Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Incidence rate ratios from Poisson regression including survey weights

    Dear all

    Still working on my Stata skills and very grateful for this forum!

    I working on estimating the diabetes incidence rate (cases per person-years) over four periods using Poisson regression models with log person-years as offset and the follow-up person cut into 1-year age bands. The four periods; 2001-2005; 2005-2009; 2009-2013; 2013-2017. It a open cohort with both left and right censoring.
    I have set up the data using stset and stsplit for followup time, age and calendar year:

    Code:
    svyset ER31997 [pweight = SW], singleunit(scaled) strata(ER31996)
    stset FUT_overall, id(id) failure(DM==1)
    stsplit splitime, every(1)
    gen FUT_split = _t - _t0
    gen age_split = Age + _t0
    gen logtimerisk = ln(FUT_split)
    gen year_split = year +_t0

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
    clear
    input float id int year byte sex float ER31996 byte ER31997 float(Age DM FUT_overall SW) byte splitime float(FUT_split age_split logtimerisk year_split) byte(_st _d) double _t byte _t0
     1 2005 1 31 2  49.32494 .         1 17082.857 0          1  49.32494           0 2005 1 0                  1 0
     1 2005 1 31 2  49.32494 .         2 17082.857 1          1  50.32494           0 2006 1 0                  2 1
     1 2005 1 31 2  49.32494 .         3 17082.857 2          1  51.32494           0 2007 1 0                  3 2
     1 2005 1 31 2  49.32494 1  3.057648 17082.857 3   .0576477  52.32494   -2.853405 2008 1 1  3.057647705078125 3
     2 2013 1 31 2  57.35999 .         1 17082.857 0          1  57.35999           0 2013 1 0                  1 0
     2 2013 1 31 2  57.35999 0  1.971794 17082.857 1   .9717941  58.35999   -.0286113 2014 1 0 1.9717941284179688 1
     3 2005 2 31 2  47.84883 .         1 21386.875 0          1  47.84883           0 2005 1 0                  1 0
     3 2005 2 31 2  47.84883 .         2 21386.875 1          1  48.84883           0 2006 1 0                  2 1
     3 2005 2 31 2  47.84883 .         3 21386.875 2          1  49.84883           0 2007 1 0                  3 2
     3 2005 2 31 2  47.84883 0 3.7354546 21386.875 3   .7354546  50.84883   -.3072665 2008 1 0  3.735454559326172 3
     4 2009 2 31 2  51.58428 .         1 21386.875 0          1  51.58428           0 2009 1 0                  1 0
     4 2009 2 31 2  51.58428 .         2 21386.875 1          1  52.58428           0 2010 1 0                  2 1
     4 2009 2 31 2  51.58428 .         3 21386.875 2          1  53.58428           0 2011 1 0                  3 2
     4 2009 2 31 2  51.58428 .         4 21386.875 3          1  54.58428           0 2012 1 0                  4 3
     4 2009 2 31 2  51.58428 0  4.012047 21386.875 4 .012046814  55.58428   -4.418955 2013 1 0  4.012046813964844 4
     5 2013 2 31 2  55.59633 .         1 21386.875 0          1  55.59633           0 2013 1 0                  1 0
     5 2013 2 31 2  55.59633 .         2 21386.875 1          1  56.59633           0 2014 1 0                  2 1
     5 2013 2 31 2  55.59633 .         3 21386.875 2          1  57.59633           0 2015 1 0                  3 2
     5 2013 2 31 2  55.59633 .         4 21386.875 3          1  58.59633           0 2016 1 0                  4 3
     5 2013 2 31 2  55.59633 0 4.2694817 21386.875 4  .26948166  59.59633   -1.311255 2017 1 0  4.269481658935547 4
     6 2005 2 31 2  45.58127 .         1  7963.667 0          1  45.58127           0 2005 1 0                  1 0
     6 2005 2 31 2  45.58127 .         2  7963.667 1          1  46.58127           0 2006 1 0                  2 1
     6 2005 2 31 2  45.58127 0 2.1224136  7963.667 2  .12241364  47.58127  -2.1003494 2007 1 0 2.1224136352539063 2
     7 2009 2 31 2  47.30933 .         1   26118.1 0          1  47.30933           0 2009 1 0                  1 0
     7 2009 2 31 2  47.30933 .         2   26118.1 1          1  48.30933           0 2010 1 0                  2 1
     7 2009 2 31 2  47.30933 .         3   26118.1 2          1  49.30933           0 2011 1 0                  3 2
     7 2009 2 31 2  47.30933 0  3.951801   26118.1 3   .9518013  50.30933  -.04939898 2012 1 0  3.951801300048828 3
