* Survival 11 years test of standard errors.sps.
* Produce file of survival times for teeth filled in period.
* Adjusted for time dependency.
* Test standard errors by using random censoring as alternative to weighted adjustment.
* Written by PSKL on 24/09/02.

* Work from file of observed restoration lives.

get file='D:\Longitudinal Data\Observed Restoration Lives 1991 to 2001.sav'.

* Restrict to restorations placed before the end of the observation period 31/12/2001.

select if (doplacn<yrmoda(2001,12,31)-yrmoda(1899,12,31)).

* Check whether time is in acceptable range, and set reint flag.

compute reint=0.
if not (missing(time)) reint=1.
if (time le 0) reint=0.
if (doplacn+time>yrmoda(2001,12,31)-yrmoda(1899,12,31)) reint=0.

execute.

* Create weights to adjust for time dependent censoring.

* Double up the cases with no re-intervention and weight by probability of re-intervention.
compute caseid=$casenum.
execute.

* Look up probability of reattendance after last visit.
sort cases by interval.
match files file=*
 /table='D:\Longitudinal Data\reattendance probability.sav'
 /by interval.
compute recid=1.
sort cases by caseid.
save outfile='d:\temp1.sav'.

select if (reint=0).
compute recid=2.
sort cases by caseid.
save outfile='d:\temp2.sav'.

add files file='d:\temp1.sav'
 /file='d:\temp2.sav'
 /by caseid .

Compute weight = 1000.
Do if (reint=0).
if (recid=1) weight = 1000-trunc(1000*prreatt+0.5).
if (recid=2) weight = trunc(1000*prreatt+0.5).
end if.

save outfile='d:\temp3.sav'.


get file='d:\temp3.sav'.

***************************************************.
* Compare and calculate survival function and statistics, and standard errors.


* Assume censored cases are lost at one year (365 days) after last visit.

* recalculate time if reint=0.
* Impute effective time of censoring.
Do if (reint=0).
If (recid=1) time=min(lastvn+365,yrmoda(2001,12,31)-yrmoda(1899,12,31))- doplacn.
if (recid=2) time=yrmoda(2001,12,31)-yrmoda(1899,12,31)- doplacn.
end if.

save outfile= 'd:\temp4.sav'.
get file='d:\temp4.sav'.

*  Test 1 Adjusted KM, but increased weights give incorrect confidence intervals.

weight by weight.

KM
  time  /STATUS=reint(1)
    /PRINT MEAN
    /PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
    /Save survival (survadj).

Kaplan-Meier

Notes
Output Created 19-NOV-2002 12:33:21
Comments
Input Data d:\temp4.sav
Filter <none>
Weight WEIGHT
Split File <none>
N of Rows in Working Data File 843599
Syntax KM
time /STATUS=reint(1)
/PRINT MEAN
/PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
/Save survival (survadj).
Resources Elapsed Time 0:01:36.44 

 Survival Analysis for TIME




          Survival Time    Standard Error   95% Confidence Interval

 Mean:      2537.38                 .09     (  2537.21,   2537.55 )
 (Limited to   4016.0 )
 Median:    3003.00                 .45     (  3002.12,   3003.88 )



                                                            Percentiles

                     45.00        50.00        55.00        60.00        65.00        70.00        75.00        80.00        85.00

 Value             3761.00      3003.00      2430.00      1964.00      1562.00      1230.00       949.00       709.00       500.00
 Standard Error        .            .45          .31          .24          .19          .14          .11          .09          .06

                                                            Percentiles

                     90.00        95.00

 Value              321.00       168.00
 Standard Error        .05          .


>Warning # 3211
>On at least one case, the value of the weight variable was zero, negative,
>or missing.  Such cases are invisible to statistical procedures and graphs
>which need positively weighted cases, but remain on the file and are
>processed by non-statistical facilities such as LIST and SAVE.


* Test 2 Unadjusted KM, for confidence intervals.

weight off.

* Censor all non-returners.
select if (recid=1).

KM
  time  /STATUS=reint(1)
   /PRINT MEAN
   /PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
   /Save survival (survunad).

Kaplan-Meier

Notes
Output Created 19-NOV-2002 12:35:01
Comments
Input Data d:\temp4.sav
Filter <none>
Weight <none>
Split File <none>
N of Rows in Working Data File 503965
Syntax KM
time /STATUS=reint(1)
/PRINT MEAN
/PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
/Save survival (survunad).
Resources Elapsed Time 0:01:30.12 

 Survival Analysis for TIME




          Survival Time    Standard Error   95% Confidence Interval

 Mean:      2483.62                2.83     (  2478.07,   2489.17 )
 (Limited to   4016.0 )
 Median:    2766.00               11.51     (  2743.43,   2788.57 )



                                                            Percentiles

                     45.00        50.00        55.00        60.00        65.00        70.00        75.00        80.00        85.00

 Value             3362.00      2766.00      2276.00      1857.00      1495.00      1188.00       924.00       696.00       496.00
 Standard Error        .          11.51         8.73         6.88         5.45         4.29         3.36         2.62         1.99

                                                            Percentiles

                     90.00        95.00

 Value              321.00       168.00
 Standard Error       1.40          .


* Test 3 Censoring using randomisation instead of weighting.

