Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: Long format data analysis confusion

 From Steve Nakoneshny <[email protected]> To "[email protected]" <[email protected]> Subject st: Long format data analysis confusion Date Mon, 10 Jun 2013 16:19:31 -0600

Dear Statalisters,

I have a long format dataset of ~34,000 observations of healthcare utilization data on a cohort of n=115 surgical patients. These patients are dichotomized by whether they had a particular intervention immediately post-surgery (n=60 "no"; n=55 "yes") or not. Our question is to determine what impact our intervention has had on post-operative healthcare utilization.

As our data are categorical, we have felt (perhaps simplistically) that using a chi2 would be sufficient. As per the below output, we compared the volume of healthcare encounters by sector vs. our pathway intervention for both a 3 month post-operative and 6 month post-operative period. Somewhat confusingly to us, the 6 month post-operative comparison is not significant despite a dramatic difference in the distribution.

Which leads me to my question(s):

1) is there perhaps an alternate analysis approach that we could (or should) be using for such data?
2) we were thinking about factoring some sort of weighting so we would be comparing equally sized groups (such that the pathway patients were weighted as 60/55), but we aren't sure of either appropriate syntax as per -h weights- or if it is even A Good Idea.

Any advice or suggestions would be appreciated.

. tab sector pathway if threepost==1,chi2 row col

Healthcare |   the Care Pathway?
Sector |        no        yes |     Total
-----------+----------------------+----------
ER |        18         17 |        35
|     51.43      48.57 |    100.00
|      0.72       1.54 |      0.97
-----------+----------------------+----------
Outpatient |       459        152 |       611
|     75.12      24.88 |    100.00
|     18.43      13.79 |     17.01
-----------+----------------------+----------
Inpatient |        13         11 |        24
|     54.17      45.83 |    100.00
|      0.52       1.00 |      0.67
-----------+----------------------+----------
MD Claims |     2,000        922 |     2,922
|     68.45      31.55 |    100.00
|     80.32      83.67 |     81.35
-----------+----------------------+----------
Total |     2,490      1,102 |     3,592
|     69.32      30.68 |    100.00
|    100.00     100.00 |    100.00

Pearson chi2(3) =  18.5821   Pr = 0.000

. tab sector pathway if sixpost==1,chi2 row col

Healthcare |   the Care Pathway?
Sector |        no        yes |     Total
-----------+----------------------+----------
ER |        47         15 |        62
|     75.81      24.19 |    100.00
|      1.19       1.48 |      1.25
-----------+----------------------+----------
Outpatient |       789        175 |       964
|     81.85      18.15 |    100.00
|     20.05      17.31 |     19.49
-----------+----------------------+----------
Inpatient |        30         11 |        41
|     73.17      26.83 |    100.00
|      0.76       1.09 |      0.83
-----------+----------------------+----------
MD Claims |     3,070        810 |     3,880
|     79.12      20.88 |    100.00
|     78.00      80.12 |     78.43
-----------+----------------------+----------
Total |     3,936      1,011 |     4,947
|     79.56      20.44 |    100.00
|    100.00     100.00 |    100.00

Pearson chi2(3) =   5.1202   Pr = 0.163

Thanks,
Steve

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/