1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
                                  META-SPSS

DISCLAIMER
----------

META-SPSS is provided "as is" without warranty of any kind. The entire risk
as to the quality, performance, and fitness for intended purpose is with you.
You assume responsibility for the selection of the program and for the use of
results obtained from that program.


DESCRIPTION
-----------

This collection of files adds meta-analytic facilities to SPSS:

1. Meta-analysis of P values (see 'MetaPValues-Info.txt' for technical details).

2. Meta-analysis of binary outcomes: RD, OR and RR, with both fixed & random
   effects (DerSimonian-Laird) models. Raw data (counts) or summary data
   (adjusted OR, RR or RD) can be used as input. Different fixed effects models:
   inverse variance (Woolf method), Mantel-Haenszel and Peto (this last only
   for OR) are available for raw data (counts); for summary data only the first.

3. Meta-analysis of continuous outcomes: unstandardised & standardised (Hedges,
   Cohen & Glass) mean differences, with both fixed (inverse variance) & random
   effects (DerSimonian-Laird) methods. Means, SD & N or Cohen's d can be used
   for input (except for Glass method, which requires means, SD & N).
   Orwin's Fail Safe N is computed for significant results. 

4. Meta-analysis of correlation coefficients: Hedges-Olkin fixed & random
   effects models (inverse variance & Dersimonian-Laird).
   Orwin's Fail Safe N is computed for significant results.

5. Meta-analysis of correlation coefficients: Schmidt-Hunter method (simplified,
   unreliability or range departure correction methods are not included). See
   Larry Lyons paper & website (details in folder "Documents") for details.

For methods 1 to 4, Higgins' I2 with its 95%CI is calculated
(see 'Higgins I2 formula.txt' for details and a reference).

For methods 2, 3 & 4, forest plots are also included (see below for a limitation). 

You must know meta-analysis before using them, although some short documents with
meta-analysis theory & formulae are included in the folder "Documents". I recommend
reading Andy Field's papers (Meta1.pdf, Metar.pdf & Metad.pdf)and taking a look at
his website. A text file with useful references (mainly from British Medical Journal)
and tutorials is also included.

A very good book on the topic is 'Systematic Reviews in Health Care' Edited by
Mathias Egger, George Davey Smith & Douglas G Altman. BMJ Books, 2nd Ed (2001).
(ISBN: 0-7279-1488-X).

Feel free to edit & modify every file. If you improve the code, I would like to
have a copy.

Questions can be sent to me: Marta Garcia-Granero (biostatistics@terra.es).


SPSS files included:
-------------------

- MetaDatasets.SPS: creates in drive A: a collection of datasets from published
  meta-analysis, mainly from British Medical Journal. Must be run once to obtain
  the datasets that MetaTests.SPS requires.

- MetaAnalysisOfxxxxxxx.SPS: Syntax files for meta-analytic methods 1 to 5.
  The name, a bit long, it's related to the method:

     Method           File
   ----------------------------------------------------------
   Odds Ratio:
     Peto           MetaAnalysisOfBinOutcomes(OR-Peto).SPS
     M-H            MetaAnalysisOfBinOutcomes(OR-MH).SPS
     Inv Var        MetaAnalysisOfBinOutcomes(OR-IV).SPS
     DerS-Laird     MetaAnalysisOfBinOutcomes(OR-DL).SPS
     Summary data   MetaAnalysisOfBinOutcomes(SumOR-IV).SPS
                    MetaAnalysisOfBinOutcomes(SumOR-DL).SPS

   Risk Ratio:
     M-H            MetaAnalysisOfBinOutcomes(RR-MH).SPS
     Inv Var        MetaAnalysisOfBinOutcomes(RR-IV).SPS
     DerS-Laird     MetaAnalysisOfBinOutcomes(RR-DL).SPS
     Summary data   MetaAnalysisOfBinOutcomes(SumRR-IV).SPS
                    MetaAnalysisOfBinOutcomes(SumRR-DL).SPS

   Risk Diff:
     M-H            MetaAnalysisOfBinOutcomes(RD-MH).SPS
     Inv Var        MetaAnalysisOfBinOutcomes(RD-IV).SPS
     DerS-Laird     MetaAnalysisOfBinOutcomes(RD-DL).SPS
     Summary data   MetaAnalysisOfBinOutcomes(SumRD-IV).SPS
                    MetaAnalysisOfBinOutcomes(SumRD-DL).SPS

  Mean Difference:
     Inv Var        MetaAnalysisOfContOutcomes(WM-IV).SPS
     DerS-Laird     MetaAnalysisOfContOutcomes(WM-DL).SPS

  Hedges G (Stand. Mean Difference):
     Inv Var        MetaAnalysisOfContOutcomes(G-IV).SPS
     DerS-Laird     MetaAnalysisOfContOutcomes(G-DL).SPS
     Summary data   MetaAnalysisOfContOutcomes(SumG-IV).SPS
                    MetaAnalysisOfContOutcomes(SumG-DL).SPS

  Cohen d (Stand. Mean Difference):
     Inv Var        MetaAnalysisOfContOutcomes(D-IV).SPS
     DerS-Laird     MetaAnalysisOfContOutcomes(D-DL).SPS
     Summary data   MetaAnalysisOfContOutcomes(SumD-IV).SPS
                    MetaAnalysisOfContOutcomes(SumD-DL).SPS

