Healthcare Research and Evaluation
Expert-defined terms from the Advanced Certificate in Healthcare Case Management course at LearnUNI. Free to read, free to share, paired with a professional course.
Adequacy of Care – A measure of whether the services provided meet establ… #
Related terms: quality of care, clinical guidelines. Example: Evaluating post‑surgical follow‑up to determine if patients received recommended wound assessments. Challenges include defining appropriate benchmarks for diverse patient populations.
Adverse Event – An unintended injury or complication caused by healthcare… #
Related terms: patient safety, harm. Example: Medication error leading to a severe allergic reaction. Researchers must distinguish between preventable and non‑preventable events to inform improvement strategies.
Algorithmic Bias – Systematic and unfair discrimination embedded in compu… #
Related terms: machine learning, fairness. Example: A predictive readmission model that underestimates risk for minority groups because training data lacked sufficient representation. Addressing bias requires transparent model documentation and diverse datasets.
Allocation Bias – A distortion in study results that occurs when particip… #
Related terms: randomization, selection bias. Example: Assigning sicker patients preferentially to a new case‑management protocol can inflate perceived effectiveness. Proper randomization procedures mitigate this risk.
Analytic Validity – The extent to which a test accurately and reliably me… #
Related terms: measurement error, reliability. Example: Confirming that a survey instrument consistently captures patient satisfaction across multiple administrations. Poor analytic validity undermines the credibility of research findings.
Application Programming Interface (API) – A set of rules that allows soft… #
Related terms: interoperability, data exchange. Example: Using an EHR API to pull real‑time utilization data for a case‑management outcome study. Challenges include authentication, version control, and data mapping.
ArcGIS – A geographic information system used to visualize spatial patter… #
Related terms: spatial analysis, geocoding. Example: Mapping readmission rates by ZIP code to identify underserved neighborhoods. Limitations involve data privacy and the need for accurate address data.
Artificial Intelligence (AI) – Computer systems that perform tasks typica… #
Related terms: machine learning, deep learning. Example: AI algorithms predicting which patients are at high risk for avoidable hospitalizations. Ethical concerns include transparency, accountability, and potential bias.
Attrition – The loss of participants over the course of a study #
Related terms: dropout, loss to follow‑up. Example: 15 % Of enrolled patients discontinue a longitudinal case‑management evaluation, potentially skewing results if attrition is systematic. Strategies to minimize attrition include frequent contact and incentives.
Baseline Characteristics – Demographic and clinical attributes of partici… #
Related terms: covariates, confounding. Example: Age, comorbidity index, and insurance status recorded at study entry. Accurate baseline data are essential for appropriate adjustment in analyses.
Benefit‑Cost Analysis (BCA) – An economic evaluation that compares the mo… #
Related terms: cost‑effectiveness, return on investment. Example: Calculating the net savings from a case‑management program that reduces readmissions. Challenges include assigning monetary value to health outcomes and accounting for indirect costs.
Bias – Any systematic error that distorts the true relationship between e… #
Related terms: confounding, selection bias. Example: Over‑reporting of positive outcomes because participants who benefit are more likely to complete surveys. Recognizing and adjusting for bias is a core skill in research appraisal.
Binary Outcome – A variable with two possible states, such as “readmitted… #
“Not readmitted.” Related terms: dichotomous variable, logistic regression. Example: Measuring the presence of a medication error (yes/no). Binary outcomes simplify analysis but may obscure nuanced clinical information.
Blinding – Concealing group allocation from participants, clinicians, or… #
Related terms: masking, double‑blind. Example: Outcome assessors unaware of whether a patient received standard or enhanced case management. In real‑world evaluations, full blinding is often impractical, requiring alternative bias‑control methods.
Bootstrapping – A resampling technique that generates many simulated samp… #
Related terms: confidence interval, non‑parametric methods. Example: Using bootstrapped means to assess the stability of average length‑of‑stay reductions. Computational intensity can be a limitation with very large datasets.
Bradford Hill Criteria – A set of nine principles for evaluating causal r… #
Related terms: causality, epidemiologic evidence. Example: Applying temporality, strength, and consistency to judge whether a new discharge protocol truly reduces readmissions. Not all criteria must be met; they serve as a heuristic.
Broad Consent – A consent process allowing future, unspecified research u… #
Related terms: informed consent, secondary use. Example: Patients agree that their de‑identified claims data may be used for multiple quality‑improvement studies. Ethical oversight must ensure privacy protections.
