Conformability error in drmeta command

[Gilbert Lazarus]

Hi everyone, I am very new to STATA.

I am currently on a dose-response meta-analysis project. I tried to execute the drmeta command, but this error keeps coming back:

drmeta logrr dose, data(persondays cases) id(id) type(ntype) se(se) reml
          greenland_fn():  3200  conformability error
                 <istmt>:     -  function returned error
r(3200);

This is the data for my study

             id   ntype dose  cases       n        logrr    se     ntype   persondays |
      |-------------------------------------------------------------------------------------------|
  1. |       1      1   8.86   20.00    94.00    1.73   0.78       1             2872      |
  2. |       1      1   5.81    6.00   129.00    0.97   0.85       1             3769      |
  3. |       1      1   4.97    2.00   132.00    0.00      .          1             3790      |
  4. |       2      1   7.40   58.00  176.00    0.83   0.22       1             4928      |
  5. |       2      1   6.50    9.00    62.00     0.54   0.28       1             1516      |
  6. |       2      1   5.60    4.00   166.00    0.00      .          1             4521      |

Does anyone know which variable caused this error? Thank you very much, any help will be much appreciated

------
UPDATE:
This also happens when I tried to perform the glst command:

glst logrr doses, se(se) cov(n cases) cc
                       *:  3200  conformability error
                glsest():     -  function returned error
                 <istmt>:     -  function returned error
r(3200);

I have seen a previous discussion mentioning the same error (ref: https://www.statalist.org/forums/for...-meta-analysis); however, I have tried replacing the error as recommended before, but to no avail

[Gilbert Lazarus]

Hi, after a while, I decided to learn from the examples first.

I am using the dataset used by Shim et al, as follows:

. list author id type dose cases n logrr se doses1 doses2

     +------------------------------------------------------------------------------------------+
     |            author   id   type    dose   cases      n    logrr     se   doses1     doses2 |
     |------------------------------------------------------------------------------------------|
  1. |           Bianchi    1      1    9.06      61    261    -.223   0.22     9.06   .8263084 |
  2. |           Bianchi    1      1      27      69    228    .0001   0.23       27   12.67886 |
  3. |           Bianchi    1      1      45      22     44     .531   0.38       45   28.97389 |
  4. |           Bianchi    1      1    64.8      19     34     .875   0.44     64.8    46.9127 |
  5. |           Bianchi    1      1       0     126    414        0      .        0          0 |
     |------------------------------------------------------------------------------------------|
  6. |             Bobak    2      1   16.05      88    413    -.431   0.22    16.05   4.050243 |
  7. |             Bobak    2      1   46.42      24    202   -1.079   0.30    46.42   30.26038 |
  8. |             Bobak    2      1   77.16      13     64    -.616   0.39    77.16   58.11095 |
  9. |             Bobak    2      1       0      77    258        0      .        0          0 |
 10. |    Malarcher-wine    3      1    1.18      46    175     -.58   0.32     1.18   .0018256 |
     |------------------------------------------------------------------------------------------|
 11. |    Malarcher-wine    3      1    8.92      17     58    -.562   0.42     8.92   .7885914 |
 12. |    Malarcher-wine    3      1   18.72       4     11     .615   0.91    18.72   5.854179 |
 13. |    Malarcher-wine    3      1       0      83    208        0      .        0          0 |
 14. |    Malarcher-beer    4      1     .95      29    117    -.288   0.33      .95   .0009526 |
 15. |    Malarcher-beer    4      1    7.43      32     81     .513   0.34     7.43   .4557471 |
     |------------------------------------------------------------------------------------------|
 16. |    Malarcher-beer    4      1    15.6      18     39    -.315   0.47     15.6   3.772958 |
 17. |    Malarcher-beer    4      1       0      83    208        0      .        0          0 |
 18. | Vliegenthart-wine    5      2       6     229    975    -.416   0.14        6        .24 |
 19. | Vliegenthart-wine    5      2      18      38    207    -.635   0.24       18   5.342597 |
 20. | Vliegenthart-wine    5      2    28.8      39    133    -.083   0.27     28.8   14.29751 |
     |------------------------------------------------------------------------------------------|
 21. | Vliegenthart-wine    5      2       0     159    480        0      .        0          0 |
 22. | Vliegenthart-beer    6      2    6.25     129    436    -.223   0.16     6.25   .2712674 |
 23. | Vliegenthart-beer    6      2   18.75      19     49     -.01   0.35    18.75   5.875869 |
 24. | Vliegenthart-beer    6      2      30      15     41    -.261   0.42       30   15.38388 |
 25. | Vliegenthart-beer    6      2       0     302   1269        0      .        0          0 |
     +------------------------------------------------------------------------------------------+

I have defined the knots as per recommended, by using the command as follows:

mkspline2 doses = dose, nk(3) cubic displayknots

             |     knot1      knot2      knot3 
-------------+---------------------------------
        dose |         0       9.06         30 

mat knots = r(knots)

However, when I tried to run the drmeta command, the command returns conformability error

drmeta logrr doses1 doses2 , data(n cases) id(id) type(type) se(se) reml 2stage
          greenland_fn():  3200  conformability error
                 <istmt>:     -  function returned error
r(3200);

Does anyone know which variable caused the error? I tried to replace the zero value non-referent logrr (ref: https://www.statalist.org/forums/for...-meta-analysis); however, however, the error still exists.

Thank you very much for the help. Any help will be much appreciated.

Best,

[Nerea Becerra]

Hi Gilbert,
the conformability error is because only the first category (reference) can have a value of 0 in the variable logrr. If you look at the study "Vliegenthart-wine" you have more than one category with logrr =0. This could be solved by adding a constant to the non-reference categories such as 0.000000001.

I hope this helps.

Nerea

[Tom Hsiung]

Hello, Gilbert

I have been looking for how to make a dose-response meta-analysis for days and I found your example useful. However, I hope to understand the drmeta procedure more deeply. Could you please tell what these parameters in the drmeta syntax mean? Thank you very much.

 
 logrr doses1 doses2 , data(n cases) id(id) type(type) se(se) reml 2stage