Question on Diagnosis Timing in the MIMIC Database for Research on Sepsis and AKI

[guangyongjin]

In the MIMIC database, ICD codes are used to identify patients with specific diagnoses. In the context of my research on building a predictive model for acute kidney injury (AKI) in sepsis patients during their ICU stay, it is crucial to clarify the timing of diagnosis, particularly with respect to ICU admission. Specifically, I need to determine whether the diagnosis of sepsis and AKI was made before, during, or after the patient was admitted to the intensive care unit. Could you please advise on the best approach to ensure that: 1. Sepsis Diagnosis: How can I confirm that the sepsis diagnosis occurred at the time of ICU admission? 2. AKI Diagnosis: How can I ensure that AKI was diagnosed during the hospitalization in the ICU, after the patient’s ICU admission, rather than before the period leading up to the ICU stay? Any guidance or suggestions on how to best address these timing issues would be greatly appreciated, as they are critical for the accuracy of my predictive model. Let me know if you'd like further revisions or additions!

[jijitoutou]

I'm a self-taught medical enthusiast and began studying medicine independently after the COVID-19 pandemic in 2020. I am also a user of MIMIC-IV from China.

Here's my opinions.

Sepsis Diagnosis Sepsis was defined as a SOFA score ≥2 within the first 24 hours after ICU admission. Therefore, users can consider sepsis to occur after admittion to the ICU.

Definition of SA-AKI Timing You can determine the time of SA-AKI by subtracting the sepsis confirmation time from the time of the septic patient's first recorded AKI. The sepsis confirmation time is the later of the two: sofa_time and infectious_time. If AKI occurs within 168 hours (7 days) after this sepsis time point, it is classified as SA-AKI. If a patient has multiple AKI records, you should only the earliest AKI record.

Exclusion Criteria You may exclude patients who had any of the following before the confirmed time of sepsis: CRRT treatment prior to sepsis confirmation time History of renal failure Long-term dialysis Chronic kidney disease stage 4 or 5

“tidy” R language may be a useful tool for data cleansing.

我是一个自学医学的爱好者。在2020年COVID-19大流行后，自学了医学知识。 我也是一位来自于中国的MIMIC-IV用户。 以下是我的观点。

1 脓毒症的诊断：使用的是入住ICU 24小时内的SOFA评分≥2，因此可以认为脓毒症是发生在入住ICU之后。 2 SA-AKI的诊断：如果患者有多条AKI的记录，你需要保留时间最早的那一条。将脓毒症患者首次AKI的时间，减去脓毒症表格中的sofa_time与infectious_time两者中更晚的那个时间（脓毒症确诊时间）。若在168小时内，即为SA-AKI。 当然，你还要考虑排除在脓毒症确诊之前接受CRRT治疗，肾衰病史，长期透析，CKD 4/5期的患者。

我使用tidy风格的R语言进行了以上的数据清洗。

希望这些信息可以帮助到大家。