RESEARCH DISCIPLINE
Statistical Analysis for Health Economics Research
Consulting for Cost, Utility & Decision-Model Studies
Health economics statistical analysis for trial-based and claims-based economic evaluations, with zero and skewed costs modeled on the mean scale, costs and QALYs analyzed jointly and decision uncertainty shown as net benefit, and results reported to CHEERS 2022 and your journal's requirements.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
Statistical Analysis for Health Economics Research
Health economics statistical analysis is the statistical design, analysis and reporting of studies that estimate the costs and health outcomes of care, using patient-level data from trials, claims and registries or a decision model built on published inputs. Costs are right-skewed with many zeros, costs and QALYs are correlated within the same patient, and both are incomplete when follow-up ends early, so a workflow that log-transforms costs, analyzes costs and QALYs separately and reports the ICER as one number misstates both the mean and its uncertainty. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides economic evaluation statistics through in-house PhD-level statisticians, who fix the perspective, the time horizon and the handling of missing costs and utilities with you before the incremental analysis begins.
WHEN TO ENGAGE
Statistical Support at Different Stages of Health Economics Research
Researchers contact us from the choice of cost and utility instruments before a trial starts to a funder's question about a finished model.
Planning an Economic Evaluation You are designing a trial-based or model-based cost-effectiveness analysis and need the perspective, time horizon, cost and utility instruments and the primary comparison fixed before any data are collected.
Analyzing Costs and QALYs Patient-level costs and utilities are collected, and the incremental analysis, the handling of skewed and zero costs and of missing values, and the uncertainty around the result must be chosen and checked.
Quantifying Decision Uncertainty Your decision model or trial-based analysis needs a probabilistic sensitivity analysis, an acceptability curve and defensible distributions for each uncertain parameter.
Answering Reviewers and Payers A journal reviewer, funder or assessment body has questioned the cost model, the interval around the ICER, the discounting or the reporting, and the analysis needs review and revision.
RESEARCH CONTEXT
Research Designs and Data We Work With in Health Economics
Economic data come from case report forms, billing systems and published inputs, and each source has a structure that determines how costs and outcomes can be analyzed.
01
Trial-Based Evaluations
Costs and utilities collected alongside a randomized trial, with resource use from diaries or records, EQ-5D at fixed visits and loss to follow-up that affects both.
02
Claims and Administrative Costs
Payer or hospital billing records, where costs are skewed with a large share of zeros, follow-up depends on enrollment and the unit is often a person-year.
03
Registry and Cohort Cost Studies
Observational cohorts with linked costs, where the treated and untreated groups differ at baseline, so the cost comparison needs adjustment for who received the care.
04
Decision-Analytic Models
Markov or decision-tree models whose inputs come from trials, registries and the literature, each parameter needing a distribution for the probabilistic analysis.
05
Utility and Quality-of-Life Data
EQ-5D, SF-6D or other preference-based scores turned into QALYs over follow-up, with ceiling effects, baseline differences and missing waves to handle.
06
Cost-of-Illness and Program Costing
Studies of services or programs where the aim is total or mean cost for a defined population, not an incremental ratio between two options.
METHODOLOGICAL CONSIDERATIONS
Statistical Issues We Consider in Health Economics Research
Your cost and utility data are examined for skew, correlation, missing values and horizon before a ratio is reported, since each can move a conclusion more than the threshold does.
Mean Cost, Not a Transformed Mean A payer's total spending depends on the arithmetic mean cost, and the mean of logged costs is not that quantity, so models that estimate the mean directly are compared on fit before one is chosen.
Costs and QALYs Move Together Sicker patients tend to cost more and gain fewer QALYs, so costs and effects are correlated, and they are analyzed jointly, for example by resampling each patient's cost and QALY together.
Baseline Utility Differences Groups can differ in baseline utility by chance, and an unadjusted QALY difference then carries that imbalance, so QALYs are estimated with adjustment for baseline utility.
Missing and Censored Costs Means from patients with complete follow-up are biased when dropout relates to health. Multiple imputation inside the resampling, or weighting for censored costs, is used with the assumptions stated.
Uncertainty Around the ICER A ratio has no sensible interval when the denominator is near zero, so incremental net benefit and an acceptability curve over a stated range of thresholds are reported beside the bootstrap region.
Time Horizon and Discounting Costs and QALYs beyond the first year are brought to present value at the rate your jurisdiction's guidance sets, and extrapolation is tested with alternative survival curves, since it rests on assumptions.
STATISTICAL METHODS
Methods Selected for the Research Question and Study Design
The method families below follow the cost, utility and design of your evaluation, and each line says why a family suits economic data.
Regression & Predictive Modeling
Costs with many zeros are split into the chance of any cost and the amount given a cost, and visit or admission counts follow count models with time at risk as the exposure.
Bayesian & Computational Methods
Resampling gives the joint uncertainty of costs and QALYs, imputation covers missing values within each resample, and simulation drives the probabilistic analysis of a model.
Longitudinal & Multilevel Analysis
Utilities recorded at several visits are modeled with baseline adjustment before the area under the curve gives QALYs, and costs from several centers keep their clustering.
Clinical, Survival & Epidemiological
Observational cost comparisons adjust for who received the treatment, and survival curves fitted to trial follow-up are extrapolated across the horizon of a decision model.