Capitation – A payment model where providers receive a fixed amount per p… #
Related terms: fee‑for‑service, global budget. Example: A health plan pays a case‑management organization a per‑member monthly fee to coordinate care. Evaluations must account for incentives that may affect utilization patterns.
Case‑Control Study – An observational design that compares individuals wi… #
Related terms: odds ratio, retrospective. Example: Investigating whether lack of post‑discharge follow‑up is associated with readmission. Selection of appropriate controls is critical to avoid bias.
Case‑Mix Index (CMI) – A relative measure of the clinical complexity and… #
Related terms: DRG weight, severity of illness. Example: A higher CMI indicates sicker patients, which can influence benchmarking of length of stay. Adjusting for case mix is essential in comparative effectiveness research.
Case Management – A collaborative process that assesses, plans, implement… #
Related terms: care coordination, patient navigation. Example: A nurse case manager arranges home health services after discharge to prevent avoidable readmission. Evaluation focuses on outcomes such as utilization, satisfaction, and cost.
Cluster Randomized Trial – A study where groups (e #
G., Clinics) rather than individuals are randomly assigned to interventions. Related terms: intraclass correlation, design effect. Example: Randomizing entire hospital wards to receive a new discharge protocol. Analyses must adjust for clustering to avoid inflated type I error.
Cohort Study – An observational design that follows a group over time to… #
Related terms: prospective, relative risk. Example: Tracking patients receiving intensive case management versus usual care to compare readmission rates. Loss to follow‑up can threaten validity.
Comparator – The control or reference condition against which a new inter… #
Related terms: standard of care, benchmark. Example: Using usual discharge planning as the comparator for a novel transitional‑care program. Selecting an appropriate comparator influences the relevance of findings.
Confidentiality – The ethical and legal obligation to protect personal he… #
Related terms: HIPAA, data security. Example: De‑identifying patient records before analysis. Breaches can erode trust and jeopardize study approval.
Confounding – A situation where an extraneous variable influences both th… #
Related terms: adjustment, stratification. Example: Age may confound the link between case‑management intensity and readmission. Multivariable regression or propensity scoring can control for confounders.
Confidence Interval (CI) – A range of values that likely contains the tru… #
Related terms: precision, statistical significance. Example: A 95 % CI for a risk reduction of 0.8–1.2 Suggests the effect may not be statistically significant. Wide intervals indicate limited sample size or variability.
Continuous Quality Improvement (CQI) – An ongoing, systematic approach to… #
Related terms: Plan‑Do‑Study‑Act (PDSA), lean. Example: Using monthly readmission dashboards to refine case‑management protocols. Success depends on data availability, staff engagement, and iterative testing.
Cost‑Effectiveness Analysis (CEA) – An economic evaluation comparing the… #
Related terms: incremental cost‑effectiveness ratio (ICER), quality‑adjusted life year (QALY). Example: Determining the cost per avoided readmission for a transitional‑care nurse program. Challenges include selecting appropriate outcome measures and discount rates.
Cost‑Utility Analysis (CUA) – A form of CEA that expresses outcomes in te… #
Related terms: QALY, utility weights. Example: Comparing a case‑management intervention to usual care by estimating gained QALYs and associated costs. Valuing health states requires robust preference data.
Cross‑Sectional Study – An observational design that captures data at a s… #
Related terms: prevalence, survey. Example: Assessing patient satisfaction with case‑management services via a one‑time questionnaire. Cannot establish causality, but useful for hypothesis generation.
Data Governance – The set of policies, procedures, and standards that ens… #
Related terms: stewardship, metadata. Example: Implementing a governance framework for sharing EHR data across research partners. Poor governance can lead to inconsistent datasets and compliance issues.
Data Linkage – The process of combining records from separate data source… #
Related terms: deterministic matching, probabilistic matching. Example: Linking hospital discharge data with pharmacy claims to assess medication adherence post‑discharge. Accurate linkage requires reliable identifiers and careful privacy safeguards.
De‑identification – Removing or masking personal identifiers so that indi… #
Related terms: HIPAA Safe Harbor, anonymization. Example: Stripping names, dates of birth, and ZIP codes from a dataset before analysis. Over‑de‑identification may reduce data utility.
Denominator – The reference population used to calculate rates or proport… #
Related terms: numerator, incidence rate. Example: Number of readmissions (numerator) divided by total discharges (denominator). Accurate denominator definition is essential for valid benchmarking.
