Replicating the Demographics Table in the MIMIC-IV Paper

[Incantatoric]

I am attempting to replicate the demographics table from the MIMIC-IV paper titled “MIMIC-IV, a freely accessible electronic health record dataset, 2023” (https://www.nature.com/articles/s41597-022-01899-x).

I have been successful in duplicating the results for the first two rows, “Number of stays” and “Unique patients”. However, I am encountering difficulties with the remaining rows. Below are the SQL queries I used and the corresponding results for the ICU case:

/* Age. avg: 63.45, std: 17.22 */
SELECT AVG(p.anchor_age) AS average_age, STDDEV(p.anchor_age) AS sd_age, count(*)
FROM mimiciv_hosp.patients p
WHERE p.subject_id IN (
    SELECT i.subject_id
    FROM mimiciv_icu.icustays i
);

/* LoS: avg: 4.96, std: 7.83 */
select count(*), avg(total_los), stddev(total_los) 
from(
    SELECT 
        subject_id, 
        SUM(los) AS total_los
    FROM 
        mimiciv_icu.icustays
    GROUP BY 
        subject_id
) as hey;

/* In-hospital mortality rate: 6,966 */
select count(*) from mimiciv_hosp.admissions a 
where a.hospital_expire_flag =1
and a.hadm_id IN (
    SELECT i.hadm_id
    FROM mimiciv_icu.icustays i
);

/* One year mortality (this is for hospital admissions): 29,076 */
select count(*) from mimiciv_hosp.patients p
where p.dod is not null;

For the one-year mortality rate, there is a notebook available (https://github.com/MIT-LCP/mimic-code/blob/main/mimic-iv/notebooks/tableone.ipynb) that calculates it. However, when I adapt the datetime function to the SQL version, the count is 11,858, which differs from the figure presented in the paper.

Additionally, I would like to understand what “Female Administrative Gender” signifies in one of the table rows.