Definition and Core Concepts in Health Research
Research is a systematic, objective and ethical process of asking questions, collecting and analysing evidence, and drawing conclusions that extend or verify knowledge. Health research investigates health, disease, healthcare delivery and the factors that influence population wellbeing.
Learning objectives
- Define research and health research.
- Explain the characteristics and stages of scientific research.
- Differentiate major research concepts and variables.
- Describe validity, reliability, bias and generalisability.
- Recognise the relationship between evidence, theory and practice.
Characteristics of sound research
Systematic
Follows an organised sequence from problem identification to reporting.
Empirical
Uses observable or measurable evidence rather than opinion alone.
Objective
Uses transparent methods designed to minimise personal influence and bias.
Ethical
Protects participants, respects autonomy and balances benefit against harm.
The research process
- Identify and define a relevant problem.
- Review existing literature and identify a knowledge gap.
- Formulate a title, objectives, questions or hypotheses.
- Select an appropriate design, population and sample.
- Develop and pretest data-collection tools.
- Obtain ethical and administrative approval.
- Collect, manage and analyse data.
- Interpret findings in relation to previous evidence.
- Draw conclusions, make justified recommendations and disseminate results.
Essential terminology
| Term | Meaning | Health example |
|---|---|---|
| Population | Complete group to which the question relates | All pregnant women attending a district’s facilities |
| Sample | Subset studied to learn about the population | 300 eligible women selected from four facilities |
| Variable | Characteristic that can take different values | Age, blood pressure, treatment group |
| Data | Recorded observations or measurements | Laboratory values and interview responses |
| Parameter | Numerical characteristic of a population | True district prevalence of anaemia |
| Statistic | Numerical value calculated from a sample | Sample prevalence of anaemia |
| Hypothesis | Testable prediction about variables | Exposure is associated with disease |
Variables and operational definitions
- Independent variable: presumed exposure, predictor or intervention.
- Dependent variable: measured outcome.
- Confounder: a third factor associated with both exposure and outcome that can distort their relationship.
- Extraneous variable: another factor that may influence observations.
- Operational definition: the exact procedure used to measure or classify a concept.
For example, “hypertension” should not remain vague. A protocol must specify the blood-pressure threshold, equipment, number and timing of measurements, position and treatment status.
Conceptual and theoretical frameworks
A concept is an abstract idea such as quality of life or adherence. A construct is a concept deliberately defined for study. A theory explains relationships among concepts, while a conceptual framework visually or logically organises the variables expected to influence the outcome. These frameworks guide measurement, analysis and interpretation.
Quality of evidence
| Concept | Key question |
|---|---|
| Validity | Does the method measure what it is intended to measure? |
| Reliability | Would repeated measurement under similar conditions give consistent results? |
| Accuracy | How close is the measurement to the true value? |
| Precision | How close are repeated measurements to one another? |
| Internal validity | Are the study conclusions credible for the participants studied? |
| External validity | Can findings reasonably apply to other relevant settings or populations? |
Bias, chance and confounding
Bias is systematic error introduced through selection, measurement, analysis or reporting. Random error is variation caused by chance and is reduced by adequate sample size and precise measurement. Confounding creates a misleading exposure–outcome estimate and may be addressed through design or analysis.
Primary and secondary research
- Primary research: investigators collect original data through surveys, observation, experiments or interviews.
- Secondary research: investigators analyse existing studies or datasets, including systematic reviews and secondary data analysis.
Research versus related activities
Research aims to generate generalisable knowledge using a protocol. Clinical audit compares current practice with an explicit standard. Monitoring routinely tracks implementation, while evaluation judges the merit or outcomes of a programme. These may use similar methods but differ in purpose.
Applied scenario: developing a research problem
A health centre reports frequent missed hypertension appointments. “Patients do not follow treatment” is an assumption, not yet a research problem. A researcher first reviews registers, confirms the scale of missed visits, examines previous studies and consults patients and staff. The refined problem may be: the proportion and determinants of missed follow-up appointments among adults receiving hypertension care at Facility X are unknown, limiting targeted retention interventions.
Possible outcome variable: missed appointment within 90 days. Possible predictors: travel time, medicine stock-outs, cost, knowledge, adverse effects and appointment reminders. Each variable requires a measurable operational definition.
From concept to measurement
| Abstract concept | Operational indicator | Possible measurement problem |
|---|---|---|
| Adherence | Percentage of prescribed doses taken in 30 days | Self-report may overestimate adherence |
| Quality of life | Score from a validated questionnaire | Tool may not be culturally appropriate |
| Poverty | Household asset index or income threshold | Income may be seasonal or underreported |
| Patient satisfaction | Multi-item rating scale after care | Courtesy bias may inflate responses |
Worked example: identifying variables
Question: “Among children under five, is household use of insecticide-treated nets associated with malaria infection?”
- Population: children under five in the defined area.
- Independent variable: verified household net use.
- Dependent variable: malaria infection defined by an approved diagnostic test.
- Potential confounders: age, season, housing, preventive treatment and local transmission intensity.
- Unit of analysis: the child, while some exposures are measured at household level.
Validity and reliability scenario
A weighing scale consistently reads 2 kg above the true weight. Measurements may be reliable because repetition gives similar values, but they are not valid because of systematic calibration error. A correctly calibrated scale that gives widely fluctuating readings lacks reliability. Good research instruments require both.
Bias in practice
- Selection bias: estimating community diabetes prevalence using only patients attending a referral clinic.
- Recall bias: cases remembering previous exposure more thoroughly than controls.
- Interviewer bias: probing one group more intensely.
- Social-desirability bias: respondents underreporting alcohol use.
