3.1 Demographic characteristics
A total of 561cases were included in the initial analysis, with 270 in the control group and 291 in the pharmacist-exposed group. A comparison of baseline characteristics between control group and pharmacist-exposed group before and after propensity score matching is shown in figure 1 and table 1, which also showed a brief overview of the main infectious diseases type diagnosed. Basic demographic characteristics in both groups are similar to each other after PSM matching including the age, gender, race and baselines of APECH-II score, CCI score and results of lab examinations. After PSM, The median age of the patients encompassed in this study was 85 years old in the control group and 84 years in the pharmacist-exposed group. The ratio of men and women is close to 1:1 in both two groups. The median values of baseline APECH-II score was 24 in the control group and 22 in the pharmacist-exposed group. The median values of baseline CCI was the same between the two groups which was 5. We also found bacterial pneumonia (29.85%), sepsis (23.79%), urinary infection (13.59%), and septic shock (11.17%) were the main infectious diseases diagnosed during the study period, and then were the fungal pneumonia (5.83%) and fungal urinary infection (1.84%). There is no significant differences of infectious diseases types between control group and the pharmacist intervention group.
Tab 1
Baseline Characteristics of study patients before and after PSM
|
Before PSM
|
After PSM
|
|
The control group(N=270)
|
The pharmacist-exposed group(N=291)
|
p
|
The control group (N=102)
|
The pharmacist-exposed group(N=102)
|
p
|
Age (Median year, IQR)
|
85.50 (79.00, 90.00)
|
83.00 (74.00, 88.00)
|
0.002*
|
85.00 (77.00, 89.00)
|
84.00 (74.00, 89.00)
|
0.716
|
65~84y, n (%)
|
96.00 (35.56%)
|
144.00 (49.48%)
|
|
36.00 (35.29%)
|
49.00 (48.04%)
|
|
≥85y, n (%)
|
146.00 (54.07%)
|
121.00 (41.58%)
|
|
54.00 (52.94%)
|
46.00 (45.10%)
|
|
Gender, male, n (%)
|
149.00 (55.19%)
|
148.00 (50.86%)
|
0.305
|
47.00 (46.08%)
|
49.00 (48.04%)
|
0.780
|
Race, ethnic Han, n (%)
|
264.00 (97.78%)
|
281.00 (96.56%)
|
0.388
|
101.00 (99.02%)
|
97.00 (95.10%)
|
0.098
|
Baseline APECH-II (Median, IQR)
|
26.50 (23.00, 28.00)
|
17.00 (13.00, 22.00)
|
0.000*
|
24.00 (19.00, 26.50)
|
22.00 (19.00, 27.00)
|
0.901
|
Baseline CCI,Median(IQR)
|
5 (4, 6)
|
6 (5, 8)
|
0.000*
|
5 (4, 6)
|
5 (4, 7)
|
0.662
|
Baseline Lab examinations
|
|
|
|
|
|
|
CRP, Median(IQR)
|
44.42 (10.05, 99.65)
|
50.96 (10.36, 118.20)
|
0.541
|
32.84 (6.41, 81.90)
|
43.23 (8.95, 100.66)
|
0.300
|
PCT, Median(IQR)
|
0.26 (0.10, 1.33)
|
0.20 (0.07, 0.92)
|
0.018
|
0.24 (0.09, 1.33)
|
0.24 (0.06, 0.92)
|
0.236
|
WBC, Median(IQR)
|
9.07 (6.73, 13.61)
|
9.45 (6.72, 13.17)
|
0.950
|
9.01 (6.76, 13.60)
|
9.94 (6.90, 13.17)
|
0.621
|
NE%, Median(IQR)
|
85.15 (77.00, 90.10)
|
84.40 (75.60, 90.40)
|
0.752
|
85.45 (75.30, 90.00)
|
82.65 (74.70, 89.30)
|
0.203
|
Main infectious diseases, n
|
927
|
1006
|
0.317
|
399
|
412
|
0.317
|
Bacterial pneumonia (J15.901*), n (%)
|
249 (26.86%)
|
323 (32.11%)
|
|
109 (27.32%)
|
123 (29.85%)
|
|
Sepsis (A41.902*), n (%)
|
232 (25.03%)
|
245 (24.35%)
|
|
100 (25.06%)
|
98 (23.79%)
|
|
Urinary infection (N39.001*), n (%)
|
133 (14.35%)
|
132 (13.12%)
|
|
58 (14.54%)
|
56 (13.59%)
|
|
Septic shock (A41.903*), n (%)
|
107 (11.54%)
|
93 (9.24%)
|
|
43 (10.78%)
|
46 (11.17%)
|
|
Fungal pneumonia (B49xx20*), n (%)
|
54 (5.83%)
|
58 (5.77%)
|
|
21 (5.26%)
|
24 (5.83%)
|
|
Fungal urinary infection (B49xx04*), n (%)
|
28 (3.02%)
|
22 (2.19%)
|
|
11 (2.76%)
|
8 (1.94%)
|
|
* ICD 10 code
3.2 Activities of clinical pharmacists
A total of 862 pharmacist-recommendations were identified, of which consensus made between the pharmacist and the ICU physicians was 833 (96.64%). The top 5 recommendations mostly occurred by the ICU pharmacist were medication regimen adjustments by diseases on progression (180/862, 20.88%) with 95.00% acceptance rate (171/180), medication regimen adjustments by microbial results (170/862, 19.72%) with 95.88% acceptance rate (163/170), drug withdrawal by full treatment courses (114/862, 13.23%) with 96.49% acceptance rate (110/114), suggestions for TDM (103/862, 11.95%) with 98.06% acceptance rate (101/103) and medication regimen adjustments by de-escalation (75/862, 8.70%) with 100.00% acceptance rate (75/75). The details were shown in table 2.
