Statistics Built Around the Evidence Question
Bioexcel supports medical device and IVD manufacturers with statistical study design, sample-size planning, Statistical Analysis Plans, clinical endpoint analysis, PMCF analysis and diagnostic-performance statistics.
Our approach connects study objectives, endpoints, data structure and regulatory evidence needs before analysis begins.
Does Bioexcel Provide Biostatistics for Medical Device and IVD Studies?
Yes. Bioexcel provides biostatistical support for medical device clinical investigations, PMCF, retrospective studies and IVD clinical performance programs, including sample-size calculation, SAP development, endpoint analysis, survival analysis, sensitivity, specificity, PPA and NPA.
Bioexcel Lifesciences & Research LLP provides biostatistical services for medical device and IVD studies, including statistical study design, sample-size calculation, Statistical Analysis Plans, endpoint analysis, PMCF statistics, survival analysis and diagnostic performance analysis.
Medical Device Studies Are Not Always Conventional Drug Trials
Statistical planning may need to account for:
- Device success
- Technical success
- Procedural success
- Learning curve
- Implant duration
- Device durability
- Reintervention
- Revision
- Long-term follow-up
- Performance goals
- Historical benchmarks
- Small study populations
- Rare events
- Single-arm designs
- Missing follow-up
- Device generations
- Operator effects
The statistical design should reflect the clinical and regulatory evidence problem.
From Clinical Question to Final Statistical Evidence
Clinical Objective
- Safety objective
- Performance objective
- Clinical-benefit objective
- Diagnostic objective
- PMCF objective
Output: Statistical Question
Endpoint Definition
- Primary endpoint
- Secondary endpoints
- Safety endpoints
- Exploratory endpoints
Output: Endpoint Framework
Statistical Design
- Single-arm
- Comparative
- Superiority
- Non-inferiority
- Performance goal
- Diagnostic-accuracy design
- Time-to-event approach
Output: Statistical Design
Sample Size
- Effect size
- Expected event/performance rate
- Precision
- Confidence
- Power
- Attrition
Output: Sample-Size Rationale
SAP Development
- Populations
- Endpoints
- Methods
- Missing data
- Subgroups
- Sensitivity analyses
Output: Statistical Analysis Plan
Database Lock
- Clean dataset
- Final coding
- Reconciled safety
- Locked database
Output: Final Analysis Dataset
Statistical Analysis
- Tables
- Listings
- Figures
- Estimates
- Confidence intervals
- Hypothesis tests
Output: Statistical Results
Clinical Interpretation
- Statistical result
- Clinical relevance
- Limitations
- Regulatory implication
Output: Evidence Conclusion
What Biostatistical Services Does Bioexcel Provide?
Bioexcel provides sample-size calculations, statistical study design, SAP development, safety and performance analysis, PMCF analysis, survival analysis, IVD diagnostic-performance statistics, subgroup analysis, sensitivity analyses and final statistical reporting.
What biostatistical services does Bioexcel provide?
Bioexcel provides statistical study design, sample-size calculation, SAP development, medical device safety/performance analysis, PMCF analysis, survival analysis and IVD diagnostic-performance statistics.
Can Bioexcel provide only statistics?
Yes. Biostatistics can be provided independently of full-service clinical study management.
What studies can Bioexcel analyze?
Bioexcel supports medical device clinical investigations, prospective and retrospective PMCF, registries and IVD clinical performance studies.
Sample Size Should Match the Study Objective
Bioexcel can support sample-size calculations for:
Inputs May Include
- Expected performance
- Event rate
- Effect size
- Precision
- Confidence level
- Statistical power
- Significance level
- Attrition
- Loss to follow-up
- Invalid/evaluable rate
Can Bioexcel Calculate Sample Size for a Medical Device Study?
Yes. Bioexcel can calculate and justify sample size based on the study objective, endpoint, expected safety or performance rate, required precision, power, confidence level and anticipated attrition.
PMCF Sample Size Should Not Be Arbitrary
Potential approaches include:
- Precision around complication rate
- Precision around device success
- Estimate of revision rate
- Estimate of reintervention
- Long-term survival
- Comparison with clinical benchmark
- Confirmation of predefined performance
Can Bioexcel Calculate Sample Size for PMCF?
Yes. Bioexcel can support PMCF sample-size planning using the specific post-market clinical objective, expected event or performance rate, required precision, follow-up and anticipated loss to follow-up.