- Publication bias: positive studies being more likely to appear in journals.
Mini case exercise
A researcher wants to determine whether night duty is associated with medication errors among nurses.
- Write a focused research question.
- Define “night duty” and “medication error” operationally.
- Identify the outcome, exposure and at least three confounders.
- Suggest one source of selection bias and one source of information bias.
- State how confidentiality could be protected.
Common proposal mistakes
- Choosing an enormous topic that cannot be studied with available time and resources.
- Using objectives that do not match the problem or analysis.
- Collecting variables with no planned purpose.
- Confusing absence of evidence with evidence of no effect.
- Making causal claims from a design that only demonstrates association.
- Ignoring local context, participant burden and dissemination of findings.
Review questions
- Define health research and list five characteristics of good research.
- Differentiate a population parameter from a sample statistic.
- Explain validity and reliability using a clinical measurement.
- Differentiate bias, random error and confounding.
- Why are operational definitions necessary?
Return to the Clinical Medicine Year 2 curriculum
References
- World Health Organization. Health Research Methodology: A Guide for Training in Research Methods.
- UAHEB. Guidelines on Research Proposal and Report Writing in Allied Health Training Institutions.
- Kothari CR, Garg G. Research Methodology: Methods and Techniques.
Educational content only. Research involving people, identifiable information or biological materials requires appropriate ethical and institutional approval.
Definition and Core Concepts in Health Research
Research is a systematic, objective and ethical process of asking questions, collecting and analysing evidence, and drawing conclusions that extend or verify knowledge. Health research investigates health, disease, healthcare delivery and the factors that influence population wellbeing.
Learning objectives
- Define research and health research.
- Explain the characteristics and stages of scientific research.
- Differentiate major research concepts and variables.
- Describe validity, reliability, bias and generalisability.
- Recognise the relationship between evidence, theory and practice.
Characteristics of sound research
Systematic
Follows an organised sequence from problem identification to reporting.
Empirical
Uses observable or measurable evidence rather than opinion alone.
Objective
Uses transparent methods designed to minimise personal influence and bias.
Ethical
Protects participants, respects autonomy and balances benefit against harm.
The research process
- Identify and define a relevant problem.
- Review existing literature and identify a knowledge gap.
- Formulate a title, objectives, questions or hypotheses.
- Select an appropriate design, population and sample.
- Develop and pretest data-collection tools.
- Obtain ethical and administrative approval.
- Collect, manage and analyse data.
- Interpret findings in relation to previous evidence.
- Draw conclusions, make justified recommendations and disseminate results.
Essential terminology
| Term | Meaning | Health example |
|---|---|---|
| Population | Complete group to which the question relates | All pregnant women attending a district’s facilities |
| Sample | Subset studied to learn about the population | 300 eligible women selected from four facilities |
| Variable | Characteristic that can take different values | Age, blood pressure, treatment group |
| Data | Recorded observations or measurements | Laboratory values and interview responses |
| Parameter | Numerical characteristic of a population | True district prevalence of anaemia |
| Statistic | Numerical value calculated from a sample | Sample prevalence of anaemia |
| Hypothesis | Testable prediction about variables | Exposure is associated with disease |
Variables and operational definitions
- Independent variable: presumed exposure, predictor or intervention.
- Dependent variable: measured outcome.
- Confounder: a third factor associated with both exposure and outcome that can distort their relationship.
- Extraneous variable: another factor that may influence observations.
- Operational definition: the exact procedure used to measure or classify a concept.
For example, “hypertension” should not remain vague. A protocol must specify the blood-pressure threshold, equipment, number and timing of measurements, position and treatment status.
Conceptual and theoretical frameworks
A concept is an abstract idea such as quality of life or adherence. A construct is a concept deliberately defined for study. A theory explains relationships among concepts, while a conceptual framework visually or logically organises the variables expected to influence the outcome. These frameworks guide measurement, analysis and interpretation.
Quality of evidence
| Concept | Key question |
|---|---|
| Validity | Does the method measure what it is intended to measure? |
| Reliability | Would repeated measurement under similar conditions give consistent results? |
| Accuracy | How close is the measurement to the true value? |
| Precision | How close are repeated measurements to one another? |
| Internal validity | Are the study conclusions credible for the participants studied? |
| External validity | Can findings reasonably apply to other relevant settings or populations? |
Bias, chance and confounding
Bias is systematic error introduced through selection, measurement, analysis or reporting. Random error is variation caused by chance and is reduced by adequate sample size and precise measurement. Confounding creates a misleading exposure–outcome estimate and may be addressed through design or analysis.
Primary and secondary research
- Primary research: investigators collect original data through surveys, observation, experiments or interviews.
- Secondary research: investigators analyse existing studies or datasets, including systematic reviews and secondary data analysis.
Research versus related activities
Research aims to generate generalisable knowledge using a protocol. Clinical audit compares current practice with an explicit standard. Monitoring routinely tracks implementation, while evaluation judges the merit or outcomes of a programme. These may use similar methods but differ in purpose.
Review questions
- Define health research and list five characteristics of good research.
- Differentiate a population parameter from a sample statistic.
- Explain validity and reliability using a clinical measurement.
- Differentiate bias, random error and confounding.
- Why are operational definitions necessary?
Return to the Clinical Medicine Year 2 curriculum
References
- World Health Organization. Health Research Methodology: A Guide for Training in Research Methods.
- UAHEB. Guidelines on Research Proposal and Report Writing in Allied Health Training Institutions.
- Kothari CR, Garg G. Research Methodology: Methods and Techniques.
Educational content only. Research involving people, identifiable information or biological materials requires appropriate ethical and institutional approval.
