Health Decision Strategies, LLC.


Survival Analysis, Markov and Cost Effectiveness

Online Calculator and Grapher


Data Submission Form
(Beta test, not for official use)

Data Entry Instructions
Use a separate row for each data case. The covariates (predictors) should come first, followed by the survival time, followed by the last-seen-status variable (1 if died, 0 if still alive or lost to follow-up). Values should be separated by commas. You can utilize datasets from another program, like a spreadsheet and save the data as a comma delimited text file that you can cut (ctrl-c)and paste (ctrl-v) into this data entry box.


Enter Number of Data Rows.
Enter Number of Covariates that are entered before the required "Time" and "Outcome" variables.

Caution: More than 1000 rows of data may run very slowly.
[--- You must at least enter "Time" and "Outcome" variables for several rows. ---]
Variables for Survival Analysis (for text box below)
Variable 1Variable 2Variable 3Var4 = TimeVar5 = Outcome
(can be group var **)covariate (optional)covariate (optional)(required)(required)
1,2 (default)0,1; 1,2 or continuous0,1; 1,2 or continuouscontinuous0=alive,1=fail,dead

** Group coding (variable 1) should be used if you wish to perform incremental cost-effectiveness analyses. As a minimum set of variables, you must enter at least numerical values for "Time" and "Outcome" separated by comma.
The default dataset is from reference 1 and 2, below.

STEPS TO FOLLOW FOR THIS SURVIVAL ANALYSIS AND MARKOV COST EFFECTIVENESS CALCULATOR
1.Enter Data, Comma Delimited
(on form 1, box above) Submit Data to Survival
Analysis Form.
2.
Plot Survival Curves
and Hazard Curves
(on form 2).
3.Calculate Transition Probabilities
and Enter Costs and Utilities
for Markov Analysis
(on form 2).
4.Submit Data to Markov Analysis
and Cost Effectiveness Form
(on form 3).

References:

1. Dalgaard, P. Statistics and Computing: Introductory Statistics with R. Chapter 12, Survival Analysis. Springer 2002.

2. Drzewiecki, KT and Andersen, PK. Survival with Malignant Melanoma, A Regression Analysis of Prognostic Factors. Cancer. 49:2414-2419, 1982.

3. Lawless, JF. Statistical Models and Methods for Lifetime Data. 1982, John Wiley & Sons, New York.

4. Pezzullo, JC. "Cox Proportional Hazards Survival Regression." at http://statpages.org/prophaz.html.

5. Briggs A. et al. Decision Modelling for Health Economic Evaluation. Oxford University Press, 2006.

6. Bewick V. et al. Statistics Review: Survival Analysis. Critical Care. 8(5):389-394, 2004.


Background
Survival Analysis, Proportional Hazards, Markov Analysis and Cost Effectiveness

General proportional hazards models (Cox regression) are a sub-class of survival models. For the purposes of this description, consider that survival models consist of two parts: the underlying hazard function, describing how hazard (risk) changes over time, and the effect parameters, describing how hazard is affected by other factors - such as the choice of treatment (group A vs B), or covariates such as age or tumor size, in a typical medical example.

With this survival analysis calculator, if you do not enter any of the variables 1,2, and 3, and thus only enter "time" and "outcome", the calculator will generate standard Kaplan-Meier data estimates and survival plot.

The proportional hazards assumption is that the effect parameters (covariates) multiply hazard: for example, if taking Treatment 'X' halves your hazard at time 0, it also halves your hazard at time 1, or time 0.5, or any given time, while the baseline hazard may vary. The effect parameter(s) estimated by any proportional hazards model can be reported as hazard ratios.

If you have two separate (treatment) groups in your survival dataset, from the survival results you can then generate "transition probabilities" (and add cost and utilities per Markov cycle) and this data can be submitted in the next form which will use Markov analysis to generate and plot incremental cost effectiveness results using "Group A" as the baseline, versus "Group B". The advantage of submitting the Survival results to the Markov calculator is that you can take survival data from a 5 year study and project out to 10, 20 years (cycles), or more. For this basic calculator there are only two Markov states: "No Event" (survived) and "Event" (died).
Health Decision Strategies, LLC.
Calculator developed by Health Decision Strategies, LLC