Datasets:
Sean Chaplin
Sean-Chaplin/
BladderCancerRecurrences
0
Downloads: 1
Diff of /README.md [f55bbf] .. [2b3461]

Switch to unified view

a/README.md b/README.md
1 
<div class="sc-cmRAlD dkqmWS"><div class="sc-UEtKG dGqiYy sc-flttKd cguEtd"><div class="sc-fqwslf gsqkEc"><div class="sc-cBQMlg kAHhUk"><h2 class="sc-dcKlJK sc-cVttbi gqEuPW ksnHgj">About Dataset</h2></div></div></div><div class="sc-jgvlka jFuPjz"><div class="sc-gzqKSP tNtjD"><div style="min-height: 80px;"><div class="sc-etVRix jqYJaa sc-bMmLMY ZURWJ"><pre class="uc-code-block">
 1 
## About Dataset
2 

  2
  
Data on recurrences of bladder cancer, used by many people to demonstrate methodology for recurrent event modelling.





  3
  
<span class="hljs-built_in">Data</span> <span class="hljs-keyword">on</span> recurrences of bladder cancer, used <span class="hljs-keyword">by</span> many people <span class="hljs-keyword">to</span> demonstrate methodology for recurrent event modelling.

  3
  





  4
  

  4
  
<table>





  5
  

  5
  
<thead>





  6
  
<table>

  6
  
<tr>





  7
  
<thead>

  7
  
<th>Column</th>





  8
  
<tr>

  8
  
<th>Description</th>





  9
  
<th>Column</th>

  9
  
<th>Format</th>





  10
  
<th>Description</th>

  10
  
</tr>





  11
  
<th>Format</th>

  11
  
</thead>





  12
  
</tr>

  12
  
<tbody>





  13
  
</thead>

  13
  
<tr>





  14
  
<tbody>

  14
  
<td><strong>Bladder Dataset 1</strong></td>





  15
  
<tr>

  15
  
<td></td>





  16
  
<td><strong>Bladder Dataset 1</strong></td>

  16
  
<td></td>





  17
  
<td></td>

  17
  
</tr>





  18
  
<td></td>

  18
  
<tr>





  19
  
</tr>

  19
  
<td>id</td>





  20
  
<tr>

  20
  
<td>Patient ID</td>





  21
  
<td>id</td>

  21
  
<td></td>





  22
  
<td>Patient ID</td>

  22
  
</tr>





  23
  
<td></td>

  23
  
<tr>





  24
  
</tr>

  24
  
<td>treatment</td>





  25
  
<tr>

  25
  
<td>Treatment received</td>





  26
  
<td>treatment</td>

  26
  
<td>Placebo, pyridoxine (vitamin B6), or thiotepa</td>





  27
  
<td>Treatment received</td>

  27
  
</tr>





  28
  
<td>Placebo, pyridoxine (vitamin B6), or thiotepa</td>

  28
  
<tr>





  29
  
</tr>

  29
  
<td>number</td>





  30
  
<tr>

  30
  
<td>Initial number of tumors</td>





  31
  
<td>number</td>

  31
  
<td>8=8 or more</td>





  32
  
<td>Initial number of tumors</td>

  32
  
</tr>





  33
  
<td>8=8 or more</td>

  33
  
<tr>





  34
  
</tr>

  34
  
<td>size</td>





  35
  
<tr>

  35
  
<td>Size (cm) of largest initial tumor</td>





  36
  
<td>size</td>

  36
  
<td></td>





  37
  
<td>Size (cm) of largest initial tumor</td>

  37
  
</tr>





  38
  
<td></td>

  38
  
<tr>





  39
  
</tr>

  39
  
<td>recur</td>





  40
  
<tr>

  40
  
<td>Number of recurrences</td>





  41
  
<td>recur</td>

  41
  
<td></td>





  42
  
<td>Number of recurrences</td>

  42
  
</tr>





  43
  
<td></td>

  43
  
<tr>





  44
  
</tr>

  44
  
<td>start</td>





  45
  
<tr>

  45
  
<td>Start time of each interval</td>





  46
  
<td>start</td>

  46
  
<td></td>





  47
  
<td>Start time of each interval</td>

  47
  
</tr>





  48
  
<td></td>

  48
  
<tr>





  49
  
</tr>

  49
  
<td>stop</td>





  50
  
<tr>

  50
  
<td>End time of each interval</td>





  51
  
<td>stop</td>

  51
  
<td></td>





  52
  
<td>End time of each interval</td>

  52
  
</tr>





  53
  
<td></td>

  53
  
<tr>





  54
  
</tr>

  54
  
<td>status</td>





  55
  
<tr>

  55
  
<td>End of interval code</td>





  56
  
<td>status</td>

  56
  
<td>0=censored, 1=recurrence, 2=death from bladder disease, 3=death other/unknown cause</td>