Design Effect – The factor by which sample size must be inflated to accou… #
Related terms: intraclass correlation coefficient (ICC), effective sample size. Example: A design effect of 1.5 Indicates that 50 % more participants are needed than in an individually randomized trial. Ignoring design effect can lead to underpowered studies.
Discharge Planning – The process of preparing a patient to leave a care s… #
Related terms: transition of care, care continuity. Example: Creating a medication reconciliation list and scheduling a follow‑up appointment before hospital exit. Effective planning reduces readmissions and adverse events.
Effect Size – A quantitative measure of the magnitude of a phenomenon #
Related terms: Cohen’s d, relative risk. Example: An effect size of 0.5 Indicates a moderate reduction in readmission risk after case‑management intervention. Reporting effect size alongside p‑values aids interpretation of clinical relevance.
Electronic Health Record (EHR) – A digital version of a patient’s paper c… #
Related terms: interoperability, clinical documentation. Example: Extracting medication lists from the EHR to evaluate adherence. Data quality varies with documentation practices and system design.
Electronic Medical Record (EMR) – A digital record used within a single o… #
Related terms: clinical workflow, data capture. Example: Using EMR data to track wound‑care follow‑up visits. Transitioning from EMR to EHR can affect longitudinal data continuity.
Eligibility Criteria – The set of inclusion and exclusion rules that defi… #
Related terms: sampling frame, generalizability. Example: Including adults aged 65 + with heart failure, excluding those with terminal illness. Strict criteria improve internal validity but may limit external applicability.
Empirical Study – Research based on observed and measured phenomena rathe… #
Related terms: quantitative research, data collection. Example: Conducting a survey of patient satisfaction after implementing a new case‑management protocol. Empirical results must be reproducible and transparent.
Endpoint – The primary outcome that a study is designed to assess #
Related terms: primary outcome, surrogate endpoint. Example: 30‑Day readmission rate as the endpoint for a transitional‑care intervention. Selecting clinically meaningful endpoints enhances relevance to stakeholders.
Ethical Review Board (ERB) – A committee that evaluates research proposal… #
Related terms: IRB, human subjects protection. Example: Submitting a protocol for a case‑management effectiveness study to the ERB for approval. Delays in review can affect project timelines.
Evidence‑Based Practice (EBP) – Clinical decision‑making that integrates… #
Related terms: clinical guidelines, knowledge translation. Example: Implementing a case‑management protocol that aligns with published readmission‑reduction evidence. Barriers include limited resources and provider resistance.
Explanatory Variable – Another term for independent variable; the factor… #
Related terms: predictor, exposure. Example: Intensity of case‑management contact (low, medium, high) as an explanatory variable for readmission. Clear definition aids model specification.
External Validity – The degree to which study findings can be generalized… #
Related terms: generalizability, transportability. Example: Results from a single academic medical center may not apply to rural hospitals. Researchers enhance external validity through diverse sampling and replication.
Factor Analysis – A statistical method for identifying underlying dimensi… #
Related terms: latent constructs, dimensionality reduction. Example: Using factor analysis to condense a 30‑item patient satisfaction survey into core domains. Requires adequate sample size and interpretability of factors.
Fidelity – The extent to which an intervention is delivered as intended #
Related terms: implementation adherence, protocol compliance. Example: Monitoring whether case managers follow the prescribed assessment checklist. Low fidelity can dilute observed effectiveness.
Fixed‑Effect Model – A meta‑analytic approach that assumes a single true… #
Related terms: random‑effects model, heterogeneity. Example: Pooling readmission reduction estimates from multiple case‑management trials using a fixed‑effect model when heterogeneity is low. Misapplication can lead to overconfident conclusions.
Fluency – In health communication research, the ease with which informati… #
Related terms: health literacy, readability. Example: Designing discharge instructions with high fluency to improve patient comprehension. Poor fluency can compromise self‑management.
Follow‑Up – The period after an intervention during which outcomes are mo… #
Related terms: post‑intervention assessment, surveillance. Example: Conducting a 30‑day phone call to assess readmission status. Inadequate follow‑up can miss delayed events.
Front‑Line Staff – Healthcare workers who interact directly with patients… #
Related terms: clinical workforce, patient-facing staff. Example: Engaging front‑line staff in quality‑improvement meetings to identify barriers to effective discharge planning. Their insights are critical for pragmatic research.