* Censor the proportion prreatt of all non-returners at the end of the observation period (instead of earlier).

Set seed = 10102002.
compute random=uniform(1).
if (reint=0 and random<prreatt) time=yrmoda(2001,12,31)-yrmoda(1899,12,31)- doplacn.

KM
  time  /STATUS=reint(1)
   /PRINT MEAN
   /PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
   /Save survival (survrand).

Kaplan-Meier

Notes
Output Created 19-NOV-2002 12:36:32
Comments
Input Data d:\temp4.sav
Filter <none>
Weight <none>
Split File <none>
N of Rows in Working Data File 503965
Syntax KM
time /STATUS=reint(1)
/PRINT MEAN
/PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95)
/Save survival (survrand).
Resources Elapsed Time 0:02:31.14 

 Survival Analysis for TIME




          Survival Time    Standard Error   95% Confidence Interval

 Mean:      2537.29                2.77     (  2531.87,   2542.71 )
 (Limited to   4016.0 )
 Median:    3001.00               14.27     (  2973.02,   3028.98 )



                                                            Percentiles

                     45.00        50.00        55.00        60.00        65.00        70.00        75.00        80.00        85.00

 Value             3761.00      3001.00      2429.00      1964.00      1562.00      1230.00       949.00       709.00       500.00
 Standard Error        .          14.27         9.89         7.57         5.91         4.55         3.53         2.72         2.04

                                                            Percentiles

                     90.00        95.00

 Value              321.00       168.00
 Standard Error       1.42          .


save outfile='D:\temp5.sav'.

* Rerun to test robustness.

get file='d:\temp4.sav'.
select if (recid=1).
compute random=uniform(1).
if (reint=0 and random<prreatt) time=yrmoda(2001,12,31)-yrmoda(1899,12,31)- doplacn.

KM
  time  /STATUS=reint(1)
   /PRINT MEAN
   /PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95).

Kaplan-Meier

Notes
Output Created 19-NOV-2002 12:40:16
Comments
Input Data d:\temp4.sav
Filter <none>
Weight <none>
Split File <none>
N of Rows in Working Data File 503965
Syntax KM
time /STATUS=reint(1)
/PRINT MEAN
/PERCENTILES = (45,50,55,60,65,70,75,80,85,90,95).
Resources Elapsed Time 0:00:40.91 

 Survival Analysis for TIME




          Survival Time    Standard Error   95% Confidence Interval

 Mean:      2537.56                2.76     (  2532.14,   2542.98 )
 (Limited to   4016.0 )
 Median:    3003.00               14.22     (  2975.13,   3030.87 )



                                                            Percentiles

                     45.00        50.00        55.00        60.00        65.00        70.00        75.00        80.00        85.00

 Value             3758.00      3003.00      2431.00      1965.00      1564.00      1231.00       949.00       709.00       500.00
 Standard Error        .          14.22         9.90         7.58         5.92         4.56         3.53         2.72         2.04

                                                            Percentiles

                     90.00        95.00

 Value              321.00       168.00
 Standard Error       1.42          .


**************************************************************************************.

* Create an Excel file for comparing survivor functions.
get file='d:\temp5.sav'.

weight off.

select if (reint=1).
sort cases by time.
aggregate outfile=*
 /presorted
 /break time
 / survadj survunad survrand
= first( survadj survunad survrand ).

SAVE TRANSLATE OUTFILE='I:\Research Projects\longevity\1991 to 2001 adj unadj and rand.xls'
  /TYPE=XLS /MAP /REPLACE /FIELDNAMES.

Data written to I:\Research Projects\longevity\1991 to 2001 adj unadj and rand.xls.
4 variables and 3783 cases written to range: SPSS.
Variable: TIME       Type: Number   Width:  8   Dec: 2
Variable: SURVADJ    Type: Number   Width: 10   Dec: 5
Variable: SURVUNAD   Type: Number   Width: 10   Dec: 5
Variable: SURVRAND   Type: Number   Width: 10   Dec: 5



* Create an Excel file for calculation of standard errors.

get file='d:\temp4.sav'.

weight by weight.

sort cases by time reint.

save outfile = 'd:\temp6.sav'.
get file='d:\temp6.sav'.

compute censored=0.
if (reint=0) censored=1.

aggregate outfile=*
 /presorted
 /break time
 /reints censored =sum(reint censored ).

>Warning # 3211
>On at least one case, the value of the weight variable was zero, negative,
>or missing.  Such cases are invisible to statistical procedures and graphs
>which need positively weighted cases, but remain on the file and are
>processed by non-statistical facilities such as LIST and SAVE.


compute reints=reints/1000.
compute censored = censored/1000.


SAVE TRANSLATE OUTFILE='I:\Research Projects\longevity\1991 to 2001 times for se.xls'
  /TYPE=XLS /MAP /REPLACE /FIELDNAMES.

Data written to I:\Research Projects\longevity\1991 to 2001 times for se.xls.
3 variables and 4012 cases written to range: SPSS.
Variable: TIME       Type: Number   Width:  8   Dec: 2
Variable: REINTS     Type: Number   Width:  8   Dec: 2
Variable: CENSORED   Type: Number   Width:  8   Dec: 2