1.5 Clinical overview of patients in the initial release of the MIMIC II database

In the first release of the MIMIC II database (January 2009) there are 26,655 patients (with unique Subject_ID's) of which 19,075 were adults (defined to be $\geq 15$ years old at time of last admission) and 6,538 were neonates ( $\leq 1$ month old at the time of first admission^1.1), These patients experienced a total of 29,545 hospital admissions (22,916 adults and 6,585 neonates). However, there were only 29,505 ICU-related hospital admissions (with 22,880 for adults and 6,582 for neonates). That is, a small fraction of the patients was admitted but never made it to the ICU for various reasons.

Distribution of major categories of primary ICD-9 codes for adult ICU-related hospital admissions (N=22,880).

Category	Code Range	Number of	%
Ischemic heart disease	410 - 414	4084	17.85%
Trauma	800 - 959	2337	10.21%
Digestive disease	520 - 579	2228	9.74%
Pulmonary disease	460 - 519	2070	9.05%
Infectious diseases	001 - 139	1603	7.01%
Neoplasms	140 - 239	1562	6.83%
Cerebrovascular disease	430 - 438	1426	6.23%
Other forms of heart disease	420 - 429	1367	5.97%
Arteries and veins	440 - 459	783	3.42%
Complications peculiar to certain specified procedures	996	674	2.95%
Heart failure	428	656	2.87%
Metabolic disorder	240 - 279	634	2.77%
Renal insufficiency	580 - 629	448	1.96%
Poisoning	960 - 989	356	1.56%
Other complications of procedures, NEC	998	349	1.53%
Chronic rheumatic heart disease	393 - 398	300	1.31%
Symptoms, signs, and ill-defined conditions	780 - 799	271	1.18%
Neurologic disease	320 - 389	255	1.11%
Diseases of the musculoskeletal system & connective tissue	710 - 739	247	1.08%
Diseases of pulmonary circulation	415 - 417	175	0.76%
Mental disorders	290 - 319	153	0.67%
Congenital anomalies	740 - 759	153	0.67%
Complications affecting specified body systems, not elsewhere classified	997	150	0.66%
Complications of pregnancy, childbirth, & puerperium	630 - 677	145	0.63%
Hypertensive disease	401 - 405	142	0.62%
Supplementary classification of factors influencing health status and contact with health services	V01 - V86	92	0.40%
Diseases of the blood and blood-forming organs	280 - 289	88	0.38%
Diseases of the skin and subcutaneous tissue	680 - 709	64	0.28%
Other and unspecified effects of external causes	990 - 995	54	0.24%
Without primary ICD9 code		6	0.03%
Complications of medical care, not elsewhere classified	999	6	0.03%
Acute Rheumatic fever	390 - 392	2	0.01%
Total		22,880	100.00%

The total number of ICU stays (number of ICUstay_ID's) was 33,361 (25,852 for adults and for 7,302 neonates), giving an average of 1.25 hospital stays per patient; 1.36 for adults and 1.12 for neonates. Details on how to calculate these numbers are given in chapter 2. Of the adult population, 38% was from the MICU, 26% from the SICU, 20% from the CSRU and 15% from the CCU. 2,430 Case_ID's (with associated waveforms and trends) have been associated with unique patients (Subject_ID's) Although we have collected neonatal waveforms, these will likely be released at a later date when the associated clinical data has been matched and verified. The rest of the description of the data in this section is therefore limited to the adult population.

Table 1.1 lists the thirty-one major primary^1.2 ICD-9 code categories for hospital admissions and the relative frequency for the adult patient population. Note that of the 22,916 adult hospital admissions (Hadm_ID), there was a total of 22,880 that were admitted to the ICU and therefore received ICD-9 codes. Note also that there are a range of codes associated with each condition. Furthermore, many non-primary codes are assigned to each ICU-related hospital admission with an average of 9 codes per admission. Table 6.1 in appendix 6.4 lists the frequency of all the ICD-9 codes (including the primary codes).

Figure 1.8 illustrates the age distribution of adult patients at the time of each ICU stay (N=25,852). Therefore, a patient will be represented more than once if they were admitted to the ICU more than once. The mean age at time of admission was 63.44 years and the median age was 65.33 years. Note small drops in admission rates for patients during their mid 20's to mid 30's, and again around 65 years of age. The distribution of length of stay for each ICU admission is illustrated in figure 1.9. Note that the distribution is heavy tailed, with a mean of 4.57 days and a median of only 2.15 days.

Figure 1.10 illustrates the distribution of patients' Simplified Acute Physiology Score (SAPS I) at each admission (lower plot) and the associated hospital mortality rate^1.3 (upper plot). Note that the mortality rate begins to climb steeply around a SAPS I value of 23. Although the mortality rate begins to fluctuate wildly above this value, this is likely an artifact of low numbers. The mean SAPS I was 13.68 and the median was 13. Figure 1.11 illustrates how this mortality varies during a hospital stay by showing the percentage of patients in the MIMIC II database that die in the ICU, and after discharge to the floor (HOSP) for all admissions (upper left) and for the subset with CMO (Comfort Measures Only) (lower right). Notice that mortality curve as a function of SAPS 1 values is not perfectly smooth, particularly above a SAPS 1 value of 20. Above a SAPS 1 value of 30, the mortality rate fluctuates wildly, but this may be an artifact of the small numbers of individuals admitted with such a value.