Hi there,
I am doing a meta-analysis on 320 patients with data from different centres (13 in total). I have different outcomes I am interested in: intracerebral haemorrhage within 5 years, ischaemic stroke within 5 years and cortical subarachnoid haemorrhage within 5 years. I am going to use ischaemic stroke on follow-up as example.
All codes are used in stata 15.1, last updated 27th June 2018.
Output:
Out of 8 centers with an event on follow-up only 4 are included:
If I then repeat it again I get a different result:
Now it includes another center and more patients but gives an error message, although I haven't change anything.
If I used the "nooverall" option 4/8 are included.
One problem could be that the information about ischaemic stroke on follow-up is missing in cases in which information about probable CAA is available. Nevertheless, we have information on ischaemic stroke in 8 centres. In some cases where its not known if anyone did NOT have an ischaemic stroke these centers are included into the forrest plot but others that we have the information on who had and who didnt have an ischaemic stroke were not (like for example Adelaide is included, whereas Lisbon and Emory are not included into the forrest plot). Another problem could be that we do know that they had an ischaemic stroke but we dont have the follow-up time, typical for a retrospective study, this is only the case in one patient though.
So Im trying to solve following issues:
I am doing a meta-analysis on 320 patients with data from different centres (13 in total). I have different outcomes I am interested in: intracerebral haemorrhage within 5 years, ischaemic stroke within 5 years and cortical subarachnoid haemorrhage within 5 years. I am going to use ischaemic stroke on follow-up as example.
All codes are used in stata 15.1, last updated 27th June 2018.
Code:
stset TimetoIS_or_FU_years_5Lim, failure(IS_FU_5y) ipdmetan, study(Center) hr : stcox Age probableCAA i.Drinker HTN OAC
Code:
stset TimetoIS_or_FU_years_5Lim, failure(IS_FU_5y)
failure event: IS_FU_5y != 0 & IS_FU_5y < .
obs. time interval: (0, TimetoIS_or_FU_years_5Lim]
exit on or before: failure
------------------------------------------------------------------------------
320 total observations
40 event time missing (TimetoIS_or_FU_years_5Lim>=.) PROBABLE ERROR
1 observation ends on or before enter()
------------------------------------------------------------------------------
279 observations remaining, representing
23 failures in single-record/single-failure data
499.948 total analysis time at risk and under observation
at risk from t = 0
earliest observed entry t = 0
last observed exit t = 5.060274
. ipdmetan, study(Center) hr : stcox Age probableCAA i.Drinker HTN OAC
Studies included: 4
Patients included: 109
Meta-analysis pooling of main (treatment) effect estimate Age
using Fixed-effects
---------------------------------------------------------------------
| Haz.
Center | Ratio [95% Conf. Interval] % Weight
-----------------------+---------------------------------------------
Klinikum Graz | 0.960 0.879 1.048 74.18
Mayo Clinic | 1.083 0.811 1.448 6.84
Royal Adelaide Hospital| 0.984 0.758 1.277 8.47
Argentina | 1.036 0.820 1.309 10.51
-----------------------+---------------------------------------------
Overall effect | 0.978 0.906 1.055 100.00
---------------------------------------------------------------------
Test of overall effect = 1: z = -0.581 p = 0.561
Heterogeneity Measures
-----------------------------------------------------------
| Value df p-value
-----------------------+-----------------------------------
Cochran Q | 0.88 3 0.830
I² (%) | 0.0%
Modified H² | 0.000
tau² | 0.0000
-----------------------------------------------------------
I² = between-study variance (tau²) as a percentage of total variance
Modified H² = ratio of tau² to typical within-study variance
Out of 8 centers with an event on follow-up only 4 are included:
Code:
table probableCAA ICH_FU_5y if TimetoIS_or_FU_years_5Lim!=., by (Center) miss
----------------------------------------
| ICH_FU_5y
Center and probableCAA | 0 1
----------------------------+-----------
UCLH |
0 | 2 .
1 | 17 2
----------------------------+-----------
CHU Caen |
0 | 2 .
1 | 10 1
----------------------------+-----------
Klinikum Graz |
0 | 10 3
1 | 10 9
----------------------------+-----------
Lisbon |
0 | 1 1
1 | 2 1
----------------------------+-----------
Mayo Clinic |
0 | 7 .
1 | 10 10
----------------------------+-----------
CHU Montpellier |
0 | 2 .
1 | 5 4
----------------------------+-----------
Sant Pau Barcelona |
0 | 13 .
