Concept Architecture
Concept
Theoretically, the Fleming-Harrington Estimator is a non-parametric estimator of the survival function based on the cumulative hazard function. It was developed as an alternative to the Kaplan-Meier estimator and is founded on counting process theory and the Nelson-Aalen estimator of cumulative hazard. In health economics, it is used to estimate survival probabilities from censored time-to-event data that inform survival models, cost-effectiveness analyses and health technology assessments.
Mathematically, the Fleming-Harrington estimator is obtained by first estimating the cumulative hazard function using the Nelson-Aalen estimator and then transforming the cumulative hazard into a survival function through the exponential relationship between survival and cumulative hazard. This approach provides a consistent estimate of the survival function under independent censoring.
In practice, the Fleming-Harrington estimator is applied to right-censored survival data in much the same way as the Kaplan-Meier estimator. It is implemented in statistical software to estimate survival curves, compare treatment groups and provide empirical survival estimates for subsequent parametric extrapolation in health economic models.
Purpose
Used to estimate survival probabilities from censored time-to-event data using a cumulative hazard approach, supporting survival analysis, evidence synthesis and economic evaluation.
Mathematical Formulae
Primary Formula
?(t) = exp(??(t))
where ?(t) is the Nelson-Aalen estimate of the cumulative hazard.
Supporting Formulae
?(t) = ?(d? / n?)
where:
- d? = number of events at time i
- n? = number at risk immediately before time i
Related Mathematical Methods
- Nelson-Aalen estimator
- Kaplan-Meier estimator
- Cumulative hazard estimation
- Counting process methods
- Survival analysis
Example
A clinical trial follows 600 patients receiving two oncology treatments with observations censored after five years. The Fleming-Harrington estimator is used to estimate the empirical survival curve by first estimating cumulative hazard and then calculating survival as ?(t) = exp(??(t)). The resulting survival estimates are subsequently used to fit parametric survival models for lifetime cost-effectiveness analysis.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| SUM | =SUM(C2:C20) | Calculate cumulative hazard by summing event-to-risk ratios. |
| EXP | =EXP(-D20) | Convert cumulative hazard into estimated survival probability. |
| IF | =IF(B2>0,A2/B2,0) | Calculate the event-to-risk contribution at each event time. |
VBA (Optional)
Automate the calculation of cumulative hazard and corresponding Fleming-Harrington survival estimates from patient-level survival data.
Sources
- Fleming TR, Harrington DP. Counting Processes and Survival Analysis.
- Klein JP, Moeschberger ML. Survival Analysis: Techniques for Censored and Truncated Data.
- Collett D. Modelling Survival Data in Medical Research.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
- NICE. Health Technology Evaluation Manual.
Related Concepts (2)
Library
Publications
1
NICE DSU Technical Support Document 21: Flexible methods for survival analysis — Rutherford, Lambert, Sweeting, Pennington, Crowther, Abrams & Latimer, TSD 21 ed., 2020 (NICE Decision Support Unit (University of Sheffield))
Guidance extending standard survival analysis to flexible parametric methods — spline-based models, fractional polynomials, mixture and cure models, and relative-survival approaches — for capturing complex hazard functions in economic evaluation.
Frequently Asked Questions (6)
What is the Fleming-Harrington estimator?
A statistical technique for estimating the survival function from censored data, more flexible than the Kaplan-Meier estimator in weighting events over time.
Source: Fleming & Harrington 1991
What does the Fleming-Harrington estimator produce?
The Fleming-Harrington estimator produces an estimate of the survival function from censored data, closely related to the Kaplan-Meier estimate but derived through the cumulative hazard, which can give a smoother result and behaves well in the tails where events are sparse. It belongs to a family that allows different weight to be placed on events at different times. In most ordinary cases it and the Kaplan-Meier estimate agree closely. Its flexibility matters more in the associated tests. Klein and Moeschberger (2003) describe it.
Source: Klein & Moeschberger 2003
How does the Fleming-Harrington estimator work?
The Fleming-Harrington estimator works within the counting-process approach to survival, estimating the cumulative hazard, as the Nelson-Aalen estimator does by summing hazard contributions at each event time, and deriving survival from it through the relationship between survival and cumulative hazard. This yields a non-parametric survival estimate from censored data. The framework also underlies a family of tests with weights that emphasise events at different follow-up times, giving flexibility beyond a single fixed weighting.
Source: Fleming & Harrington 1991
How does the Fleming-Harrington estimator relate to Kaplan-Meier?
The Fleming-Harrington estimator and the Kaplan-Meier estimator both estimate the survival function non-parametrically from censored data and give similar results, but they arise from different approaches: Kaplan-Meier is the product-limit estimator built from conditional survival probabilities at each event, while Fleming-Harrington derives survival from the Nelson-Aalen cumulative hazard within the counting-process framework. The two estimators are closely related and asymptotically equivalent, with the Fleming-Harrington approach providing a framework that also supports flexibly weighted comparisons of survival.
Source: Kalbfleisch & Prentice 2002
What is the advantage of the flexible weighting in the Fleming-Harrington estimator?
The advantage of flexible weighting, associated with the Fleming-Harrington framework, is that events at different follow-up times can be given different importance in comparing survival, so that differences arising early or late can be emphasised as appropriate. The log-rank test weights all events equally, but where a treatment's effect is concentrated early or late, a weighted test can be more powerful. Flexible weighting thus allows the comparison to be tuned to the expected pattern of difference, improving sensitivity to it.
Source: Fleming & Harrington 1991
Where is the Fleming-Harrington approach used?
The Fleming-Harrington approach is used in survival analysis for estimating survival and cumulative hazard from censored data within the counting-process framework, and particularly for its family of weighted tests comparing survival between groups. These weighted tests are applied where treatment effects may differ over time, such as being larger early or late, so that a suitably weighted comparison is more powerful than the standard log-rank test. The approach is part of the modern theoretical foundation of survival analysis based on counting processes.
Source: Fleming & Harrington 1991
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 20 Oct 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-EM-SM-021
Stable URI · Machine-readable · Resolvable · CC BY 4.0