RESEARCH DISCIPLINE
Statistical Analysis for Clinical Trials
Consulting for Randomized & Pragmatic Studies
Clinical trial statistical analysis for parallel-group, cluster-randomized and pragmatic trials, with the analysis following the prespecified estimand and plan, every departure from it recorded, and the results prepared as CONSORT 2025 and your target journal require.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
Statistical Analysis for Clinical Trials
Clinical trial statistical analysis is the statistical design, analysis and reporting of randomized studies in which participants are allocated to interventions and followed on a prespecified schedule, so that the primary comparison answers a question fixed in advance. Participants stop treatment, miss visits or receive other therapy after randomization, and each of these events changes what a treatment comparison means, so an analysis that excludes them or runs a complete-case t-test at the final visit answers a different question from the one the trial was designed to ask. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides clinical trial statistician support through in-house PhD-level statisticians, who work from your protocol to settle the estimand, the analysis populations and the model for repeated outcomes before the data are examined.
WHEN TO ENGAGE
Statistical Support at Different Stages of Clinical Trial Research
Trial teams usually reach out at a recognizable point, from the drafting of the analysis plan to a journal's questions about a completed trial report.
Writing the Statistical Analysis Plan Your protocol is approved or nearly so, and the primary estimand, analysis populations, missing-data strategy and multiplicity rules need to be written into a plan before unblinding.
Analyzing the Trial Data Follow-up is complete or an interim database is ready, and the primary and secondary analyses must be run as prespecified, together with the sensitivity analyses that the plan lists.
Preparing the CONSORT Report Results have to be written for a journal, with participant flow, prespecified outcomes, effect estimates with intervals and any departure from the plan reported openly.
Reviewing a Trial Analysis A trial was analyzed on a per-protocol set or on complete cases, and a coauthor, journal or funder has asked whether the analysis follows the prespecified estimand.
RESEARCH CONTEXT
Research Designs and Data We Work With in Clinical Trials
Clinical trial data analysis starts from how participants were allocated and followed, and each design carries its own unit of analysis and its own risks for the comparison.
01
Parallel-Group Randomized Trials
Participants allocated to two or more arms, with baseline measures and a primary endpoint assessed at scheduled follow-up visits.
02
Cluster-Randomized Trials
Clinics, wards or schools allocated as groups, with outcomes measured on individuals who share the allocation of their cluster.
03
Crossover and Within-Person Designs
Participants who receive several treatments in sequence, so that period and carryover effects have to be addressed in the analysis.
04
Pilot and Feasibility Trials
Small preliminary trials that test recruitment, retention and procedures, and whose estimates inform a later sample size but are not definitive.
05
Non-Inferiority and Equivalence Trials
Trials built around a prespecified margin, in which the confidence interval is read against the margin and the choice of population carries more weight.
06
Adaptive and Group Sequential Trials
Trials with planned interim analyses or design modifications, whose stopping rules and error control are fixed before the first look.
METHODOLOGICAL CONSIDERATIONS
Statistical Issues We Consider in Clinical Trials
These checks protect the comparison that randomization created, because most threats to a trial result arise after allocation, not before it.
Estimand and Intercurrent Events Population, treatment, outcome variable, summary measure and the handling of events such as stopping treatment, rescue medication or death are stated together, as ICH E9(R1) describes.
Analysis Populations The primary analysis follows the intention-to-treat principle and analyzes participants as randomized, and the per-protocol and safety populations are defined in the plan before unblinding.
Repeated Outcomes and Missing Visits Outcomes measured at several visits are analyzed with a repeated-measures model that uses every observed visit, not a t-test on those with a final visit, and the missing-data assumption is probed.
Randomization Structure Stratification factors used in randomization and the baseline value of the outcome enter the model as prespecified, and a cluster-randomized trial is analyzed at the level of allocation.
Multiplicity and Interim Looks Your plan names the primary endpoint, orders the secondary ones in a testing hierarchy and sets the alpha-spending rule for any interim look, and an unplanned look is reported as such.
Plan Signing and Data Locking The plan is signed and the database locked before unblinding, and any later change to the analysis is documented as a deviation with its reason and its effect on the result.
STATISTICAL METHODS
Methods Selected for the Research Question and Study Design
Each method family is chosen for the outcome and the data structure of your study, and the line under each title explains why it is used with trial data.
Longitudinal & Multilevel Analysis
Mixed models for repeated measures estimate the treatment difference at the primary visit from all observed assessments, and cluster-randomized trials use mixed models or GEE.
Comparative & Hypothesis Testing
A single-visit endpoint is compared by ANCOVA adjusted for baseline and the stratification factors, with exact tests where event counts are small.
Clinical, Survival & Epidemiological
Time-to-event endpoints such as progression or relapse use Kaplan-Meier curves and Cox models, with a competing-risks analysis when death prevents the outcome.
Regression & Predictive Modeling
Binary responders are modeled with baseline adjustment to give risk differences and ratios, and event counts such as exacerbations are modeled as rates with extra variation allowed.
Bayesian & Computational Methods
Missing outcomes are handled under a stated assumption with sensitivity analyses, and simulation checks the error rates and power of a design with interim looks.
Qualitative & Mixed Methods
A process evaluation run alongside a trial combines coded interviews with the trial data, so that findings on delivery and uptake can be read with the effect estimate.