  57
  
<td>End of interval code</td>

  57
  
</tr>





  58
  
<td>0=censored, 1=recurrence, 2=death from bladder disease, 3=death other/unknown cause</td>

  58
  
<tr>





  59
  
</tr>

  59
  
<td>rtumor</td>





  60
  
<tr>

  60
  
<td>Number of tumors found at recurrence</td>





  61
  
<td>rtumor</td>

  61
  
<td></td>





  62
  
<td>Number of tumors found at recurrence</td>

  62
  
</tr>





  63
  
<td></td>

  63
  
<tr>





  64
  
</tr>

  64
  
<td>rsize</td>





  65
  
<tr>

  65
  
<td>Size of largest tumor at recurrence</td>





  66
  
<td>rsize</td>

  66
  
<td></td>





  67
  
<td>Size of largest tumor at recurrence</td>

  67
  
</tr>





  68
  
<td></td>

  68
  
<tr>





  69
  
</tr>

  69
  
<td>enum</td>





  70
  
<tr>

  70
  
<td>Event number (observation number within patient)</td>





  71
  
<td>enum</td>

  71
  
<td></td>





  72
  
<td>Event number (observation number within patient)</td>

  72
  
</tr>





  73
  
<td></td>

  73
  
<tr>





  74
  
</tr>

  74
  
<td><strong>Bladder Dataset 0</strong></td>





  75
  
<tr>

  75
  
<td></td>





  76
  
<td><strong>Bladder Dataset 0</strong></td>

  76
  
<td></td>





  77
  
<td></td>

  77
  
</tr>





  78
  
<td></td>

  78
  
<tr>





  79
  
</tr>

  79
  
<td>id</td>





  80
  
<tr>

  80
  
<td>Patient ID</td>





  81
  
<td>id</td>

  81
  
<td></td>





  82
  
<td>Patient ID</td>

  82
  
</tr>





  83
  
<td></td>

  83
  
<tr>





  84
  
</tr>

  84
  
<td>rx</td>





  85
  
<tr>

  85
  
<td>Treatment received</td>





  86
  
<td>rx</td>

  86
  
<td>1=placebo, 2=thiotepa</td>





  87
  
<td>Treatment received</td>

  87
  
</tr>





  88
  
<td>1=placebo, 2=thiotepa</td>

  88
  
<tr>





  89
  
</tr>

  89
  
<td>number</td>





  90
  
<tr>

  90
  
<td>Initial number of tumors</td>





  91
  
<td>number</td>

  91
  
<td>8=8 or more</td>





  92
  
<td>Initial number of tumors</td>

  92
  
</tr>





  93
  
<td>8=8 or more</td>

  93
  
<tr>





  94
  
</tr>

  94
  
<td>size</td>





  95
  
<tr>

  95
  
<td>Size (cm) of largest initial tumor</td>





  96
  
<td>size</td>

  96
  
<td></td>





  97
  
<td>Size (cm) of largest initial tumor</td>

  97
  
</tr>





  98
  
<td></td>

  98
  
<tr>





  99
  
</tr>

  99
  
<td>stop</td>





  100
  
<tr>

  100
  
<td>Recurrence or censoring time</td>





  101
  
<td>stop</td>

  101
  
<td></td>





  102
  
<td>Recurrence or censoring time</td>

  102
  
</tr>





  103
  
<td></td>

  103
  
<tr>





  104
  
</tr>

  104
  
<td>enum</td>





  105
  
<tr>

  105
  
<td>Which recurrence (up to 4)</td>





  106
  
<td>enum</td>

  106
  
<td></td>





  107
  
<td>Which recurrence (up to 4)</td>

  107
  
</tr>





  108
  
<td></td>

  108
  
<tr>





  109
  
</tr>

  109
  
<td><strong>Bladder Dataset 2</strong></td>





  110
  
<tr>

  110
  
<td></td>





  111
  
<td><strong>Bladder Dataset 2</strong></td>

  111
  
<td></td>





  112
  
<td></td>

  112
  
</tr>





  113
  
<td></td>

  113
  
<tr>





  114
  
</tr>

  114
  
<td>id</td>





  115
  
<tr>

  115
  
<td>Patient ID</td>





  116
  
<td>id</td>

  116
  
<td></td>





  117
  
<td>Patient ID</td>

  117
  
</tr>





  118
  
<td></td>

  118
  
<tr>





  119
  
</tr>

  119
  
<td>rx</td>





  120
  
<tr>

  120
  
<td>Treatment received</td>





  121
  
<td>rx</td>

  121
  
<td>1=placebo, 2=thiotepa</td>





  122
  
<td>Treatment received</td>

  122
  
