RESEARCH DISCIPLINE
Cancer Research Statistical Analysis
Endpoints Defined and Analyzed Correctly
For oncology trials, cohorts and registries: overall and progression-free survival, response rates and competing events, with each endpoint defined in the plan and every effect reported with its confidence interval.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Does Statistical Analysis in Cancer Research Involve?
Cancer research statistical analysis estimates how treatments, prognostic factors and exposures relate to outcomes such as overall survival, progression-free survival and tumor response, in data that are censored, often subject to competing events and frequently examined in subgroups. It covers the precise definition of each endpoint, the time-to-event models that suit the design, the handling of events that preclude the one of interest and the reporting of effects with their uncertainty. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, supports oncology investigators and graduate researchers: an in-house PhD-level statistician applies the endpoint definitions of your protocol and states how far the conclusions depend on them. What the findings mean for patients is a clinical judgment, and it stays with you.
WHEN TO ENGAGE
Stages of Oncology Research That Need a Statistician
Oncology projects tend to reach a statistician through a protocol, a dataset, a locked trial database or a reviewer, and each stage asks something different of the analysis.
Endpoints Being Defined A protocol or grant names overall survival, progression-free survival or response, and the event definitions, censoring rules and assessment schedule must be fixed before the analysis.
Registry or Cohort in Hand A retrospective cohort or registry has follow-up and outcomes, and the question is how to analyze them with competing events, missing stage and treatments that changed over time.
Trial Data to Analyze Trial data are ready for the primary and secondary endpoints, with stratification factors, interim looks and subgroup questions that the plan must handle consistently.
Reviewer Questions on Survival A reviewer doubts proportional hazards, the censoring of patients lost to follow-up or the reading of a hazard ratio, and the reply calls for added analyses, not a restated result.
WHAT THE SUPPORT COVERS
Statistical Components of Oncology Studies We Support
Oncology data analysis turns dated clinical records into endpoints and estimates, starting with the definition of each endpoint and ending with the reported table.
Endpoint Definitions
Oncology trial endpoints are each defined with a start date, an event, a censoring rule and an assessment schedule, so that overall and progression-free survival mean the same thing throughout.
Time-to-Event Estimation
Kaplan-Meier estimates and median times with their intervals are produced through Survival Analysis, and groups are compared with a test suited to the hazards.
Prognostic Models
Prognostic factors and treatment effects are modeled with Cox Regression, with proportional hazards examined and the hazard ratio reported with its interval.
Competing Risks
Where death from other causes precludes the event of interest, cumulative incidence is estimated through Competing Risks Analysis and not as one minus Kaplan-Meier.
Response and Safety Data
Response rates are reported with exact confidence intervals and adverse events by grade and arm, within the conventions of Clinical Trials.
Reporting Oncology Studies
Results follow CONSORT for trials and REMARK for prognostic marker studies, as expected in Medicine research, with subgroup findings labeled as such.
COMMON ISSUES
What Weakens Oncology Endpoint Analyses
Time-to-event data hold traps that simple comparisons do not, and these six recur in oncology manuscripts and in the reviewer reports that follow them.
Censoring Not Examined Patients lost to follow-up are censored without asking whether they left because their disease progressed, which undermines the assumption that censoring is uninformative.
Responders Against Nonresponders Survival of responders is compared with nonresponders from the start of treatment, though response is known only later, so the comparison favors patients who lived long enough to respond.
Hazard Ratio Read as Time Gained A hazard ratio is described as extending survival by a stated amount, though it is a ratio of rates, and a survival probability or restricted mean survival time says more.
One Ratio for Crossing Curves Curves that overlap early and separate late, or cross, as can happen with immunotherapy, are summarized by a single hazard ratio, which hides the way the effect changes over time.
Subgroups Mined Many subgroups are examined for a treatment effect with no prespecified plan and the few that look significant are reported, though interaction tests and multiplicity were never considered.
Surrogate Read as Survival An improvement in progression-free survival is reported as a survival benefit, though the link between them varies by cancer type and treatment and must be argued and not assumed.
DELIVERABLES
What You Receive for Your Oncology Analysis
Every survival time in these files is derived from documented dates and rules, so that a coinvestigator or a journal can audit each endpoint.
Statistical Analysis Plan A plan stating the endpoints, estimands, censoring rules, stratification, subgroup and multiplicity strategy, ready to accompany a protocol or a registry application.
Derived Endpoint Dataset A dataset with event times and censoring indicators derived from the raw dates, each rule documented so that the endpoints can be checked line by line.
