Definition
Longitudinal study: a research design that collects data from the same participants repeatedly, at multiple time points over an extended period of time, to track changes in health, behavior or other variables. Longitudinal studies are usually observational, such as cohort and panel studies, but repeated measurements in clinical trials follow the same logic.
- The opposite is a cross-sectional study, which measures each person only once, at a single point in time.
- Main types: prospective cohort, retrospective cohort, panel study, birth cohort and repeated cross-sectional study (different people at each wave).
- The biggest limitation is attrition: participants who drop out over the years can bias the results.
Updated October 21, 2026
Working with follow-up visits and long-term data? Trial design, CRFs and data queries are part of the VIARES Clinical Research Associate course.
Contents
- What is a longitudinal study?
- Types of longitudinal studies
- Longitudinal vs cross-sectional study
- Longitudinal study design and data collection
- Analysis of longitudinal data
- Famous longitudinal studies
- Benefits and limitations of longitudinal studies
- Longitudinal studies in clinical research
- Example of a longitudinal study: a hypothetical patient cohort
- What a CRA checks in long-running studies
- FAQ
- Sources
- Where you learn this at VIARES
A single blood test shows where a patient is today, not whether the value is rising or falling. For that, you must measure the same person again. That is the core idea of longitudinal research.
What is a longitudinal study? Definition and meaning
Longitudinal research involves repeated measures of the same participants at multiple time points: at baseline, then at planned follow-up points over months, years or decades. Each person’s results are compared with their own earlier results.
The NCI Dictionary of Cancer Terms describes a longitudinal cohort study as research that follows large groups of people over a long time, assessing the same people many times. A review in the Journal of Thoracic Disease (Caruana et al., 2015) adds that longitudinal studies are generally observational, collect quantitative and/or qualitative data, and may also be experimental.
Because the same individuals are measured over time, longitudinal data show change over time within each person. That is why this research method is used to study how diseases develop, how risk factors relate to outcomes and how treatments perform long term.
Types of longitudinal studies
Longitudinal research designs differ in when data are collected and whether the same people are measured at every wave. In medicine, the main type is the cohort study.
| Type of longitudinal study | How it works | Typical use |
|---|---|---|
| Prospective cohort study | Follows a group forward; exposures are measured before outcomes occur | Risk factors for disease, long-term outcomes of treatment |
| Retrospective longitudinal study | Uses existing records (health records, registries) covering past time points | Fast, cheaper questions with good historical data |
| Panel study | A sample answers the same questionnaire or longitudinal surveys at regular intervals | Health behavior, quality of life, social research |
| Birth cohort | Follows people born in the same period across the life course | Child development, aging, life-course epidemiology |
| Repeated cross-sectional study | Same questions at several time points, but different people each time | Population trends; cannot show change within individuals |
| Clinical trial with repeated visits | Assigned intervention, measured at scheduled visits | Efficacy and safety over the treatment and follow-up period |
Caruana et al. also describe linked panels, which link data collected for other purposes, such as health records or registries, into individual-level longitudinal patient data.
Longitudinal vs cross-sectional study
A cross-sectional study measures a population at a single point in time. It is quick but static: it shows associations, not sequences.
| Longitudinal study | Cross-sectional study | |
|---|---|---|
| Measurements | Same participants, multiple time points | Each participant once |
| Shows | Change over time and sequence of events | Prevalence and associations at one moment |
| Cause and effect | Stronger support, because exposure is measured before outcome | Weak; the direction of an association is often unclear |
| Time and cost | High; risk of dropout | Low; no follow-up needed |
Cross-sectional studies are faster to set up and often come first, as a preliminary step before a longitudinal study.
Longitudinal study design and data collection
A longitudinal study design starts with a precise research question, then decides what is measured, how often, for how long and in whom.
- Time points: measurement frequency should match the expected speed of change.
- Duration: long enough for the outcome to occur. There is no fixed minimum.
- Methods to collect data: questionnaires, interviews, exams, lab tests, imaging and record linkage. A longitudinal survey design keeps questions identical across waves.
- Consistency: Caruana et al. stress that methods must be identical across sites and consistent over time, with unique codes linking each person’s data.
- Retention: keep participants in the study; exit interviews show why people leave.
In clinical research, longitudinal tracking happens through scheduled visits recorded in a case report form. Good data collection protects data quality: a new lab method halfway through can look like real change.
Analysis of longitudinal data
Longitudinal data analysis must respect one fact: measurements from the same person are correlated. Caruana et al. warn that ignoring this can lead to wrong conclusions. Common approaches:
- Within-subject change: comparing each participant with their own baseline, as in a within-subjects design.
- Mixed-effects models: an average trajectory plus individual variation; they can use participants with missed visits.
- Generalized estimating equations (GEE): population-average effects that account for repeated measures.
- Time-to-event (survival) analysis: whether and when an event occurs, such as progression or death.
Missing data are the central problem. ICH E9 calls missing values a potential source of bias, asks for handling methods pre-defined in the protocol and for checking how sensitive results are to them. ICH E9(R1) adds planning for intercurrent events, such as treatment discontinuation, additional treatment or death. These choices belong in the statistical analysis plan.
Famous longitudinal studies (examples)
Some well-known examples of a longitudinal study have run for generations:
- Framingham Heart Study: started in 1948 with 5,209 men and women aged 30 to 62 from Framingham, Massachusetts. According to the NHLBI, participants return about every two to six years; their children joined in 1971 and a third generation in 2002.