</tr>





  123
  
<td>1=placebo, 2=thiotepa</td>

  123
  
<tr>





  124
  
</tr>

  124
  
<td>number</td>





  125
  
<tr>

  125
  
<td>Initial number of tumors</td>





  126
  
<td>number</td>

  126
  
<td>8=8 or more</td>





  127
  
<td>Initial number of tumors</td>

  127
  
</tr>





  128
  
<td>8=8 or more</td>

  128
  
<tr>





  129
  
</tr>

  129
  
<td>size</td>





  130
  
<tr>

  130
  
<td>Size (cm) of largest initial tumor</td>





  131
  
<td>size</td>

  131
  
<td></td>





  132
  
<td>Size (cm) of largest initial tumor</td>

  132
  
</tr>





  133
  
<td></td>

  133
  
<tr>





  134
  
</tr>

  134
  
<td>start</td>





  135
  
<tr>

  135
  
<td>Start of interval (0 or previous recurrence time)</td>





  136
  
<td>start</td>

  136
  
<td></td>





  137
  
<td>Start of interval (0 or previous recurrence time)</td>

  137
  
</tr>





  138
  
<td></td>

  138
  
<tr>





  139
  
</tr>

  139
  
<td>stop</td>





  140
  
<tr>

  140
  
<td>Recurrence or censoring time</td>





  141
  
<td>stop</td>

  141
  
<td></td>





  142
  
<td>Recurrence or censoring time</td>

  142
  
</tr>





  143
  
<td></td>

  143
  
<tr>





  144
  
</tr>

  144
  
<td>enum</td>





  145
  
<tr>

  145
  
<td>Which recurrence (up to 4)</td>





  146
  
<td>enum</td>

  146
  
<td></td>





  147
  
<td>Which recurrence (up to 4)</td>

  147
  
</tr>





  148
  
<td></td>

  148
  
</tbody>





  149
  
</tr>

  149
  
</table>





  150
  
</tbody>

  150
  





  151
  
</table>

  151
  
Bladder is the data set that appears most commonly in the literature. It uses only the 85 subjects with nonzero follow-up who were assigned to either thiotepa or placebo, and only the first four recurrences for any patient. The status variable is 1 for recurrence and 0 for everything else (including death for any reason). The data set is laid out in the competing risks format of the paper by Wei, Lin, and Weissfeld.





  152
  
<pre class="uc-code-block"><code>Bladder is <span class="hljs-keyword">the</span> data <span class="hljs-built_in">set</span> that appears most commonly <span class="hljs-keyword">in</span> <span class="hljs-keyword">the</span> literature. It uses only <span class="hljs-keyword">the</span> <span class="hljs-number">85</span> subjects <span class="hljs-keyword">with</span> nonzero follow-up who were assigned <span class="hljs-built_in">to</span> either thiotepa <span class="hljs-keyword">or</span> placebo, <span class="hljs-keyword">and</span> only <span class="hljs-keyword">the</span> <span class="hljs-keyword">first</span> <span class="hljs-literal">four</span> recurrences <span class="hljs-keyword">for</span> <span class="hljs-keyword">any</span> patient. The status <span class="hljs-built_in">variable</span> is <span class="hljs-number">1</span> <span class="hljs-keyword">for</span> recurrence <span class="hljs-keyword">and</span> <span class="hljs-number">0</span> <span class="hljs-keyword">for</span> everything <span class="hljs-keyword">else</span> (including death <span class="hljs-keyword">for</span> <span class="hljs-keyword">any</span> reason). The data <span class="hljs-built_in">set</span> is laid out <span class="hljs-keyword">in</span> <span class="hljs-keyword">the</span> competing risks <span class="hljs-built_in">format</span> <span class="hljs-keyword">of</span> <span class="hljs-keyword">the</span> paper <span class="hljs-keyword">by</span> Wei, Lin, <span class="hljs-keyword">and</span> Weissfeld.