Generalizability – Synonymous with external validity; the applicability o… #
Related terms: population validity, transferability. Example: A case‑management program proven effective in a Medicare population may need adaptation for commercial insurers. Researchers report demographic and setting characteristics to aid readers in judging generalizability.
Goal‑Attainment Scaling (GAS) – A method for measuring individualized out… #
Related terms: patient‑centered outcomes, custom metrics. Example: Assigning each patient a GAS score based on achievement of medication adherence, mobility, and social support goals. While flexible, GAS requires rigorous scaling to ensure comparability.
Health Economics – The study of how health resources are allocated, inclu… #
Related terms: econometrics, resource utilization. Example: Applying health‑economic modeling to estimate long‑term savings from an intensive case‑management program. Requires assumptions about discount rates and willingness‑to‑pay thresholds.
Health Literacy – The capacity to obtain, process, and understand basic h… #
Related terms: patient education, communication. Example: Assessing literacy levels before providing written discharge instructions. Low health literacy is linked to higher readmission rates.
Health Information Exchange (HIE) – The electronic sharing of health info… #
Related terms: interoperability, data sharing. Example: Using an HIE to obtain outpatient follow‑up data for patients discharged from an acute care hospital. Technical standards and consent policies affect data completeness.
Health Services Research (HSR) – A multidisciplinary field that examines… #
Related terms: policy analysis, system evaluation. Example: Studying the impact of a statewide case‑management initiative on overall hospital utilization. HSR often integrates clinical, economic, and sociological perspectives.
Hierarchical Linear Modeling (HLM) – A statistical technique for analyzin… #
G., Patients within providers). Related terms: multilevel modeling, random effects. Example: Modeling readmission risk while accounting for clustering within case‑management teams. Requires sufficient clusters to estimate variance components reliably.
Hipaa – The Health Insurance Portability and Accountability Act, a U #
S. Law governing privacy and security of health information. Related terms: privacy rule, security rule. Example: Implementing encryption and access controls for a research database containing PHI. Non‑compliance can result in legal penalties and loss of trust.
Incidence – The number of new cases of a condition occurring in a defined… #
Related terms: incidence rate, new events. Example: Calculating the incidence of 30‑day readmission among patients who received case management. Distinct from prevalence, which includes existing cases.
Incidence Rate Ratio (IRR) – A measure comparing incidence rates between… #
Related terms: relative risk, rate ratio. Example: An IRR of 0.75 Indicates a 25 % reduction in readmission incidence for the intervention group. Confidence intervals convey precision.
Index of Disparity – A summary measure that quantifies differences in hea… #
Related terms: inequity metric, health disparities. Example: Calculating the index to compare readmission rates among racial/ethnic groups. Highlights areas needing targeted interventions.
Intention‑to‑Treat (ITT) Analysis – A principle that includes all partici… #
Related terms: per‑protocol, analysis set. Example: An ITT analysis of a case‑management trial preserves randomization benefits and reflects real‑world effectiveness. Missing data can complicate ITT implementation.
Interoperability – The ability of different information systems to exchan… #
Related terms: standards, FHIR. Example: Achieving interoperability between a hospital EHR and a community‑based case‑management platform enables seamless data flow for outcome tracking. Lack of interoperability hampers comprehensive evaluation.
Instrument Validity – The degree to which a measurement tool accurately c… #
Related terms: construct validity, content validity. Example: Validating a patient‑reported outcome measure for care coordination satisfaction. Invalid instruments produce misleading results.
Interval Censoring – A form of censoring where the exact time of an event… #
Related terms: survival analysis, right censoring. Example: Knowing that a readmission occurred between day 10 and day 15 post‑discharge but not the precise day. Specialized statistical methods are required.
Kaplan‑Meier Estimate – A non‑parametric method for estimating survival p… #
Related terms: survival curve, censoring. Example: Plotting time to readmission for patients with and without intensive case management. Provides visual comparison but does not adjust for covariates.
Key Performance Indicator (KPI) – A quantifiable metric used to evaluate… #
Related terms: benchmark, dashboard. Example: 30‑Day readmission rate as a KPI for a case‑management department. Selecting meaningful KPIs aligns measurement with strategic goals.
Kuder‑Richardson Formula 20 (KR‑20) – A reliability coefficient for dicho… #
Related terms: Cronbach’s alpha, internal consistency. Example: Calculating KR‑20 for a yes/no checklist assessing discharge completeness. Values above 0.7 Are generally acceptable.