Positive and Negative Cohorts Need Separate Planning
For an IVD study, sample size may need to be justified for:
Positive Cohort
- Expected sensitivity
- Expected PPA
- Precision
- Confidence interval
Negative Cohort
- Expected specificity
- Expected NPA
- Precision
- Confidence interval
Additional Cohorts
- Weak positives
- Near-cutoff samples
- Serotypes
- Genotypes
- Age groups
- Cross-reactive conditions
- Interfering conditions
Can Bioexcel Calculate IVD Sample Size?
Yes. Bioexcel supports IVD sample-size planning based on expected sensitivity, specificity, PPA or NPA, required precision, confidence level, subgroup requirements and anticipated invalid or non-evaluable samples.
One Primary Endpoint Should Answer the Main Study Question
A primary endpoint should be:
- Clinically meaningful
- Measurable
- Predefined
- Statistically analyzable
- Relevant to intended purpose
- Relevant to the regulatory question
Medical Device
- Device success
- Major adverse event rate
- Reintervention-free survival
- Functional improvement
IVD
- Sensitivity
- Specificity
- PPA
- NPA
Secondary Endpoints Add Clinical Context
Potential secondary analyses:
- Safety
- Technical success
- Procedural success
- Quality of life
- Reintervention
- Hospitalization
- Subgroup outcomes
- Device deficiencies
- User performance
- Invalid-test rate
Secondary endpoints should support rather than dilute the main evidence objective.
Single-Arm Medical Device Study Design
Single-arm studies may be considered where:
- Randomization is not practical
- The clinical benchmark is established
- The device is highly specialized
- Historical controls are scientifically defensible
The benchmark should be justified using reliable external evidence.
Compare Device Performance With a Predefined Clinical Benchmark
A performance goal may be derived from:
- Published literature
- Registry data
- Previous clinical studies
- State of the art
- Clinically accepted performance thresholds
with appropriate confidence.
Can Bioexcel Support a Performance-Goal Study?
Yes. Bioexcel can support performance-goal selection, sample-size calculation, hypothesis definition and statistical analysis where an appropriately justified external benchmark is available.
Demonstrating Better Performance
A superiority study tests whether one treatment/device outcome is statistically better than the comparator.
- Null hypothesis
- Alternative hypothesis
- Effect size
- Power
- Significance level
- Sample size
Demonstrating That Performance Is Not Unacceptably Worse
A non-inferiority design may be appropriate where the objective is to demonstrate that the investigational device is not worse than a comparator beyond a predefined margin.
- Comparator selection
- Non-inferiority margin
- Clinical justification
- Analysis population
- Missing data
- Sensitivity analysis
Can Bioexcel Design a Non-Inferiority Medical Device Study?
Yes. Bioexcel can support non-inferiority study design, including margin justification, sample-size calculation, analysis populations, Statistical Analysis Plan and sensitivity analyses.
Comparative Medical Device Analysis
Potential Comparisons
- Investigational vs predicate
- Investigational vs standard of care
- Two device generations
- Device vs alternative procedure
- User group comparisons
Potential Methods
- Difference in proportions
- Difference in means
- Relative risk
- Odds ratio
- Hazard ratio
- Regression models
Predefine the Analysis Before Looking at the Final Results
A SAP may include:
- Study objectives
- Endpoints
- Analysis populations
- Sample-size assumptions
- General statistical methods
- Safety analysis
- Performance analysis
- Missing-data handling
- Protocol deviations
- Subgroups
- Sensitivity analyses
- Interim analysis where applicable
- Tables/listings/figures
Can Bioexcel Prepare a Statistical Analysis Plan?
Yes. Bioexcel can develop study-specific SAPs for medical device, PMCF, retrospective and IVD studies, defining analysis populations, endpoints, statistical methods, missing-data rules and planned outputs.
Define Who Is Included in Each Analysis
Enrolled Population
All enrolled subjects.
Safety Population
Subjects with relevant device exposure.
Full Analysis Set
Subjects meeting predefined analysis criteria.
Per-Protocol Population
Subjects without specified major protocol deviations.
Evaluable IVD Population
Specimens meeting predefined clinical-performance criteria.
The definitions should be study specific.
Quantify and Interpret Clinical Safety
- Adverse-event incidence
- Serious adverse events
- Device-related events
- Procedure-related events
- Device deficiencies
- Reintervention
- Revision
- Mortality
- Event timing
Potential Outputs
Demonstrate Whether the Device Performs as Intended
Potential Endpoints
- Device success
- Technical success
- Procedural success
- Functional improvement
- Anatomical outcome
- Durability
- Clinical success
Potential Analysis
- Proportions
- Means
- Change from baseline
- Confidence intervals
- Hypothesis tests
- Repeated measures
Analyze Outcomes Across Follow-Up
Medical device studies often include:
Potential Methods
- Change from baseline
- Repeated measures
- Mixed models where appropriate
- Trend analysis
- Time-by-treatment effects
Long-Term Device Outcomes Need Time-Based Analysis
Relevant Outcomes
- Device survival
- Revision-free survival
- Reintervention-free survival
- Event-free survival
- Time to complication
Potential Methods
Can Bioexcel Perform Kaplan-Meier Analysis for Medical Device Studies?