     8 2013 2 31 2  51.26113 .         1   26118.1 0          1  51.26113           0 2013 1 0                  1 0
     8 2013 2 31 2  51.26113 .         2   26118.1 1          1  52.26113           0 2014 1 0                  2 1
     8 2013 2 31 2  51.26113 .         3   26118.1 2          1  53.26113           0 2015 1 0                  3 2
     8 2013 2 31 2  51.26113 .         4   26118.1 3          1  54.26113           0 2016 1 0                  4 3
     8 2013 2 31 2  51.26113 0 4.5734596   26118.1 4   .5734596  55.26113  -.55606776 2017 1 0  4.573459625244141 4
     9 2017 2 31 2  55.83459 .         1   26118.1 0          1  55.83459           0 2017 1 0                  1 0
     9 2017 2 31 2  55.83459 .         2   26118.1 1          1  56.83459           0 2018 1 0                  2 1
     9 2017 2 31 2  55.83459 .         3   26118.1 2          1  57.83459           0 2019 1 0                  3 2
     9 2017 2 31 2  55.83459 0  3.510887   26118.1 3  .51088715  58.83459   -.6716065 2020 1 0 3.5108871459960938 3
    10 2005 2 31 2  41.39121 .         1 26216.125 0          1  41.39121           0 2005 1 0                  1 0
    10 2005 2 31 2  41.39121 .         2 26216.125 1          1  42.39121           0 2006 1 0                  2 1
    10 2005 2 31 2  41.39121 .         3 26216.125 2          1  43.39121           0 2007 1 0                  3 2
    10 2005 2 31 2  41.39121 0 3.7628365 26216.125 3   .7628365  44.39121   -.2707116 2008 1 0  3.762836456298828 3
    11 2009 2 31 2  45.15405 .         1 26216.125 0          1  45.15405           0 2009 1 0                  1 0
    11 2009 2 31 2  45.15405 .         2 26216.125 1          1  46.15405           0 2010 1 0                  2 1
    11 2009 2 31 2  45.15405 .         3 26216.125 2          1  47.15405           0 2011 1 0                  3 2
    11 2009 2 31 2  45.15405 0  3.776531 26216.125 3   .7765312  48.15405  -.25291842 2012 1 0  3.776531219482422 3
    12 2013 2 31 2  48.93058 .         1 26216.125 0          1  48.93058           0 2013 1 0                  1 0
    12 2013 2 31 2  48.93058 .         2 26216.125 1          1  49.93058           0 2014 1 0                  2 1
    12 2013 2 31 2  48.93058 .         3 26216.125 2          1  50.93058           0 2015 1 0                  3 2
    12 2013 2 31 2  48.93058 .         4 26216.125 3          1  51.93058           0 2016 1 0                  4 3
    12 2013 2 31 2  48.93058 0 4.5734634 26216.125 4  .57346344  52.93058   -.5560611 2017 1 0  4.573463439941406 4
    13 2017 2 31 2  53.50404 .         1 26216.125 0          1  53.50404           0 2017 1 0                  1 0
    13 2017 2 31 2  53.50404 .         2 26216.125 1          1  54.50404           0 2018 1 0                  2 1
    13 2017 2 31 2  53.50404 .         3 26216.125 2          1  55.50404           0 2019 1 0                  3 2
    13 2017 2 31 2  53.50404 0  3.990139 26216.125 3    .990139  56.50404 -.009909934 2020 1 0 3.9901390075683594 3
    14 2005 1 31 2 36.346706 .         1     13876 0          1 36.346706           0 2005 1 0                  1 0
    14 2005 1 31 2 36.346706 .         2     13876 1          1 37.346706           0 2006 1 0                  2 1
    14 2005 1 31 2 36.346706 .         3     13876 2          1  38.34671           0 2007 1 0                  3 2
    14 2005 1 31 2 36.346706 0  3.705326     13876 3   .7053261  39.34671  -.34909505 2008 1 0 3.7053260803222656 3
    15 2009 1 31 2  40.05203 .         1     13876 0          1  40.05203           0 2009 1 0                  1 0
    15 2009 1 31 2  40.05203 .         2     13876 1          1  41.05203           0 2010 1 0                  2 1
    15 2009 1 31 2  40.05203 .         3     13876 2          1  42.05203           0 2011 1 0                  3 2
    15 2009 1 31 2  40.05203 .         4     13876 3          1  43.05203           0 2012 1 0                  4 3
    15 2009 1 31 2  40.05203 0  4.611805     13876 4    .611805  44.05203   -.4913417 2013 1 0  4.611804962158203 4
    16 2013 1 31 2  44.66384 .         1     13876 0          1  44.66384           0 2013 1 0                  1 0
    16 2013 1 31 2  44.66384 0 1.4733658     13876 1   .4733658  45.66384   -.7478868 2014 1 0 1.4733657836914063 1