1 | 3 5
----------------------------+-----------
Royal Adelaide Hospital |
0 | 15 .
1 | 9 5
----------------------------+-----------
Argentina |
0 | 9 1
1 | 10 1
----------------------------+-----------
Beth Israel Harvard |
0 | 4 .
1 | 7 .
----------------------------+-----------
Monash University Australia |
0 | 2 .
1 | 11 12
----------------------------+-----------
Emory Clinic |
0 | 7 1
1 | 2 1
----------------------------------------
Code:
stset TimetoIS_or_FU_years_5Lim, failure(IS_FU_5y)
failure event: IS_FU_5y != 0 & IS_FU_5y < .
obs. time interval: (0, TimetoIS_or_FU_years_5Lim]
exit on or before: failure
------------------------------------------------------------------------------
320 total observations
40 event time missing (TimetoIS_or_FU_years_5Lim>=.) PROBABLE ERROR
1 observation ends on or before enter()
------------------------------------------------------------------------------
279 observations remaining, representing
23 failures in single-record/single-failure data
499.948 total analysis time at risk and under observation
at risk from t = 0
earliest observed entry t = 0
last observed exit t = 5.060274
. ipdmetan, study(Center) hr : stcox Age probableCAA i.Drinker HTN OAC
Studies included: 5
Patients included: 121
Meta-analysis pooling of main (treatment) effect estimate Age
using Fixed-effects
---------------------------------------------------------------------
| Haz.
Center | Ratio [95% Conf. Interval] % Weight
-----------------------+---------------------------------------------
CHU Caen | 2160.292 0.000 . 0.00
Klinikum Graz | 0.960 0.879 1.048 74.18
Mayo Clinic | 1.083 0.811 1.448 6.84
Royal Adelaide Hospital| 0.984 0.758 1.277 8.47
Argentina | 1.036 0.820 1.309 10.51
-----------------------+---------------------------------------------
Overall effect | 0.978 0.906 1.055 100.00
---------------------------------------------------------------------
Test of overall effect = 1: z = -0.581 p = 0.561
Heterogeneity Measures
-----------------------------------------------------------
| Value df p-value
-----------------------+-----------------------------------
Cochran Q | 0.88 4 0.927
I² (%) | 0.0%
Modified H² | 0.000
tau² | 0.0000
-----------------------------------------------------------
I² = between-study variance (tau²) as a percentage of total variance
Modified H² = ratio of tau² to typical within-study variance
macro substitution results in line that is too long
The line resulting from substituting macros would be longer than allowed. The maximum allowed length is 645,216
characters, which is calculated on the basis of set maxvar.
You can change that in Stata/SE and Stata/MP. What follows is relevant only if you are using Stata/SE or
Stata/MP.
The maximum line length is defined as 16 more than the maximum macro length, which is currently 645,200
characters. Each unit increase in set maxvar increases the length maximums by 129. The maximum value of set
maxvar is 32,767. Thus, the maximum line length may be set up to 4,227,159 characters if you set maxvar to its
largest value.
r(920);
If I used the "nooverall" option 4/8 are included.
Code:
ipdmetan, study(Center) hr nooverall: stcox Age probableCAA i.Drinker HTN OAC
Studies included: 4
Patients included: 109
Meta-analysis pooling of main (treatment) effect estimate Age
using Fixed-effects
---------------------------------------------------------------------
| Haz.
Center | Ratio [95% Conf. Interval] % Weight
-----------------------+---------------------------------------------
Klinikum Graz | 0.960 0.879 1.048 74.18
Mayo Clinic | 1.083 0.811 1.448 6.84
Royal Adelaide Hospital| 0.984 0.758 1.277 8.47
Argentina | 1.036 0.820 1.309 10.51
---------------------------------------------------------------------
One problem could be that the information about ischaemic stroke on follow-up is missing in cases in which information about probable CAA is available. Nevertheless, we have information on ischaemic stroke in 8 centres. In some cases where its not known if anyone did NOT have an ischaemic stroke these centers are included into the forrest plot but others that we have the information on who had and who didnt have an ischaemic stroke were not (like for example Adelaide is included, whereas Lisbon and Emory are not included into the forrest plot). Another problem could be that we do know that they had an ischaemic stroke but we dont have the follow-up time, typical for a retrospective study, this is only the case in one patient though.
So Im trying to solve following issues:
- Why do I get for the same input two different outputs with a different amount of patients?
- Why are only 109 or 121 included out of 279 patients (missing values in some of the variables?)
- Why are not all centres with an event included?
- What I am doing wrong and how can I correct it?
Comment