a b/README.md 

  1
  
# Predicted risk of lung cancer based on the UK Biobank risk prediction model





  
  

  2
  
This repository provides Stata code to calculate the predicted risk of





  
  

  3
  
lung cancer based on supplied information and the UK Biobank risk





  
  

  4
  
prediction model.





  
  

  5
  





  
  

  6
  
## How to use this code





  
  

  7
  
You will need a copy of Stata, version 12 or higher, with the





  
  

  8
  
user-written commands stpm2 and rcsgen installed (`ssc install stpm2`, `ssc install rcsgen`). Open stata





  
  

  9
  
and change directory to the root of this repository. Running





  
  

  10
  
```





  
  

  11
  
do predict_lca_risk.do





  
  

  12
  
```





  
  

  13
  
will calculate the predicted risk of lung cancer. By default this will





  
  

  14
  
take covariate information from the file 'input.csv', calculate the





  
  

  15
  
2-year cumulative probability of lung cancer for each record therein,





  
  

  16
  
and save the results in the file 'output.csv'. These default input,





  
  

  17
  
output, and time-horizons can be changed by editing the parameters in





  
  

  18
  
the configuration block at the beginning of the file





  
  

  19
  
'predict\_lca\_risk.do'.





  
  

  20
  





  
  

  21
  
## Input specification





  
  

  22
  
The input file must be an ASCII plain text CSV file, and must contain





  
  

  23
  
the following variables:





  
  

  24
  





  
  

  25
  
variable name | description | type | valid values





  
  

  26
  
--------------|:------------|:-----|:-------------





  
  

  27
  
age | age in years | real | [40,70]





  
  

  28
  
smoke\_stat | smoking status | integer | 0 (never smoker), 1 (former smoker), 2 (current smoker)





  
  

  29
  
male | male sex | integer | 0 (female), 1 (male)





  
  

  30
  
previous\_cancer | Previously diagnosed with an invasive cancer | integer | 0 (no), 1 (yes)





  
  

  31
  
fhist\_lungca | one or more first-degree relatives who have been diagnosed with lung cancer | integer | 0 (no), 1 (yes)





  
  

  32
  
emph\_bronch | history of emphysema or bronchitis | integer | 0 (no), 1 (yes)





  
  

  33
  
allergy | history of hayfever, allergic rhinitis, or eczema | integer | 0 (no), 1 (yes)





  
  

  34
  
fev1\_max | forced expiratory volume in one second (FEV1) | real | [0.5,10]





  
  

  35
  
smkfmr\_quitage | for former smokers only, the age at which they quit | real | less than current age





  
  

  36
  
ncig | for current and former smokers only, the average number of cigarettes smoked per day | real | (0,80]





  
  

  37
  
stop1day\_easy | for current smokers only, how difficult would it be to not smoke for one day | integer | 0 (difficult or very difficult), 1 (easy or very easy)





  
  

  38
  





  
  

  39
  
Other variables can be included in the input file, but they are





  
  

  40
  
ignored by the program.





  
  

  41
  





  
  

  42
  
## Output file





  
  

  43
  
The output file contains the same variables as the input file, with





  
  

  44
  
the addition of the variable 'cif\_lung' which contains the predicted





  
  

  45
  
probability (Cumulative Incidence Function evaluated at a given time





  
  

  46
  
horizon) for each input observation.