Kaplan-Meier and Incidence Figures Kaplan-Meier and cumulative incidence curves with numbers at risk, with median or restricted mean survival times and their confidence intervals.
Model Output Tables Hazard ratios or subdistribution hazard ratios with 95% confidence intervals, the covariates in each model and the checks of proportional hazards.
Censoring and Endpoint Checks Results rerun under alternative censoring rules, endpoint definitions and handling of missed scans, so that a reader can see which conclusions hold under each choice.
Endpoint Programs and Reviewer Notes The programs that derive each endpoint from the raw dates and fit each model, with notes on how the findings and their limits can be described in a manuscript or a reply to reviewers.
ILLUSTRATIVE EXAMPLE
A Randomized Trial With a Delayed Treatment Effect
An illustration of how oncology endpoints are analyzed when the Kaplan-Meier curves do not behave as the standard model assumes.
ILLUSTRATIVE EXAMPLE · RANDOMIZED ONCOLOGY TRIAL RESEARCH · UNITED STATES
A clinical investigator had completed a randomized phase 3 trial of an immunotherapy against standard chemotherapy and needed the analysis of progression-free and overall survival for a manuscript.
The statistical plan fixed the endpoint definitions, the stratified log-rank comparison and the censoring rules before database lock, and it named restricted mean survival time as the alternative if the hazards were not proportional. The complication appeared in the curves, which overlapped for the first months and separated later, so a single hazard ratio would have misdescribed the effect. The prespecified restricted mean survival time comparison was reported beside the hazard ratio, the departure from proportional hazards was stated plainly, and patients lost to follow-up were reviewed to judge whether censoring could be informative. Subgroup results were labeled exploratory.
Outcome: The investigator received the hazard ratio and the restricted mean survival time difference, each with its 95% confidence interval, the curves with numbers at risk, the censoring review, the tables and the scripts.
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.
SOFTWARE
Survival Software Behind the Estimates
Trial units often work in SAS and academic groups in R or Stata, so the analysis is done where the endpoint derivations can be validated and rerun.
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.
RESEARCH RESPONSIBILITY
Statistical Support Within Your Own Oncology Research
OnlineSPSS works on the endpoints and the models, and every judgment about patients, treatments and clinical meaning is made by you and your coinvestigators.
Agreed Statistical Support OnlineSPSS provides the agreed methodological, analytical and statistical reporting support, defined in a scope that is confirmed before any work begins.
No Intellectual Contribution OnlineSPSS does not claim any intellectual contribution to your research. Authorship of the manuscript, thesis or trial report remains yours.
Your Research Decisions The research question, the clinical interpretation, the discussion and the conclusions remain yours, with overall responsibility.
Confidential Handling Research data and project details are handled confidentially, project information is not published, and an NDA is available where required.
FREQUENTLY ASKED QUESTIONS
Questions About Cancer Research Analysis
Responder comparisons, the limits of a hazard ratio and progression dated by scheduled scans are three points oncology reviewers press on.
What does statistical analysis for oncology and cancer research cover?
Statistical analysis for oncology and cancer research covers endpoint definitions such as overall and progression-free survival, time-to-event models, competing risks, response rates and effects reported with confidence intervals. Trials, registries and hospital cohorts are all supported, and observational findings are worded as associations.
How should responders and nonresponders be compared?
Not by comparing survival from the start of treatment, because response is observed only after time has passed, so responders must have lived long enough to respond. A landmark analysis from a fixed time point, or a model in which response is treated as time-dependent, avoids that bias.
Is a hazard ratio enough to describe a survival benefit?
A hazard ratio describes a relative difference in event rates under stated assumptions, and it says little about how much time or how many patients are gained. Reporting it with survival probabilities at fixed times or a restricted mean survival time difference, each with intervals, gives a fuller account.
How are progression dates handled when scans are scheduled at intervals?
Progression is found only at a scan, so its recorded date is the visit at which it was seen, not the day it began. When arms share a scan schedule the comparison stays fair; when schedules differ, an evaluation-time bias can arise, and interval-censored methods or a sensitivity analysis on the scan dates show how much it matters.
How do you protect my research data and confidentiality?
Your 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 share your research question, design and relevant materials, such as a protocol, endpoint definitions or a dataset with event dates, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Plan the Analysis
Around Your Endpoints
Share the protocol or research question, how progression, death and censoring were recorded and any reviewer comments, and the assessment will outline the endpoint analyses your data can support.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014