Lag Time – The delay between an intervention and the observable effect on… #
Related terms: delay, temporal relationship. Example: A case‑management program may show reductions in readmission only after several months as processes stabilize. Recognizing lag time informs appropriate follow‑up periods.
Learning Health System – An integrated ecosystem that continuously genera… #
Related terms: rapid cycle improvement, data‑to‑knowledge. Example: Embedding real‑time analytics of readmission data to refine case‑management protocols. Requires culture of collaboration and robust data infrastructure.
Logistic Regression – A statistical model used for predicting binary outc… #
Related terms: odds ratio, binary predictor. Example: Modeling the odds of 30‑day readmission based on case‑management intensity, comorbidity score, and discharge disposition. Assumes linearity in the logit and independence of observations.
Loss Function – A mathematical expression that quantifies the error of a… #
Related terms: cost function, optimization. Example: Using cross‑entropy loss to train a neural network that predicts high‑risk patients for readmission. Selecting appropriate loss functions influences model performance.
Machine Learning (ML) – A subset of AI that enables computers to learn pa… #
Related terms: supervised learning, unsupervised learning. Example: Applying a random‑forest algorithm to identify predictors of avoidable readmissions. Requires careful validation to prevent overfitting.
Mixed‑Methods Research – An approach that combines quantitative and quali… #
Related terms: triangulation, convergent design. Example: Quantitative analysis of readmission rates complemented by focus groups with case managers to explore implementation barriers. Integration of findings can reveal nuanced insights.
Multicollinearity – A condition in which two or more predictors in a regr… #
Related terms: variance inflation factor (VIF), redundancy. Example: Including both length of stay and DRG weight may cause multicollinearity. Remedies include variable selection or dimensionality reduction.
Multimorbidity – The presence of two or more chronic conditions in an ind… #
Related terms: comorbidity, complex patients. Example: Patients with heart failure and diabetes often require intensive case management. Multimorbidity complicates risk adjustment and outcome attribution.
National Quality Forum (NQF) – An organization that endorses evidence‑bas… #
Related terms: measure endorsement, quality metrics. Example: Using NQF‑endorsed readmission measures to benchmark case‑management performance. Adoption promotes comparability across institutions.
Negative Predictive Value (NPV) – The probability that subjects with a ne… #
Related terms: specificity, screening test. Example: An NPV of 92 % for a risk‑stratification tool indicates most patients flagged as low‑risk indeed avoid readmission. NPV varies with prevalence.
Network Analysis – A set of methods for examining relationships among ent… #
Related terms: social network, graph theory. Example: Mapping referral patterns between primary care and specialty services to identify gaps in care coordination. Data completeness and privacy are common challenges.
Non‑Response Bias – Systematic error arising when individuals who do not… #
Related terms: survey bias, attrition. Example: Patients with poorer health may be less likely to complete a post‑discharge satisfaction survey, skewing results upward. Weighting adjustments can mitigate bias.
Observational Study – Research that assesses outcomes without manipulatin… #
Related terms: cohort, case‑control. Example: Analyzing existing EHR data to compare readmission rates before and after a case‑management policy change. Causal inference requires careful control for confounding.
Odds Ratio (OR) – A measure of association that compares the odds of an o… #
Related terms: logistic regression, relative odds. Example: An OR of 0.65 Suggests a 35 % reduction in odds of readmission with the intervention. For common outcomes, OR may overstate effect size relative to risk ratio.
Outcome Measure – A variable used to assess the effect of an intervention #
Related terms: primary outcome, secondary outcome. Example: 30‑Day all‑cause readmission is a common outcome measure in case‑management research. Selecting clinically meaningful and measurable outcomes is essential for impact.
Patient‑Reported Outcome Measure (PROM) – A tool that captures health sta… #
Related terms: quality‑of‑life, survey instrument. Example: Using a PROM to assess perceived care coordination after discharge. Validation ensures reliability across diverse patient groups.
Patient‑Reported Experience Measure (PREM) – An instrument that records p… #
Related terms: patient satisfaction, experience survey. Example: A PREM evaluating timeliness of follow‑up appointments. Complementary to PROMs, PREMs inform service‑delivery improvements.
Peer Review – The evaluation of scholarly work by experts in the same fie… #
Related terms: manuscript assessment, quality control. Example: Submitting a case‑management effectiveness study to a peer‑reviewed journal for publication. Peer review enhances credibility but can introduce publication bias.
