1): p.8-18.doi: 10.6030/1939-067x-3.1.8 . | Open in Read by QxMD
  26. Umscheid CA, Margolis DJ, Grossman CE. Key Concepts of Clinical Trials: A Narrative Review. Postgrad Med. 2011; 123 (5): p.194-204.doi: 10.3810/pgm.2011.09.2475 . | Open in Read by QxMD
  27. Farrokhyar F, Karanicolas PJ, Thoma A, et al. Randomized Controlled Trials of Surgical Interventions. Ann Surg. 2010; 251 (3): p.409-416.doi: 10.1097/sla.0b013e3181cf863d . | Open in Read by QxMD
  28. Owen Faris, Jeffrey Shuren. An FDA Viewpoint on Unique Considerations for Medical-Device Clinical Trials. N Engl J Med. 2017; 376 (14): p.1350-1357.doi: 10.1056/nejmra1512592 . | Open in Read by QxMD
  29. Policy for Data and Safety Monitoring. Updated: June 10, 1998. Accessed: April 20, 2023.
  30. Reviewing and Reporting Unanticipated Problems Involving Risks to Subjects or Others and Adverse Events: OHRP Guidance (2007). Updated: January 15, 2007. Accessed: April 20, 2023.
  31. Coates S, Täubel J, Lorch U. Practical risk management in early phase clinical trials. Eur J Clin Pharmacol. 2018; 75 (4): p.483-496.doi: 10.1007/s00228-018-02607-8 . | Open in Read by QxMD
  32. Gupta S. Intention-to-treat concept: A review. Perspect Clin Res. 2011; 2 (3): p.109-112.doi: 10.4103/2229-3485.83221 . | Open in Read by QxMD
  33. Ranganathan P, Pramesh C, Aggarwal R. Common pitfalls in statistical analysis: Intention-to-treat versus per-protocol analysis. Perspectives in Clinical Research. 2016; 7 (3): p.144.doi: 10.4103/2229-3485.184823 . | Open in Read by QxMD
  34. Shah PB. Intention-to-treat and per-protocol analysis. Can Med Assoc J. 2011; 183 (6): p.696-696.doi: 10.1503/cmaj.111-2033 . | Open in Read by QxMD
  35. Smith VA, Coffman CJ, Hudgens MG. Interpreting the Results of Intention-to-Treat, Per-Protocol, and As-Treated Analyses of Clinical Trials. JAMA. 2021; 326 (5): p.433.doi: 10.1001/jama.2021.2825 . | Open in Read by QxMD
  36. Schuhlen H. Pre-specified vs. post-hoc subgroup analyses: are we wiser before or after a trial has been performed?. European Heart Journal. 2014; 35 (31): p.2055-2057.doi: 10.1093/eurheartj/ehu202 . | Open in Read by QxMD
  37. Mathes T, Pieper D. Clarifying the distinction between case series and cohort studies in systematic reviews of comparative studies: potential impact on body of evidence and workload. BMC Med Res Methodol. 2017; 17: p.107.doi: 10.1186/s12874-017-0391-8 . | Open in Read by QxMD
  38. Hart J. On Ecological Studies: A Short Communication. Dose-Response. 2011; 9 (4): p.dose-response.1.doi: 10.2203/dose-response.10-046.hart . | Open in Read by QxMD
  39. Bhopal RS. Concepts of Epidemiology. Oxford University Press ; 2016
  40. Kelley K. Good practice in the conduct and reporting of survey research. International Journal for Quality in Health Care. 2003; 15 (3): p.261-266.doi: 10.1093/intqhc/mzg031 . | Open in Read by QxMD
  41. Messiah A, Castro G, Rodríguez de la Vega P, Acuna JM. Random sample community-based health surveys: does the effort to reach participants matter?. BMJ Open. 2014; 4 (12): p.e005791.doi: 10.1136/bmjopen-2014-005791 . | Open in Read by QxMD
  42. Cadoret RJ. Adoption Studies.. Alcohol Health Res World. 1995; 19 (3): p.195-200.
  43. Nyati KK, Nyati R. Role of Campylobacter jejuni Infection in the Pathogenesis of Guillain-Barré Syndrome: An Update. Biomed Res Int. 2013; 2013: p.1-13.doi: 10.1155/2013/852195 . | Open in Read by QxMD
  44. Bours MJL. Tutorial: A nontechnical explanation of the counterfactual definition of effect modification and interaction. J Clin Epidemiol. 2021; 134: p.113-124.doi: 10.1016/j.jclinepi.2021.01.022 . | Open in Read by QxMD
  45. B. Burt Gerstman. Epidemiology Kept Simple. John Wiley & Sons ; 2013
  46. Alexander et al.. Epidemiologic Research and Information Center Notebook. Epidemiologic Research and Information Center (ERIC) Notebook. 2015.
  47. Fedak KM, Bernal A, Capshaw ZA, Gross S. Applying the Bradford Hill criteria in the 21st century: how data integration has changed causal inference in molecular epidemiology.. Emerg Themes Epidemiol. 2015; 12: p.14.doi: 10.1186/s12982-015-0037-4 . | Open in Read by QxMD
  48. Kummar S, Doroshow JH. Phase 0 Trials: Expediting the Development of Chemoprevention Agents. Cancer Prevention Research. 2011; 4 (3): p.288-292.doi: 10.1158/1940-6207.capr-11-0013 . | Open in Read by QxMD
  49. Sharp et al.. STAT 509 Design and Analysis of Clinical Trials. Department of Statistics Online Programs. 2018.
  50. $An Introduction to Applied Epidemiology and Biostatistics.
  51. Kochanek et al.. Deaths: Final Data for 2017. National Vital Statistics Reports CDC. 2019; 68 (Nine).
  52. Martin J.. Case Control Odds Ratio. CTSPedia. 2012.
  53. Kanchanaraksa S.. Evaluation of Diagnostic and Screening Tests: Validity and Reliability. The Johns Hopkins University Bloomberg School of Public Health. 2008.
  54. Parikh R, Mathai A, Parikh S, Chandra Sekhar G, Thomas R. Understanding and using sensitivity, specificity and predictive values. Indian J Ophthalmol. 2008; 56 (1): p.45.doi: 10.4103/0301-4738.37595 . | Open in Read by QxMD
  55. Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests.. The Clinical biochemist. Reviews. 2008; 29 Suppl 1: p.S83-7.
  56. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve.. Radiology. 1982; 143 (1): p.29-36.doi: 10.1148/radiology.143.1.7063747 . | Open in Read by QxMD
  57. Masic I, Miokovic M, Muhamedagic B. Evidence Based Medicine - New Approaches and Challenges. Acta Informatica Medica. 2008; 16 (4): p.219.doi: 10.5455/aim.2008.16.219-225 . | Open in Read by QxMD
  58. Szumilas M. Explaining odds ratios.. Journal of the Canadian Academy of Child and Adolescent Psychiatry = Journal de l'Academie canadienne de psychiatrie de l'enfant et de l'adolescent. 2010; 19 (3): p.227-9.
  59. U.S. Preventive Services Task Force Procedure Manual. Updated: December 1, 2015. Accessed: August 10, 2020.

Icon of a lock3 free articles remaining

You have 3 free member-only articles left this month. Sign up and get unlimited access.
 Evidence-based content, created and peer-reviewed by physicians. Read the disclaimer