Tab 2
Top 5 pharmacist activities mostly occurred and the acceptance by the ICU clinicians
No.
|
Contents
|
No.
|
Proportion (%)
|
Consensus No. with ICU physicians
|
Acceptance (%)
|
1
|
Medication regimen adjustments by diseases on progression
|
180
|
20.88%
|
171
|
95.00%
|
2
|
Medication regimen adjustments by microbial results
|
170
|
19.72%
|
163
|
95.88%
|
3
|
Drug withdrawal by full treatment courses
|
114
|
13.23%
|
110
|
96.49%
|
4
|
Suggestions for TDM*
|
103
|
11.95%
|
101
|
98.06%
|
5
|
Medication regimen adjustments by de-escalation
|
75
|
8.70%
|
75
|
100.00%
|
*TDM, Therapeutic drug monitoring
3.3 The positive influence to the ICU clinicians on drug selection tendency
We are interested in the impact of pharmacist intervention on antibiotic usage in ICU patients. In a summary, the AUD of all antibiotics consumed in the ICU decreased from 241.91 DDD/100 bed days to 176.64 DDD/100 bed days after implementing the intervention (p=0.018). By agent, AUD proportion was observed decreasingly in carbapenems (23.07% vs 14.43%), triazole and other antimycotics for systemic use (16.98% vs 18.53%), glycopeptides and linezolid (15.32% vs 14.83%), tetracyclines (11.56% vs 6.26%) and third-generation cephalosporins (6.91% vs 4.57%). The descender was 48.59%, 10.27%, 20.44%, 55.52% or 45.61%, respectively. Whereas, AUD proportion was increased in aminoglycosides (9.41% vs 11.77%), monobactams (4.76% vs 9.47%), combinations of penicillin and beta-lactamase inhibitors (3.71% vs 10.65%) and fluoroquinolones (2.42% vs 0.91%). The descender was -2.77%, -63.68%, -135.65% or -14.92%, respectively. The corresponding results were shown in Figure 2.
3.4 Clinical and economic outcomes with and without ICU pharmacists
As shown in Table 3, the ICU mortality (p=0.607) and LOS (p=0.163) with or without pharmacist exposed had no statistical difference. About the cost, pharmacist intervention significantly decreased the antibiotic charges from $836.3(IQR 426.88, 1682.09) to $362.15 (IQR 148.23, 1034.4) (p< 0.001). Correspondingly, total cost of all medications were reduced from $2868.18 (IQR $1268.44, $5059.00) vs $1941.5 (IQR $1092.89, $3538.97) (p=0.016).
Tab 3
Clinical and economic outcomes of critically ill patients with infectious diseases with and without clinical pharmacists in ICUs.
Clinical and economic outcomes
|
The control group
|
The pharmacist-exposed group
|
P
|
Mortality, No (%)
|
20.00 (19.61%)
|
23.00 (22.55%)
|
0.607
|
LOS, Median(IQR)
|
8.00 (5.00, 13.00)
|
6.90 (4.50, 12.15)
|
0.163
|
Antibiotic charges per stay ($/case), Median(IQR)
|
836.3 (426.88, 1682.09)
|
362.15 (148.23, 1034.4)
|
< 0.001
|
Drug charges per stay ($/case), Median(IQR)
|
2868.18 (1268.44, 5059.00)
|
1941.5 (1092.89, 3538.97)
|
0.016
|
3.5 The relationship between mortality and pharmacist interventions
Univariate analyses showed that there was no statistically difference in pharmacist intervention between the groups of survival and death (p=0.288), but in the factors of age (p=0.001), baseline APECH-II score (p=0.029), baseline CCI score (p=0.032), numbers of admissions (p=0.001), numbers of antibacterial drugs adoptions (p=0.007), numbers of antibacterial drugs combinations (p=0.005), discharge diagnostic items (p<0.001), total number of bacteria detected (p=0.001), numbers of gram-positive cocci detected (p=0.038) and number of fungi detected (p=0.002) were as significant variables affecting the morality. Results of bivariate logistic regression was shown that the mortality outcome was related to the patients’ numbers of admissions (OR=1.095, 95%CI 1.028-1.165), discharge diagnostic items (OR=1.109, 95%CI 1.054-1.166) and numbers of gram-positive cocci detected (OR= 3.392, 95%CI 1.348-8.531). The details were shown in supplement 1 and 2.