Yes. Bioexcel can perform Kaplan-Meier and other appropriate time-to-event analyses for outcomes such as device survival, revision, reintervention and long-term clinical events.
Historical Data Require Different Statistical Thinking
Retrospective datasets may include:
Challenges
- Different follow-up durations
- Missing variables
- Incomplete records
- Variable clinical practice
- Confounding
- Selection bias
Potential Methods
- Descriptive analysis
- Survival analysis
- Regression
- Stratification
- Sensitivity analysis
- Multivariable adjustment
- Propensity methods where justified
Can Bioexcel Analyze Retrospective PMCF Data?
Yes. Bioexcel can analyze retrospective PMCF datasets using methods appropriate to variable follow-up, missing data, confounding, survival outcomes and the predefined clinical objective.
Missing Data Should Be Planned for Before Analysis
Questions to Address
- How much data are missing?
- Which variables are affected?
- Why are data missing?
- Does missingness threaten the primary endpoint?
- What assumptions are reasonable?
Potential Methods
- Complete-case analysis
- Imputation where appropriate
- Sensitivity analysis
- Mixed models
- Time-to-event censoring
The selected method should be justified.
Determine Which Deviations Affect the Analysis
Potential Categories
- Eligibility deviations
- Device-use deviations
- Follow-up deviations
- Endpoint timing deviations
- Major protocol deviations
The SAP Should Define Impact On
- Full analysis set
- Per-protocol population
- Sensitivity analysis
Understand Whether Performance Differs Across Important Populations
- Age
- Sex
- Risk category
- Device size
- Investigator/site
- Disease severity
- Device generation
- Serotype/genotype for IVD
- Geographic region
Subgroup analysis should be interpreted cautiously when sample sizes are small.
Multi-Site Studies May Need Site-Level Assessment
Considerations
- Enrollment by site
- Outcome by site
- Protocol deviations
- Operator experience
- Clinical-practice differences
- Recruitment bias
Approaches
- Site-stratified analysis
- Random/fixed effects where justified
- Sensitivity analysis
Some Devices Depend on Operator Experience
Relevant For
- Surgical robotics
- Novel procedures
- Implantable devices
- Operator-dependent systems
Potential Assessment
- Procedure time over cases
- Technical success over cases
- Complications by operator experience
- Early vs later cases
Core Diagnostic Performance Measures
Diagnostic Sensitivity
The proportion of reference-positive cases correctly identified by the index test.
Diagnostic Specificity
The proportion of reference-negative cases correctly identified by the index test.
Potential Reporting
Agreement When the Comparator Is Not an Accepted Reference Standard
PPA
Positive Percentage Agreement.
NPA
Negative Percentage Agreement.
Can Bioexcel Calculate PPA and NPA?
Yes. Bioexcel provides PPA, NPA, overall agreement and associated confidence intervals for IVD studies where agreement measures are appropriate to the comparator strategy.
Diagnostic Performance Structure
| Reference Positive | Reference Negative | |
|---|---|---|
| Index Positive | True Positive | False Positive |
| Index Negative | False Negative | True Negative |
From This Structure, Appropriate Measures May Include
Evaluate Test Performance Across Thresholds
For quantitative or semi-quantitative assays, ROC analysis may help evaluate:
- Sensitivity
- Specificity
- Decision thresholds
- Area under the curve
- Cut-off performance
Only use ROC methods where relevant to the assay and intended diagnostic claim.
Predefine How Discordance Is Handled
- Index positive / reference negative
- Index negative / reference positive
- Repeat result
- Additional characterization result
The primary analysis should follow the predefined study methodology rather than changing retrospectively based on discordant outcomes.
Track the Performance of the Test System Itself
- Invalid tests
- Repeat tests
- Non-evaluable specimens
- Operator-related invalids
- Device-related invalids
These may be relevant to both clinical performance and usability.
Intended-User Performance
- Successful test completion
- Correct sample collection
- Correct device operation
- Correct result interpretation
- IFU comprehension
- User-error rate
- Questionnaire results
Show the Precision of the Estimate
A single percentage does not show statistical uncertainty.
