Treatment-Switching Adjustment for HTA

Definition and scope

Methods estimating a specified treatment-effect or survival contrast under a hypothetical switching strategy, distinguished from the randomized assignment effect and chosen according to the decision problem and identification assumptions. The target population, use, and analysis remain protocol decisions; this card explains the choices and their consequences rather than approving a study.

Operational specification

Define the trial, randomized arms, switch opportunities, reasons/timing, therapies after switching, prognostic histories, administrative cut, outcome, and target hypothetical strategy. State why adjustment is relevant to the HTA decision: not every treatment change should be removed. Distinguish rank-preserving structural failure-time, inverse-censoring-weighted, and two-stage approaches rather than treating them as interchangeable corrections.

Analysis and diagnostics

Choose a method using the switching process, data availability, and assumptions, such as a common treatment effect/structural model or measured predictors of switching with adequate support. Estimate uncertainty including the adjustment procedure and any re-censoring or extrapolation decisions. Compare plausible methods and report adjusted and unadjusted contrasts with clear targets.

When NOT to use and common errors

Do not simply drop switchers or censor them without addressing informative selection. Do not claim that the adjusted effect remains a randomized ITT effect, apply RPSFT assumptions without clinical justification, or extrapolate an adjusted tail without assessing support and structural uncertainty.

Worked example and implementation boundary

For a synthetic structural-time teaching transformation, a person has ten untreated months and five treated months. Assume, without estimating, a treatment time factor of two; transform treated time back by dividing by that factor. The hypothetical transformed time is 12.5 months. This is not an estimated RPSFT parameter, a complete re-censoring analysis, or a treatment-effect result. The code performs that explicitly stated calculation with synthetic inputs. It is not a complete fitted analysis and does not demonstrate the identification assumptions, estimate sampling uncertainty, or certify application in another population.

Data-source and reporting requirements

Preserve the provenance, dates, units, and observation rule for each input. Claims capture services and transactions under observable benefits; absence of a record is not proof that an event or resource use did not occur. EHR and registry measurements depend on the collection process and may miss care outside the system. Linked sources add reconciliation and selection questions. Primary collection can improve a missing domain but still requires valid measurement and follow-up. Report which of these limitations affects the stated use, which evidence supports the chosen inputs, and what remains uncertain.

Worked example

Scenario

For a synthetic structural-time teaching transformation, a person has ten untreated months and five treated months. Assume, without estimating, a treatment time factor of two; transform treated time back by dividing by that factor.

Dataset

Synthetic inputs with units defined in the scenario; no patient data.

Steps

Result

The hypothetical transformed time is 12.5 months. This is not an estimated RPSFT parameter, a complete re-censoring analysis, or a treatment-effect result.

Trade-offs

Code example

untreated_months = 10
treated_months = 5
assumed_factor = 2
result = untreated_months+treated_months/assumed_factor
assert abs(result - 12.5) < 1e-9
print(result)

untreated_months <- 10
treated_months <- 5
assumed_factor <- 2
result <- untreated_months+treated_months/assumed_factor
stopifnot(abs(result - 12.5) < 1e-9)
print(result)

data worked_example;
  untreated_months = 10;
  treated_months = 5;
  assumed_factor = 2;
  result = untreated_months+treated_months/assumed_factor;
  put result=;
run;

Citations

1. [1]Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. American Journal of Epidemiology. 2016;183(8):758-764.
2. [2]NICE Decision Support Unit. Survival analysis technical support documents: TSD 24, adjusting for treatment switching in clinical trials; update to TSD 16. Maintained official methods resource.