Phenotype – Observable characteristics or traits resulting from the inter… #
Related terms: clinical phenotype, genotype. Example: Defining a high‑risk heart‑failure phenotype based on ejection fraction, BNP levels, and prior admissions. Accurate phenotyping improves risk stratification.
Placebo Effect – Improvement in outcomes attributable to patients’ expect… #
Related terms: expectancy, control group. Example: Patients receiving enhanced discharge counseling may report better adherence due to perceived attention. In non‑blinded studies, placebo effects can confound results.
Policy Analysis – The systematic evaluation of policy options, impacts, a… #
Related terms: health policy, stakeholder analysis. Example: Assessing the financial implications of a statewide incentive for case‑management programs. Requires integration of epidemiologic data, cost estimates, and political context.
Population Attributable Fraction (PAF) – The proportion of incidents in t… #
Related terms: attributable risk, public health impact. Example: Estimating the PAF of inadequate discharge planning for overall readmissions. Guides prioritization of interventions.
Predictive Modeling – The use of statistical or ML techniques to forecast… #
Related terms: risk score, algorithm. Example: Developing a readmission risk model using demographics, comorbidities, and prior utilization. Model calibration and validation are essential to ensure reliability.
Propensity Score Matching (PSM) – A technique that creates comparable gro… #
Related terms: causal inference, confounding control. Example: Matching patients who received case management with similar patients who did not, based on age, diagnosis, and prior admissions. Balance diagnostics confirm matching quality.
Prospective Cohort – A cohort study in which participants are followed fo… #
Related terms: longitudinal, temporal sequence. Example: Enrolling patients at hospital discharge and tracking readmissions over the next 90 days. Provides stronger evidence of temporality than retrospective designs.
Quality‑Adjusted Life Year (QALY) – A composite measure that combines len… #
Related terms: utility, cost‑utility analysis. Example: A case‑management program that adds 0.05 QALYs per patient at a cost of $500 yields an ICER of $10,000 per QALY. Thresholds for cost‑effectiveness vary by country.
Randomized Controlled Trial (RCT) – An experimental study where participa… #
Related terms: gold standard, randomization. Example: Randomizing hospitals to implement a bundled case‑management protocol versus usual care. RCTs minimize bias but can be costly and may have limited external validity.
Randomization – The process of assigning participants to groups by chance… #
Related terms: allocation concealment, balance. Example: Using a computer‑generated sequence to allocate patients to intensive versus standard case management. Proper randomization ensures comparability.
Rate Ratio – The ratio of incidence rates between two groups #
Related terms: incidence rate ratio, relative risk. Example: A rate ratio of 0.80 Indicates a 20 % reduction in readmission rate for the intervention group. Preferred when follow‑up time varies among participants.
Readmission – The return of a patient to an acute care setting within a s… #
Related terms: re‑hospitalization, post‑acute care. Example: 30‑Day all‑cause readmission is a common quality metric. Distinguishing avoidable from unavoidable readmissions is critical for fair evaluation.
Real‑World Evidence (RWE) – Data generated outside of controlled clinical… #
Related terms: observational data, pragmatic research. Example: Analyzing RWE to assess the impact of a case‑management program across multiple health systems. RWE offers broader applicability but may have data quality limitations.
Recall Bias – Systematic error caused by differences in participants’ abi… #
Related terms: self‑report, memory error. Example: Patients may underreport medication non‑adherence during a follow‑up interview, leading to inaccurate exposure assessment. Prospective data collection can reduce recall bias.
Recursive Partitioning – A decision‑tree method that splits data into hom… #
Related terms: classification tree, CART. Example: Using recursive partitioning to identify patient subgroups with highest readmission risk. Trees are intuitive but can overfit without pruning.
Reference Standard – The best available method for determining the true s… #
Related terms: gold standard, criterion validity. Example: Chart review by clinicians serves as a reference standard for confirming readmission causes. Imperfect reference standards can bias test performance estimates.
Regression to the Mean – The statistical phenomenon where extreme values… #
Related terms: statistical artifact, baseline effect. Example: Patients selected for high readmission risk may show improvement even without intervention, partly due to regression to the mean. Control groups help distinguish true effects.
Reliability – The consistency of a measurement instrument across time, it… #
Related terms: test‑retest, inter‑rater reliability.