Sensitivity = 95%
is more informative when presented with:
95% confidence interval
The required precision should inform sample-size planning.
A Small P-Value Is Not the Clinical Conclusion
Bioexcel statistical interpretation should consider:
- Effect size
- Confidence interval
- Clinical relevance
- Benchmark
- Study design
- Limitations
- Benefit-risk
Does Bioexcel Help With Clinical Interpretation of Statistics?
Yes. Bioexcel can support interpretation of statistical findings in the context of the clinical endpoint, study objective, state of the art and intended regulatory evidence use.
Structured Statistical Outputs
Tables
- Demographics
- Baseline
- Safety
- Device performance
- Primary endpoint
- Secondary endpoint
- Subgroups
Listings
- Serious events
- Device deficiencies
- Protocol deviations
- Reinterventions
Figures
- Kaplan-Meier curves
- Outcome trends
- Forest plots where appropriate
- ROC curves
From Analysis to Documented Evidence
Potential statistical deliverables:
- Statistical Analysis Plan
- Statistical output package
- Tables/listings/figures
- Statistical analysis report
- Interpretation summary
- Dataset documentation
These Can Support
Statistics Start With a Clean Dataset
This integration helps avoid common issues such as missing endpoint variables or inconsistent data definitions.
Statistical Design Should Be Finalized Before Recruitment
- Endpoint definition
- Sample-size calculation
- Hypothesis
- SAP
- Final analysis
Statistics Should Answer the Residual Clinical Question
Evidence Gap: Long-term revision risk.
Endpoint: Revision-free survival.
Method: Kaplan-Meier.
Evidence Output: 3-year revision-free survival estimate with confidence interval.
Statistical Results Become Clinical Evidence
Final analyses can support the CER by providing:
- Safety estimates
- Performance estimates
- Clinical-benefit outcomes
- Confidence intervals
- Long-term outcomes
- Subgroup findings
- Statistical limitations
Already Have the Study and Data?
Bioexcel can provide functional biostatistics support for:
- Sponsor-run studies
- Another CRO's study
- Existing databases
- PMCF datasets
- Retrospective datasets
- IVD datasets
Potential Scope
Can Bioexcel Provide Biostatistics Without Running the Study?
Yes. Bioexcel can provide standalone biostatistical services for sponsor-managed or externally managed studies, subject to availability and quality of the required study data and documentation.
Need a Second Opinion on the Statistical Design?
Bioexcel can review:
- Sample-size rationale
- Primary endpoint
- Hypothesis
- Performance goal
- Non-inferiority margin
- SAP
- Missing-data method
- Final analysis
Potential Situations
- Notified Body query
- Protocol revision
- Underpowered design concern
- Inconsistent sample size
- Incorrect sensitivity/specificity calculations
- Retrospective-analysis concerns
Biostatistics for India and European Programs
Bioexcel biostatistics connects directly with regional clinical delivery.
Why MedTech Sponsors Use Bioexcel for Biostatistics
Medical Device & IVD Focus
Statistical designs reflect device and diagnostic evidence needs.
Study Design Integration
Biostatistics begins during protocol development.
Data Integration
Statistics connect directly with Bioexcel EDC and data-management workflows.
PMCF Expertise
Support for prospective, retrospective and long-term post-market analysis.
Diagnostic Expertise
Sensitivity, specificity, PPA, NPA and related IVD analyses.
Regulatory Interpretation
Results are presented in a form useful for clinical and regulatory documentation.
Biostatistics Deliverables
Depending on scope:
- Statistical consultation
- Sample-size report
- Statistical-design rationale
- Randomization specifications where applicable
- Statistical Analysis Plan
- Analysis population definitions
- TLF shells
- Statistical programming outputs
- Tables
- Listings
- Figures
- Statistical analysis report
- Analysis datasets
- Statistical interpretation
- Regulatory query response support
Biostatistics in Practice
Cardiovascular Device
Question: Long-term event-free outcome
Method: Time-to-event analysis — Output: Kaplan-Meier estimates and clinical interpretation
Retrospective PMCF
Challenge: Variable follow-up and missing records
Bioexcel Role: Analysis strategy, survival analysis, sensitivity assessment
IVD Clinical Performance
Question: Diagnostic performance
Bioexcel Role: Positive/negative cohort calculation, sensitivity, specificity and confidence intervals
Frequently Asked Questions About Biostatistics
Yes. Bioexcel calculates sample size for medical device clinical investigations, PMCF, retrospective studies and IVD performance studies.
Bioexcel integrates biostatistics into the clinical evidence strategy from study design through regulatory interpretation.