  Glass delta (Stand. Mean Difference):
     Inv Var        MetaAnalysisOfContOutcomes(Delta-IV).SPS
     DerS-Laird     MetaAnalysisOfContOutcomes(Delta-DL).SPS

  Correlation (Hedges-Olkin):
     Inv Var        MetaAnalysisOfCorrCoef(HO-IV).SPS
     DerS-Laird     MetaAnalysisOfCorrCoef(HO-DL).SPS

  Correlation (Schmidt-Hunter):
                    MetaAnalysisOfCorrCoef(SH).SPS

  P Values:
                    MetaAnalysisOfPvalues.SPS
  ----------------------------------------------------------

- ForOR.SPS:  For binary outcomes the information included assumes that RR
  and RD are the summary data of interest. This syntax has to be run after
  loading binary outcomes files if OR is the summary measure of interest.
  ForOR.SPS creates 4 extra variables, called: caex, canex, coex & conex.

- MetaTests.SPS: a syntax file that shows the use of the different method
  files.

- XXXXXX.sct: several chart templates that work with SPSS 11.5 or older.

- Two SAV files in "Temp" folder: they are created by files of methods 2-4.
  They were originally included in the ZIP file to be sure that the "Temp"
  folder is created during installation, because several syntax files store
  information in Temp and it must exist to avoid errors.


INSTALLATION
------------

- Download META-SPSS.ZIP and unzip its contents to a floppy disk (drive A:).
- Run MetaDatasets.SPS once to create the 12 example datasets.
- Open MetaTests.SPS and run it to see how the files work.


ACCURACY OF THE METHODS
-----------------------

I have tried to check the output with published datasets and other programs
(Review Manager, Meta, Easy Meta Analysis, Combine...) and, to my knowledge,
everything is OK.

An extra check was performed comparing META-SPSS output with tables 15.4 & 15.6
of 'Systematic Reviews in Health Care' (p 303-306), using the same datasets.
If you want to check that results with META-SPSS are correct, the datasets are
included as 'MetaDataset(XII).sav' (binary outcomes) & 'MetaDataset(XIII).sav'
(continuous outcomes).

                           Table 15.4
 Method         Estimate (95% CI)        Sig of effect(*)  Heterogeneity
 ---------------------------------------------------------------------------
 Odds Ratio:
   Peto         0.59 (0.51 to 0.69)      z=7.05 (p<0.001)  X2=38.5 (p<0.001)
   M-H          0.59 (0.51 to 0.69)      z=6.98 (p<0.001)  X2=38.9 (p<0.001)
   Inv Var      0.60 (0.52 to 0.70)      z=6.70 (p<0.001)  X2=38.8 (p<0.001)
   DerS-Laird   0.54 (0.34 to 0.85)      z=2.64 (p=0.008)

 Risk Ratio:
   M-H          0.79 (0.74 to 0.85)      z=6.95 (p<0.001)  X2=29.8 (p<0.001)
   Inv Var      0.80 (0.75 to 0.85)      z=7.14 (p<0.001)  X2=29.7 (p<0.001)
   DerS-Laird   0.77 (0.64 to 0.92)      z=2.93 (p=0.003)

 Risk Diff:
   M-H         -0.117 (-0.149 to -0.085) z=7.13 (p<0.001)  X2=33.1 (p<0.001)
   Inv Var     -0.127 (-0.158 to -0.095) z=7.86 (p<0.001)  X2=32.7 (p<0.001)
   DerS-Laird  -0.124 (-0.211 to -0.038) z=2.81 (p=0.005)
 ---------------------------------------------------------------------------

                           Table 15.6
 Method         Estimate (95% CI)        Sig of effect(*)  Heterogeneity
 ---------------------------------------------------------------------------
 Fixed effect models:
   Cohen's d   -0.16 (-0.41 to 0.08)     z=1.29 (p=0.20)   X2=2.34 (p=0.31)
   Hedges' g   -0.16 (-0.41 to 0.08)     z=1.29 (p=0.20)   X2=2.31 (p=0.32)
   Glass's d   -0.16 (-0.40 to 0.09)     z=1.24 (p=0.21)   X2=2.28 (p=0.32)

 Random effects models:
   Cohen's d   -0.15 (-0.42 to 0.12)     z=1.12 (p=0.26)
   Hedges' g   -0.15 (-0.42 to 0.11)     z=1.12 (p=0.26)
   Glass's d   -0.15 (-0.42 to 0.12)     z=1.10 (p=0.27)
---------------------------------------------------------------------------
 (*) META-SPSS gives a X2 test, mathematically equivalent (X2=(Z)2).
---------------------------------------------------------------------------

LIMITATIONS & KNOWN BUGS
------------------------

Cumulative meta-analysis is not possible... yet (I'm working on that). Check
this website from time to time to see if I have updated META-SPSS.

The syntax files assume that everything runs from a floppy disk (A:). You will
have to edit every 'Meta...SPS' file (including MetaTests.SPS & MetaDatasets.SPS)
and replace every "A:" by a suitable drive&path if you want to run them from
other location.

This code was developed before SPSS 12 was released, and the chart templates
included will only work with SPSS 11.5 or older. The graphs will need formatting
if SPSS 12 is used (good luck, I never succeeded trying to apply chart templates
in v12, and some options, available in older versions, have dissapeared).