Introduction
Materials and methods
Patient data
Staffing model
Study design and statistics
Results
Basic clinical data
Clinical data | All (N = 2,891) | Admission group | |||
---|---|---|---|---|---|
NT (n = 175) | OH (n = 2,716) |
P
| |||
Mean age (years) | 61 | 62 | 61 | 0.515b | |
Gender (n) | Male | 1,944 | 124 (71.9%) | 1,820 (67.0%) | 0.293c |
Female | 947 | 51 (29.1%) | 896 (33.0%) | ||
Emergency (n) | No | 2,455 | 71 (40.6%) | 2,384 (87.8%) | <0.001c |
Yes | 436 | 104 (59.4%) | 332 (12.2%) | ||
Source (n) | Medical | 105 | 15 (8.6%) | 90 (3.3%) | <0.001c |
Surgical | 2,786 | 160 (91.4%) | 2,626 (96.7%) | ||
Transfusion (n) | No | 2,571 | 151 (86.3%) | 2,420 (89.1%) | 0.250c |
Yes | 320 | 24 (13.7%) | 296 (10.9%) | ||
Mechanical ventilation (n) | No | 2,550 | 149 (85.1%) | 2,401 (88.4%) | 0.201d |
NIV | 21 | 1 (0.6%) | 20 (0.7%) | ||
Intubation | 246 | 22 (12.6%) | 224 (8.2%) | ||
Tracheotomy | 74 | 3 (1.7%) | 71 (2.7%) | ||
Median APACHE II score | 8 | 14 | 8 | <0.001b | |
ICU outcome (n) | Dead | 125 | 27 (15.4%) | 98 (3.6%) | <0.001c |
Alive | 2,766 | 148 (74.6%) | 2,618 (96.4%) | ||
Median LOSHOS (days) | 14 | 16 | 14 | 0.581b | |
Median LOSICU (hours) | 24 | 42 | 24 | 0.011b | |
Mortality (28 days), n (%) | 111 (3.8%) | 25 (14.3%) | 86 (3.2%) | <0.001c |
Mortality at 28 days
Variables | Dead | Alive |
P
| |
---|---|---|---|---|
Admission time, n (%) | NT | 25 (14.3) | 150 (85.7) | <0.001b |
OH | 86 (3.2) | 2,630 (96.8) | ||
Emergency admission, n (%) | No | 36 (1.5) | 2,419 (98.5) | <0.001b |
Yes | 75 (17.2) | 361 (82.8) | ||
Source, n (%) | Medical | 27 (25.7) | 78 (74.3) | <0.001c |
Surgical | 84 (3.0) | 2,702 (97) |
95% CI | ||||
---|---|---|---|---|
Variables by cohort |
P
| Odds ratio | Lower | Upper |
All patients | ||||
Admission time (NT vs. OH) | 0.010 | 1.920 | 1.171 | 3.150 |
Emergency admission (yes vs. no) | <0.001 | 51.241 | 36.234 | 72.463 |
Source (medical vs. surgical) | 0.959 | 1.014 | 0.606 | 1.697 |
APACHE II score | <0.001 | 1.032 | 1.017 | 1.047 |
PSM cohort | ||||
Admission time (NT vs. OH) | 0.022 | 2.187 | 1.119 | 4.271 |
Emergency admission (yes vs. no) | 0.788 | 1.135 | 0.452 | 2.848 |
Source (medical vs. surgical) | 0.071 | 0.404 | 0.151 | 1.079 |
APACHE II score | <0.001 | 1.161 | 1.103 | 1.222 |
Survival analysis
Univariate and multivariate analyses in the entire cohort
Variables |
P
| OR | 95% CI for HR | |
---|---|---|---|---|
Lower | Upper | |||
Age (years) | 0.199 | 1.008 | 0.996 | 1.021 |
Gender (female/male) | 0.997 | 0.999 | 0.648 | 1.542 |
Source (medical/surgical) | 0.153 | 1.408 | 0.880 | 2.251 |
Emergency admission (yes/no) | 0.254 | 1.270 | 0.842 | 1.914 |
Transfusion (no/yes) | 0.541 | 0.878 | 0.578 | 1.333 |
Mechanical ventilation (no/NIV/intubation/tracheotomy) | 0.060 | 0.829 | 0.682 | 1.008 |
Admission day (weekday/weekend) | 0.198 | 0.724 | 0.443 | 1.183 |
Admission time (NT/OH) | 0.010 | 1.725 | 1.118 | 2.744 |
APACHE II score (>8/≤8) | <0.001 | 1.113 | 1.082 | 1.145 |
Stratified analyses
Propensity score matching analysis
OH | NT | |||||
---|---|---|---|---|---|---|
Total (dead) | Median LOSICU(hours) | Total (dead) | Median LOSICU(hours) |
P
| ||
APACHE II score (median) | 14 (23) | 65 | 14 (27) | 42 | 0.983a | |
Emergency admission (n) | No | 71 (2) | 24 | 71 (1) | 38 | 1.0b |
Yes | 104 (21) | 90 | 104 (26) | 58 | ||
Source (n) | Medical | 15 (5) | 95 | 15 (3) | 37 | 1.0b |
Surgical | 160 (18) | 48 | 160 (24) | 43 | ||
Mean age (years) | 65 (23) | 65 | 62 (27) | 42 | 0.139a | |
Gender (n) | Male | 115 (16) | 66 | 124 (22) | 60 | 0.301b |
Female | 60 (7) | 49 | 51 (5) | 37 | ||
Mechanical ventilation (n) | No | 145 (6) | 43 | 149 (8) | 37 | 0.840c |
NIV | 2 (0) | 145 | 1 (0) | 59 | ||
Intubation | 23 (14) | 87 | 22 (18) | 58 | ||
Tracheotomy | 5 (3) | 307 | 3 (1) | 200 | ||
Transfusion (n) | No | 147 (13) | 46 | 151 (12) | 38 | 0.548b |
Yes | 28 (10) | 178 | 24 (15) | 59 |
Weekend admission
Discussion
Different patient outcomes between office hours and nighttime admissions
Higher mortality in nighttime admissions might be related to staffing model
ICU training for residents may be a promising choice in China
Limitations and perspectives of the study
Conclusions
Key messages
-
NT admission is related to ICU mortality even after PSM analysis.
-
Comparisons between weekend and weekday admissions supported our notion that higher mortality at night may be related to differences in staffing coverage.
-
Given the current shortage of financial and staffing resources, as well as the insufficient ICU training courses for residents, upgrading our CCM training system and improving our clinical protocols so that dedicated trainees can accomplish NT coverage with on-call intensivist supervision is a primary necessity.