    17 2005 2 31 2 27.583185 .         1   13071.7 0          1 27.583185           0 2005 1 0                  1 0
    17 2005 2 31 2 27.583185 .         2   13071.7 1          1 28.583185           0 2006 1 0                  2 1
    17 2005 2 31 2 27.583185 .         3   13071.7 2          1 29.583185           0 2007 1 0                  3 2
    17 2005 2 31 2 27.583185 .         4   13071.7 3          1 30.583185           0 2008 1 0                  4 3
    17 2005 2 31 2 27.583185 0 4.0449123   13071.7 4  .04491234 31.583185   -3.103043 2009 1 0  4.044912338256836 4
    18 2009 2 31 2   31.6281 .         1   13071.7 0          1   31.6281           0 2009 1 0                  1 0
    18 2009 2 31 2   31.6281 .         2   13071.7 1          1 32.628098           0 2010 1 0                  2 1
    18 2009 2 31 2   31.6281 1 2.9891815   13071.7 2   .9891815 33.628098 -.010877427 2011 1 1 2.9891815185546875 2
    19 2005 1 31 2 26.463097 .         1     16000 0          1 26.463097           0 2005 1 0                  1 0
    19 2005 1 31 2 26.463097 .         2     16000 1          1 27.463097           0 2006 1 0                  2 1
    19 2005 1 31 2 26.463097 .         3     16000 2          1 28.463097           0 2007 1 0                  3 2
    19 2005 1 31 2 26.463097 0  3.729975     16000 3   .7299747 29.463097   -.3147453 2008 1 0 3.7299747467041016 3
    20 2009 1 31 2  30.19307 .         1     16000 0          1  30.19307           0 2009 1 0                  1 0
    20 2009 1 31 2  30.19307 .         2     16000 1          1  31.19307           0 2010 1 0                  2 1
    20 2009 1 31 2  30.19307 .         3     16000 2          1  32.19307           0 2011 1 0                  3 2
    20 2009 1 31 2  30.19307 .         4     16000 3          1  33.19307           0 2012 1 0                  4 3
    20 2009 1 31 2  30.19307 0 4.1982746     16000 4   .1982746  34.19307  -1.6181023 2013 1 0  4.198274612426758 4
    21 2005 1 31 2  25.55662 .         1  8697.875 0          1  25.55662           0 2005 1 0                  1 0
    21 2005 1 31 2  25.55662 .         2  8697.875 1          1  26.55662           0 2006 1 0                  2 1
    21 2005 1 31 2  25.55662 .         3  8697.875 2          1  27.55662           0 2007 1 0                  3 2
    21 2005 1 31 2  25.55662 0  3.740929  8697.875 3   .7409286  28.55662   -.2998509 2008 1 0 3.7409286499023438 3
    22 2009 1 31 2  29.29755 .         1  8697.875 0          1  29.29755           0 2009 1 0                  1 0
    22 2009 1 31 2  29.29755 1 1.0119114  8697.875 1 .011911392  30.29755    -4.43026 2010 1 1  1.011911392211914 1
    23 2017 1 31 2 37.844723 .         1  8697.875 0          1 37.844723           0 2017 1 0                  1 0
    23 2017 1 31 2 37.844723 .         2  8697.875 1          1  38.84472           0 2018 1 0                  2 1
    23 2017 1 31 2 37.844723 1  2.889221  8697.875 2   .8892212  39.84472  -.11740927 2019 1 1   2.88922119140625 2
    24 2005 1 31 2  24.87197 .         1   11872.5 0          1  24.87197           0 2005 1 0                  1 0
    24 2005 1 31 2  24.87197 .         2   11872.5 1          1  25.87197           0 2006 1 0                  2 1
    24 2005 1 31 2  24.87197 .         3   11872.5 2          1  26.87197           0 2007 1 0                  3 2
    24 2005 1 31 2  24.87197 .         4   11872.5 3          1  27.87197           0 2008 1 0                  4 3
    24 2005 1 31 2  24.87197 0  4.064083   11872.5 4   .0640831  28.87197  -2.7475746 2009 1 0  4.064083099365234 4
    25 2009 1 31 2 28.936054 .         1   11872.5 0          1 28.936054           0 2009 1 0                  1 0
    25 2009 1 31 2 28.936054 .         2   11872.5 1          1 29.936054           0 2010 1 0                  2 1
    25 2009 1 31 2 28.936054 .         3   11872.5 2          1 30.936054           0 2011 1 0                  3 2
    25 2009 1 31 2 28.936054 .         4   11872.5 3          1 31.936054           0 2012 1 0                  4 3
    25 2009 1 31 2 28.936054 0 4.0093117   11872.5 4 .009311676 32.936054   -4.676486 2013 1 0  4.009311676025391 4
    end
    label values sex sex2
    label def sex2 1 "male", modify
    label def sex2 2 "female", modify

