Docs Methodology Parametric distributions
Methodology

Parametric survival distributions

EasyHTA fits parametric survival models using maximum likelihood estimation via the flexsurv package (Jackson, 2016). Six standard distributions are supported, consistent with the recommendations of NICE DSU Technical Support Document 14 (Latimer, 2011). For the reconstruction step that precedes model fitting, see The Guyot algorithm.

Supported distributions

DistributionParametersHazard shapeTypical use in HTA
ExponentialrateConstantUseful as a baseline or sensitivity model; appropriate when a constant hazard is plausible.
Weibullshape, scaleMonotone increasing, monotone decreasing, or constant (exponential as a special case)Commonly used and frequently selected in oncology HTA; simple and supports monotone hazards.
Log-normalmeanlog, sdlogUnimodal: initial increase followed by later declineAppropriate when a unimodal hazard is plausible (early rise, later decline).
Log-logisticshape, scaleDecreasing if shape ≤ 1; unimodal if shape > 1Heavier tails than log-normal, so long-tail behaviour must be justified.
Gompertzshape, rateMonotone increasing, monotone decreasing, or constant (shape > 0, < 0, or = 0)Can represent exponentially changing hazards; background mortality should be assessed or incorporated separately rather than implicitly absorbed.
Generalised gammamu, sigma, QFlexible; nests Weibull, gamma, and log-normal as special casesUseful flexible reference model; nested special cases can be compared via constrained likelihood comparisons where statistically valid.

Model selection

EasyHTA computes AIC and BIC for all fitted models. Lower values indicate better fit relative to model complexity. AIC and BIC are internal fit diagnostics against the observed data only; they do not validate the plausibility of the extrapolated tail. Model choice should also weigh hazard plausibility, external evidence, and uncertainty in the extrapolated period.

Statistical fit is not sufficient for model choice. Model selection must also consider the biological plausibility of the extrapolated hazard.

Parameter estimation

All models are fitted by maximum likelihood estimation. The full variance–covariance matrix is included in the Excel export, supporting propagation of correlated parameter uncertainty in downstream probabilistic sensitivity analyses (PSA), provided the parameter scale (flexsurv estimates positive parameters on the log scale), ordering, and any back-transformations are handled correctly by the receiving model.

TSD14 alignment

  • All six standard distributions mentioned in TSD14 are available
  • Model comparison outputs include log-likelihood, AIC, and BIC
  • Visual inspection overlays (fitted curves vs KM) are included in all results views
  • The variance–covariance matrix is exported for PSA use
  • Analysis metadata (run date, reconstruction method) is recorded in the Summary export sheet
  • For complex hazards, non-proportional hazards, cure assumptions, or situations where background mortality materially affects extrapolation, analysts should also consider methods discussed in NICE DSU TSD21 alongside TSD14

References

Guyot P, Ades AE, Ouwens MJNM, Welton NJ (2012). Enhanced secondary analysis of survival data: reconstructing the data from published Kaplan–Meier survival curves. BMC Medical Research Methodology, 12, 9. doi:10.1186/1471-2288-12-9

Jackson C (2016). flexsurv: A Platform for Parametric Survival Modeling in R. Journal of Statistical Software, 70(8), 1–33. doi:10.18637/jss.v070.i08

Latimer NR (2011, last updated 2013). NICE DSU Technical Support Document 14: Undertaking survival analysis for economic evaluations alongside clinical trials, extrapolation with patient-level data. Sheffield: NICE DSU. sheffield.ac.uk/nice-dsu

Parametric survival analysis and
cure-fraction modelling for health
technology assessment teams.

LinkedIn

Company


Graphics and branding by Ilari Ilvessalo · EasyHTA Oy 2026. All rights reserved.