- Harvard Study of Adult Development: the Harvard longitudinal study has followed two groups of men for more than 80 years: 268 Harvard graduates from the classes of 1939 to 1944 (Grant Study) and 456 men who grew up in inner-city Boston (Glueck Study). It studies predictors of healthy aging.
- 1958 National Child Development Study: a British birth cohort following 17,415 people born in England, Scotland and Wales in one week of March 1958, with data sweeps from birth to age 62.
Benefits and limitations of longitudinal studies
The benefits of longitudinal studies come from time; so do most of the disadvantages.
| Benefits | Limitations |
|---|---|
| Establish the sequence of events: exposure before outcome | Attrition: loss to follow-up reduces sample size and can bias results if dropouts differ |
| Track changes within individuals, not just group averages | Time and cost: years of funding, staff and infrastructure |
| Prospective data collection avoids recall bias | Confounding: without randomization, other factors can explain associations (see confounding variable) |
| Separate age, period and cohort effects | Panel conditioning: repeated questioning can change how participants answer or behave |
| Stronger evidence on causal relationships than a single snapshot | Changing methods: new lab assays, definitions or questionnaire versions break comparability |
Attrition threatens internal and external validity: the remaining sample may no longer represent the original population, and group comparisons may become unfair.
Longitudinal studies in clinical research
How can longitudinal studies help in the understanding of diseases? They show how a disease develops without treatment (its natural history) and how fast it worsens, the basis of disease progression modeling and endpoint selection. Uses include:
- Natural history studies before trials in rare diseases.
- Long-term outcomes after a trial ends, through an extended follow-up period.
- Registries and post-marketing studies that monitor the safety of approved treatments.
- Long-term follow-up after gene therapy: FDA’s 2020 guidance recommends observing participants for delayed adverse events for up to 15 years for some product types, such as integrating vectors and genome editing products, and up to five years for AAV vectors.
Most are observational studies. A randomized controlled trial also produces longitudinal outcomes, but randomization lets it test cause-and-effect relationships directly.
Example of a longitudinal study: a hypothetical patient cohort
Hypothetical example with invented numbers. A university hospital starts a prospective longitudinal cohort study of 400 adults with early chronic kidney disease. The research question: which factors at baseline predict faster decline in kidney function over 5 years?
- Visits every 6 months: 11 time points from baseline to year 5.
- At each visit: blood and urine tests, blood pressure, medication review and a quality-of-life questionnaire.
- Retention: 372 participants (93 %) at year 2, 312 (78 %) at year 5.
A cross-sectional analysis would only show that patients with higher blood pressure tend to have lower kidney function. The longitudinal approach shows who declines faster after starting with high blood pressure.
The 88 participants lost by year 5 matter. If the sickest left because they started dialysis, the remaining cohort looks healthier than it is. The analysis must treat dialysis and death as events, not missing data.
FAQ
What is the opposite of a longitudinal study?
The opposite of a longitudinal study is a cross-sectional study. A cross-sectional study measures each participant once, at a single point in time: a snapshot. A longitudinal study measures the same participants repeatedly, so it shows how variables change and in which order events happen.
Is a longitudinal study observational?
Usually, yes. Most longitudinal studies, such as cohort and panel studies, observe participants without assigning treatment. But longitudinal research can also be experimental: a clinical trial that measures participants at repeated visits collects longitudinal data. What defines it is repeated measurement of the same people.
How long does a longitudinal study have to be?
There is no fixed minimum length. A study is longitudinal when it measures the same participants at two or more time points. Duration depends on the research question: weeks for a fast-changing symptom, years for a chronic disease, decades for studies of aging such as the Framingham Heart Study.
What research methods are commonly used in longitudinal studies?
Common methods include repeated questionnaires and longitudinal surveys, interviews, physical examinations, laboratory tests, imaging and linkage to health records or registries. Quantitative methods dominate in medicine; qualitative interviews can explain why people change. Analysis uses methods for repeated measures, such as mixed-effects models.
What is a prospective longitudinal cohort study?
It is a study that recruits a group of people, measures their exposures at the start, and follows them forward to see who develops the outcome of interest. Because exposures are recorded before outcomes occur, it avoids recall bias and shows the sequence of events. The Framingham Heart Study is a classic example.
Sources
- National Cancer Institute: Definition of longitudinal cohort study (NCI Dictionary of Cancer Terms)
- Caruana EJ, Roman M, Hernández-Sánchez J, Solli P. Longitudinal studies. J Thorac Dis. 2015;7(11):E537–E540 (PubMed Central)
- ICH E9: Statistical Principles for Clinical Trials (ICH guideline)
- ICH E9(R1) Addendum: Estimands and Sensitivity Analysis in Clinical Trials (ICH guideline)
- FDA: Long Term Follow-up After Administration of Human Gene Therapy Products, Guidance for Industry (2020)
- National Heart, Lung, and Blood Institute: Framingham Heart Study (FHS)
- Harvard Study of Adult Development: History of the Study (Grant and Glueck Study)
- UCL Centre for Longitudinal Studies: 1958 National Child Development Study
Where you learn this at VIARES
Section 6, “Clinical Trial Design & Clinical Research Roles and Responsibilities”, of the VIARES Clinical Research Associate course covers the study protocol, trial design and methodologies. Section 8, “Monitoring and Closing Clinical Trial Sites”, covers conducting and documenting monitoring visits. Section 11, “Data Management for Clinical Research Associates”, covers the purpose and structure of the case report form, accurate data collection, data queries, and record keeping and archiving. VIARES also offers the Micro Certificates COHORT (prospective and retrospective cohort studies, design, data collection, analysis, limitations) and OBSERVATIONAL STUDY.