    My first problem arises when I want to compare the incidence rates across two years as I get two quite different IRR if I include year_split (calendar year) or not:

    Code:
    * Age, sex and calenderyear adjusted
    * without calendaryear
    svy: poisson _d i.year i.sex age_split, offset(logtimerisk)
    margins, over(year) expression(predict(n)/FUT_split)
    nlcom exp(_b[2017.year] - _b[2005.year]) /*1,166*/
    
    * with calendaryear [year_split]
    svy: poisson _d i.year i.sex age_split year_split, offset(logtimerisk)
    margins, over(year) expression(predict(n)/FUT_split)
    nlcom exp(_b[2017.year] - _b[2005.year]) /*2,554*/
    My second problem arises when I want to look at the incidence rate ratio by sex:

    Code:
    svy: poisson _d i.year i.sex age_split year_split, offset(logtimerisk)
    margins, over(year sex) expression(predict(n)/FUT_split)
    nlcom exp(_b[2017.year##1.sex] - _b[2005.year##1.sex]) /*invalid matrix stripe;/
    Kind regards
    Jannie

  • #2
    Hi again,
    I'm still trying to figure out how to estimate incidence ratio when there is multiple specifications in the Poisson model. I've tried various things, but nothing works. I looked at the 'testnl_example', but still cant make it work.
    Fingers cross someone knows the way forward.
    BW, Jannie
    Code:
     svy: poisson _d i.year i.sex age_split year_split, offset(logtimerisk)  margins, over(year sex) expression(predict(n)/FUT_split)  nlcom exp(_b[2017.year#1.sex] - _b[2005.year#1.sex]) /*invalid matrix stripe;/

    Comment

    Working...
    X