DELIVERABLES
Statistical Outputs Prepared for Your Research
The outputs follow the structure of a trial, from the review of the protocol to the programs that reproduce every reported result from the locked data.
Data & Study Review The protocol, randomization structure, visit schedule, endpoint definitions and data dictionary are reviewed, and the draft analysis plan is checked against them before analysis.
Statistical Analysis The primary and secondary analyses as prespecified on the intention-to-treat population, with the sensitivity, per-protocol and subgroup analyses the plan lists, and model diagnostics.
Effect Estimates & Uncertainty Treatment effects with confidence intervals on the prespecified scale, absolute and relative differences for binary outcomes and, where relevant, the interval read against the margin.
Tables & Figures A CONSORT flow diagram, baseline characteristics by arm, primary and secondary outcome tables, visit-by-visit profile plots and forest plots of prespecified subgroups.
Statistical Interpretation Results are explained against the estimand, secondary and exploratory findings are kept apart from the primary one, and deviations from the plan are described, not smoothed over.
Analysis Files The analysis plan, programs in SAS, R, Stata or SPSS syntax and a documented analysis dataset, so that each table can be reproduced from the locked data.
REPORTING
Reporting Clinical Trials
The guideline follows the design of the trial, and each extension is applied only where it fits.
Randomized trials report to CONSORT 2025, and the protocol and analysis plan are written to SPIRIT 2025. The extensions apply by design: cluster randomized trials, pilot and feasibility trials and non-inferiority trials each add items on the unit of allocation, the purpose of the estimates and the margin. The participant flow, the prespecified primary outcome with its estimand, the effect estimate with its interval and every departure from the analysis plan are reported from the locked analysis, so each number can be matched to the plan. Trial registration and data sharing statements follow the ICMJE recommendations and the instructions of the target journal.
ILLUSTRATIVE EXAMPLE
How the Analysis May Look in Practice
An illustration of how the analysis of a pragmatic trial is fixed in advance and carried through to a result that can be reported openly.
ILLUSTRATIVE EXAMPLE · PRIMARY CARE TRIAL RESEARCH · UNITED STATES
A primary care research network ran a pragmatic randomized trial of a structured self-management program for adults with type 2 diabetes, with glycemic control measured at several scheduled visits.
Some participants stopped attending the program and some missed later visits, and the draft plan had proposed a t-test on participants with a complete final visit. Before the database was unblinded, the plan was revised to define a treatment-policy estimand under the intention-to-treat principle, with a mixed model for repeated measures adjusted for the baseline value and the randomization stratum. Missing visits were assumed to be missing at random in the primary analysis, and a tipping-point analysis examined how different the missing outcomes would have to be to overturn the conclusion. A per-protocol analysis was listed as supplementary.
Outcome: The network received the treatment difference at the primary time point with its 95% confidence interval, the sensitivity analyses, a CONSORT 2025 flow diagram, the signed analysis plan and the programs.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
SOFTWARE
Statistical Software for Clinical Trials
Trial analyses are run in the package your team or sponsor already uses, with programs or syntax provided where applicable.
Statistical Support Within Your Research OnlineSPSS provides the agreed methodological, analytical, and statistical reporting support. The researcher or research team retains the scientific question, substantive interpretation, conclusions, and overall responsibility for the research.
FREQUENTLY ASKED QUESTIONS
Questions About Statistical Analysis for Clinical Trials
Short answers to what trial teams ask before sharing a study, from analysis populations and missing visits to the analysis plan, confidentiality and the assessment.
What statistical methods are used in clinical trials?
RCT statistical analysis follows the primary endpoint and the visit schedule: ANCOVA or a mixed model for repeated measures for continuous outcomes, logistic or Poisson-type models for binary and count outcomes, and Kaplan-Meier and Cox methods for time-to-event endpoints. Cluster-randomized trials add cluster-level or mixed models. The Statistical Methods pages cover each of these in turn.
Should the primary analysis be intention-to-treat or per-protocol?
The primary analysis should follow the prespecified estimand, which for most superiority trials means intention-to-treat. Analyzing participants as randomized keeps the groups comparable, while excluding participants after randomization can bias the comparison. Per-protocol results are therefore sensitivity analyses, except in non-inferiority trials, where both are examined.
How are missing visits handled in a trial analysis?
A complete-case t-test at the final visit drops participants with partial data and assumes that, within each arm, those without a final visit would have had outcomes like those with one. A mixed model for repeated measures uses every observed visit and is valid when data are missing at random given the model. Sensitivity analyses under missing-not-at-random assumptions, such as delta-adjusted multiple imputation or a tipping-point analysis, then show how far the conclusion depends on that assumption.
Can you write the statistical analysis plan for my trial?
We prepare the plan from your protocol, defining the estimand, analysis populations, models, missing-data handling and multiplicity rules, and it is finalized and dated before the data are unblinded. The clinical aims, the protocol and responsibility for the trial remain with your research team.
How is confidential trial data handled?
Research data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You send your research question, the protocol and any data or draft plan you have. A statistically trained Research Coordinator reviews them and responds the same day, and the analysis, deliverables, timeline and one fixed project fee are confirmed before work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Discuss Your
Clinical Trial Research With Us
Share your protocol, any draft analysis plan and available data, even if the trial is still being designed. We check the estimand against the design and confirm the approach before work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014