Comparative & Hypothesis Testing
Mean costs are compared with intervals that allow unequal variances, and ANCOVA adjusts the QALY difference for baseline utility, since rank tests do not compare means.
Evidence Synthesis & Meta-Research
A decision model needs treatment effects from several trials, so a network meta-analysis supplies relative effects and their correlations for the probabilistic analysis.
DELIVERABLES
Statistical Outputs Prepared for Your Research
The outputs follow the structure of an economic evaluation, from the review of cost categories to the files that reproduce each result.
Data & Study Review Perspective, time horizon, cost categories, price year and the utility instrument are reviewed, and the pattern of zero, skewed and missing costs and utilities is documented.
Statistical Analysis The agreed cost, QALY and incremental analyses, with comparison of models for skewed costs, imputation inside the resampling and the planned scenario and subgroup analyses.
Effect Estimates & Uncertainty Incremental costs and QALYs with confidence intervals, the ICER with a bootstrap region, incremental net benefit and cost-effectiveness acceptability curves.
Tables & Figures Cost breakdown tables, cost-effectiveness planes, acceptability curves, tornado diagrams for deterministic sensitivity and the parameter table of a decision model.
Statistical Interpretation Results are explained for decision makers within the stated perspective and horizon, uncertainty is set out plainly, and a threshold is shown as a range, never as a verdict.
Analysis Files R, Stata, SAS or Python programs and, for a model, the parameter distributions and random seeds, so the probabilistic results can be regenerated.
REPORTING
Reporting Health Economics Research
Reporting starts from guidance written for economic evaluations, and the trial, claims or model behind the evaluation brings its own guidelines.
Economic evaluations report to CHEERS 2022, which asks for the perspective, time horizon, the rate applied to future costs and outcomes, currency and price year, the model structure and the sensitivity analyses. An evaluation run alongside a randomized trial adds CONSORT 2025 for the trial itself, and one built on claims or registry data follows STROBE with RECORD for routinely collected data. Modeling studies also draw on the ISPOR-SMDM good research practices for model structure, parameter estimation and uncertainty. Costs are stated in a currency and price year, uncertainty is given as intervals or acceptability curves, and the instructions of your target journal or assessment body are checked at the end.
ILLUSTRATIVE EXAMPLE
How the Analysis May Look in Practice
An illustration of how a trial-based economic question moves from patient-level costs and utilities to a result that states its uncertainty.
ILLUSTRATIVE EXAMPLE · COMMUNITY HEALTH CENTER ECONOMIC EVALUATION · UNITED STATES
A group of community health centers ran a pragmatic trial of a nurse-led diabetes care program and wanted to know whether it was cost-effective from the health system perspective over 12 months.
Costs were right-skewed with many zeros in some categories, and some patients left before follow-up ended, so costs and EQ-5D scores were incomplete. Before outcomes were examined, we fixed the perspective, the horizon and the primary analysis: QALYs from the area under the utility curve adjusted for baseline utility, and mean costs from a log-link model checked against a two-part model. Missing values were imputed within each bootstrap resample, with complete-case results as a check. Uncertainty was reported as incremental net benefit across a range of willingness-to-pay values and an acceptability curve, not as one ratio.
Outcome: The group received incremental costs and QALYs with 95% confidence intervals, the ICER with its bootstrap region, the acceptability curve, a report structured to CHEERS 2022 with CONSORT 2025 flow counts, 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.
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 Health Economics Research
Economic analyses are run in the software your group uses for patient-level and model data, with syntax or scripts 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 Health Economics
Answers to common questions from health economists, on skewed costs, ICER intervals, sample size, review, confidentiality and the assessment.
How are skewed costs with many zeros analyzed?
Costs are modeled on their original scale, because the decision depends on the mean, using a generalized linear model with a log link and a suitable variance function, or a two-part model for the chance of any cost and the amount given a cost. Both are compared, and a simple transformation is used only where it can be back-transformed to the mean. Our Statistical Methods pages describe the families involved.
Can an ICER be given a confidence interval?
An ICER is reported with a bootstrap cost-effectiveness plane, the incremental net benefit across a range of willingness-to-pay values and an acceptability curve, because a ratio is not a sufficient summary of uncertainty. The ICER itself is still given, beside the plane that shows how uncertain it is, and a ratio interval is avoided where the QALY difference may be near zero.
Does a trial need to be powered for costs and QALYs?
A trial is usually powered for its clinical primary outcome, and the economic outcomes are analyzed within that sample. We can estimate the precision the sample gives for incremental costs and QALYs, which is often wide because costs vary so much, and the results are reported as estimates with intervals, not as a test that decides the question.
Can you review a completed economic evaluation or model?
Yes. The review covers the perspective and horizon, the cost model, the handling of missing data, the distributions in the probabilistic analysis, the discounting and the reporting against CHEERS 2022, and states what stands and what should change. A model or script produced with an AI tool is reviewed in the same way.
How is confidential cost and claims data handled?
Study datasets, protocols and analysis files are handled confidentially, and project information is not published. 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, 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
Discuss Your
Health Economics Research With Us
Share your research question, perspective, available cost and utility data, even if only a protocol or a model outline exists. We review the structure of your data and confirm the approach before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014