Treatment-arm transition probability from a hazard ratio

Applies a hazard ratio to a comparator transition probability for one model cycle. The comparator probability is converted to a rate, the rate is multiplied by the hazard ratio, and the result is converted back, which reduces to a single power. The conversions are set out on the Transition Probability page.

Signature

p_T = 1 - (1 - p_C)^HR
Inputs
InputsDefinitionUnit
p_CProbability that a comparator person at risk at the start of the cycle has the event within the cycleprobability from 0 to 1
HRHazard ratio of treatment to comparator, assumed constant within the cycleratio, no unit
Output
p_TProbability that a treated person at risk at the start of the cycle has the event within the cycleprobability from 0 to 1

Function

Relative hazard function

Maps the hazards of an event in a treatment group and a comparator group at the same time t to their ratio. Under proportional hazards the ratio is constant over time, and it can then be applied to a baseline survival curve or a baseline transition probability to obtain absolute outcomes for the treatment group.

Try this function

Implementations

  • Excel

    Treatment probability from named cells

    Excel raises the comparator survival probability for the cycle to the power of the hazard ratio and subtracts the result from 1.

    =1-(1-ComparatorProb)^HazardRatio

Assumptions

  • Proportional hazards within the cycle

    The hazard ratio applies throughout the cycle. The comparator hazard need not be constant, because the relationship holds for the cumulative hazard over the cycle.

  • Single exit or cause-specific use

    The probability refers to one event from the starting state. When a state has competing exits, the ratio is applied to the cause-specific hazard of the event it describes, and the row is then rebuilt with the competing-hazards formula on the Transition Probability page.

Worked examples

  • Annual cycle with a comparator hazard of 0.10

    A comparator probability of about 0.0952 per annual cycle, from a hazard of 0.10 per year, and a hazard ratio of 0.70 give a treatment probability of about 0.0676, as in the article. Multiplying 0.0952 by 0.70 would give about 0.0666.

    p_C = 0.0952; HR = 0.70; p_T = 0.0676
  • Comparator probability of 0.20 per cycle

    With a comparator probability of 0.20 per cycle and a hazard ratio of 0.70, the treatment probability is about 0.1446, compared with 0.14 from direct multiplication, as in the Transition Probability article.

    p_C = 0.20; HR = 0.70; p_T = 0.1446

Common errors

  • Multiplying a probability by the hazard ratio

    Direct multiplication treats the hazard ratio as a ratio of probabilities. The error is small at low probabilities, 0.0666 against 0.0676 in the first example, but grows as probabilities rise or cycles lengthen, and a hazard ratio above 1 can then push the product above 1.

Sources

  • Relative rates applied under proportional hazards

    Jones E, Epstein D, García-Mochón L. A procedure for deriving formulas to convert transition rates to probabilities for multistate Markov models. Medical Decision Making. 2017;37(7):779-789. Introduction (relative rates from randomised trials are incorporated into rate estimates using the assumption of proportional hazards, and the simple rate to probability formula).

    View source →

  • Rate and probability conversion used with a hazard ratio

    Gidwani R, Russell LB. Estimating transition probabilities from published evidence: a tutorial for decision modelers. PharmacoEconomics. 2020;38(11):1153-1164. Equations 8 and 9.

    View source →

Canonical Identity

Stable URI · Machine-readable · Resolvable · CC BY 4.0