nus-mornin-lab / echo-mimiciii

Transthoracic echocardiography and mortality in sepsis: analysis of the MIMIC-III database
https://doi.org/10.1007/s00134-018-5208-7
49 stars 36 forks source link

Regarding Angus Script Results #2

Closed jueurousseau closed 6 years ago

jueurousseau commented 6 years ago

We used the angus.sql script (https://github.com/MIT-LCP/mimic-code/blob/master/concepts/sepsis/angus.sql) and produced the following.

1) select count(distinct subject_id) from mimiciii_angus_sepsis where angus = 1; -- 12,636 DISTINCT PATIENTS 2) select count(*) from mimiciii_angus_sepsis where angus = 1; -- 15,254 ROW COUNT

This is inconsistent with your results in your paper (https://link.springer.com/article/10.1007%2Fs00134-018-5208-7). I also found that terms admissions and patients were used interchangeably, but I believe these are distinct variables in MIMIC. Do you mean admissions or patients for your query (i.e. Figure 1)?

kiendang commented 6 years ago

We mean icu admission actually.

Using subject_id (patient)

select count(distinct subject_id) from angus_sepsis where angus = 1;
# 12636

Using hadm_id (hospital admission)

select count(distinct hadm_id) from angus_sepsis where angus = 1;
# 15254

Using icustay_id (ICU admission, this is what we used in our paper)

select count(distinct icustay_id) from icustays a left join angus_sepsis b on a.hadm_id = b.hadm_id where angus = 1;
# 17420