  152
  
content_copy





  153
  
</code>

  153
  
Bladder1 is the full data set from the study. It contains all three treatment arms and all recurrences for 118 subjects; the maximum observed number of recurrences is 9.





  154
  

  154
  
Bladder2 uses the same subset of subjects as bladder, but formatted in the (start, stop] or Anderson-Gill style. Note that in transforming from the WLW to the AG style data set there is a quite common programming mistake that leads to extra follow-up time for 12 subjects: all those with follow-up beyond their 4th recurrence. This "follow-up" is a side effect of throwing away all events after the fourth while retaining the last follow-up time variable from the original data. The bladder2 data set found here does not make this mistake, but some analyses in the literature have done so; it results in the addition of a small amount of immortal time bias and shrinks the fitted coefficients towards zero.





  155
  
<pre class="uc-code-block"><code>Bladder1 is <span class="hljs-keyword">the</span> full data <span class="hljs-built_in">set</span> <span class="hljs-built_in">from</span> <span class="hljs-keyword">the</span> study. It <span class="hljs-keyword">contains</span> all <span class="hljs-literal">three</span> treatment arms <span class="hljs-keyword">and</span> all recurrences <span class="hljs-keyword">for</span> <span class="hljs-number">118</span> subjects; <span class="hljs-keyword">the</span> maximum observed <span class="hljs-built_in">number</span> <span class="hljs-keyword">of</span> recurrences is <span class="hljs-number">9.</span>

  
  





  156
  
</code>]

  
  





  157
  
<pre class="uc-code-block">

  
  





  158
  

  
  





  159
  
<code>Bladder2 uses <span class="hljs-keyword">the</span> same subset <span class="hljs-keyword">of</span> subjects <span class="hljs-keyword">as</span> bladder, but formatted <span class="hljs-keyword">in</span> <span class="hljs-keyword">the</span> (<span class="hljs-built_in">start</span>, <span class="hljs-built_in">stop</span>] <span class="hljs-keyword">or</span> Anderson-Gill style. Note that <span class="hljs-keyword">in</span> transforming <span class="hljs-built_in">from</span> <span class="hljs-keyword">the</span> WLW <span class="hljs-built_in">to</span> <span class="hljs-keyword">the</span> AG style data <span class="hljs-built_in">set</span> there is <span class="hljs-keyword">a</span> quite common programming mistake that leads <span class="hljs-built_in">to</span> extra follow-up <span class="hljs-built_in">time</span> <span class="hljs-keyword">for</span> <span class="hljs-number">12</span> subjects: all those <span class="hljs-keyword">with</span> follow-up beyond their <span class="hljs-number">4</span>th recurrence. This <span class="hljs-string">"follow-up"</span> is <span class="hljs-keyword">a</span> side effect <span class="hljs-keyword">of</span> throwing away all events <span class="hljs-keyword">after</span> <span class="hljs-keyword">the</span> <span class="hljs-keyword">fourth</span> <span class="hljs-keyword">while</span> retaining <span class="hljs-keyword">the</span> <span class="hljs-keyword">last</span> follow-up <span class="hljs-built_in">time</span> <span class="hljs-built_in">variable</span> <span class="hljs-built_in">from</span> <span class="hljs-keyword">the</span> original data. The bladder2 data <span class="hljs-built_in">set</span> found here does <span class="hljs-keyword">not</span> make this mistake, but some analyses <span class="hljs-keyword">in</span> <span class="hljs-keyword">the</span> literature have done so; <span class="hljs-keyword">it</span> results <span class="hljs-keyword">in</span> <span class="hljs-keyword">the</span> addition <span class="hljs-keyword">of</span> <span class="hljs-keyword">a</span> small amount <span class="hljs-keyword">of</span> immortal <span class="hljs-built_in">time</span> bias <span class="hljs-keyword">and</span> shrinks <span class="hljs-keyword">the</span> fitted coefficients towards <span class="hljs-literal">zero</span>.

  
  





  160
  
</code><div class="uc-code-block-copy-button-wrapper">