BACKGROUND
Optimizing antibiotic treatment is a promising method to reduce the length of intensive care unit (ICU) stay and therefore reduce ICU costs.1,2 3 4,5 target attainment for antibiotics is essential for therapeutic success.6,7 target attainment for widely used beta-lactam antibiotics can be as low as 40%–60%.8,9 10
Individualizing treatment by using therapeutic drug monitoring (TDM) has been proposed to optimize the dosing of selected antibiotics.6 target attainment . The benefits of TDM have been established for vancomycin and aminoglycosides.11,12 4
Few studies have examined the effects of target attainment on ICU costs. To explore this relationship, we evaluated data from EXPAT trial.8 target attainment on the costs of intensive care and identified factors that may have contributed to any differences in costs.
MATERIALS AND METHODS
Primary Aim
The primary aim of this study was to describe the difference in total ICU costs between patients with beta-lactam target attainment and those with target nonattainment.
Secondary Aims
The secondary aim was to describe the difference in daily ICU costs between patients with beta-lactam target attainment and those with target nonattainment, as well as to identify cost determinants.
Population
We conducted our analyses using data from the EXPAT trial, a prospective, observational pharmacokinetic/pharmacodynamic (PK/PD) study including 2 centers.8
Target attainment was defined as reaching the pharmacodynamic target (PDT) for the antibiotic. The PDT of these antibiotics was defined as an unbound plasma concentration above at least one time the minimal inhibitory concentration (MIC) for 100% of the dosing interval (100%ƒT > 1MIC). MIC is the minimum concentration required to prevent the visible growth of a bacterium in vitro. Patients were divided into 2 study groups based on blood sampling: target attainment and target nonattainment.
Data Extraction
Patient characteristics were extracted from the electronic health records on the first day of antibiotic administration in the EXPAT trial. ICU costs between January 2016 and June 2017 were extracted from the hospital administration system. All declared hospital costs during ICU admission of included patients were collected and categorized into 10 categories: specialist consultation, renal replacement therapy (RRT), bedside procedures, laboratory diagnostics, microbiology, surgery, pathology, radiology, transfusion (blood-derived products), and fixed ICU admission costs.
Economic Evaluation
Cost analyses were performed from a health care perspective using the Dutch guidelines for cost studies.13 14
All costs are described as the total costs during ICU admission and daily costs. Daily costs were calculated by dividing the total costs by ICU LOS. ICU costs consist of variable and fixed costs. Fixed ICU admission costs were the declared daily costs of admission and included aggregated costs of staff, maintenance and acquisition of devices, hospital space, and preparation of medication. Variable costs were defined as total costs minus the fixed ICU admission costs. Costs that were not declared in 2016 were adjusted to the standard inflation defined by the Dutch Central Bureau of Statistics (CBS) to match the costs in 2016.15
Statistical Analysis
Patient characteristics are expressed as mean values with SDs for normally distributed data; alternatively, they are expressed as median values with interquartile ranges (IQRs). Categorical data are expressed as counts with percentages. Normality was tested using the Shapiro–Wilk test. Differences in patient characteristics were calculated with an independent 2-tailed Student t test or Mann–Whitney U test, as appropriate. Categorical differences were tested using the χ2 or Fisher exact test, as appropriate.
For our primary aim, we assessed differences between costs using the Mann–Whitney U test in addition to a 2-tailed t test with bootstrapping (×1000). As costs generally tend to present a right-skewed distribution, they are expressed in Euro (€) as median with IQR.
For our secondary aims, we explored whether beta-lactam target attainment is a cost determinant. We performed a general linear regression analysis of the total costs. Both patients with target attainment and target nonattainment were included in this analysis. For the regression model, we selected 4 relevant variables for costs: sex, age, Sequential Organ Failure Assessment (SOFA) score, and RRT. Target attainment was included as the most important factor in the model. Furthermore, we analyzed a model without RRT because RRT strongly affects both reaching the target attainment and LOS. Effects were reported as mean differences or odds ratios (ORs) with corresponding confidence intervals (CIs). McFadden R2 was used to determine the proportion of the variance in the total cost predicted by the model. Analysis of variance was used to calculate statistical differences.
All analyses were performed with “R” version 3.6.3 (2020, Vienna, Austria). In all analyses, a P value below 0.05 was considered significant, unless stated otherwise.
RESULTS
Patient Characteristics
In total, 79 patients were included in the analysis. Based on serum antibiotic analyses, 50 patients were allocated to antibiotic target attainment and 29 to antibiotic target nonattainment. The population demonstrating target attainment presented a higher age, higher SOFA score at the initiation of antibiotic therapy, and increased RRT (Table 1 ). Although not significant, there was an important difference in ICU LOS, which was 15 days (IQR 7–28) in patients with target attainment when compared with 7 days (IQR 5–18) in those with target nonattainment. Although not significant, but important, mortality at 30 days was higher in patients with target attainment (24%) than in those with target nonattainment (13.8%).
TABLE 1. -
Baseline Characteristics and Clinical Outcomes
Target Attainment (N = 50)Target Nonattainment (N = 29)
P
Age
63.0 [56.3–68.8]
58.0 [51.0–64.0]
0.047
†
Male
28 (56.0%)
24 (82.8%)
0.026
‡
BMI (kg/m2 )
26.0 [23.6–28.9]
24.9 [21.8–26.3]
0.127†
Target antibiotic
Trough concentration (mg/L)
0.001
‡
Cefotaxim
12 (24.0%)
16 (55.2%)
18.0 [10.7–27.8]
2.13 [1.10–3.75]
Ceftazidim
4 (8.0%)
0 (0%)
72.2 [50.9–104]
Ceftriaxon
15 (30.0%)
1 (3.4%)
5.55 [2.73–9.56]
1.94 [1.94–1.94]
Cefuroxime
0 (0%)
2 (6.9%)
5.22 [4.06–6.38]
Augmentin
4 (8.0%)
4 (13.8%)
30.8 [17.9–54.5]
3.78 [3.20–7.12]
Meropenem
15 (30.0%)
6 (20.7%)
12.2 [3.83–37.0]
1.36 [0.966–1.52]
Sepsis at admission
10 (20.0%)
2 (6.9%)
0.116‡
APACHE Score
22.2 (6.24)
20.8 (5.95)
0.331*
SOFA score
8.70 (3.48)
7.00 (3.47)
0.040
*
Albumin (g/L)
23.5 [20.3–29.0]
31.0 [25.0–35.0]
0.009
†
Serum creatinine (umol/L)
118 [70.3–158]
77.0 [60.0–100]
0.019
†
Leukocyte count (×109 /L)
12.3 [7.63–18.1]
16.3 [11.1–18.0]
0.106†
CRP (mg/L)
86.5 [47.5–221]
62.0 [10.0–144]
0.038
†
Transfusion received
40 (80%)
15 (51.7%)
0.011
‡
RRT
11 (22.0%)
1 (3.4%)
0.047
‡
30-day mortality
12 (24.0%)
4 (13.8%)
0.386‡
ICU LOS
15.0 [7.00–28.0]
9.00 [5.00–18.0]
0.133†
Bold P values are significant (lower than 0.05).
* t test.
† Mann–Whitney U test.
‡ Fisher exact test.
APACHE, Acute Physiology And Chronic Health Evaluation version 2; BMI, body mass index; CRP, C-reactive protein.
Hospital Costs
Fixed admission costs accounted for 60.1% of the total costs, leaving the variable costs at 39.9%. Target attainment , compared with target nonattainment, showed a trend toward higher total ICU costs (€44.600 and €28.200, P = 0.103) (Table 2 ). Furthermore, the same trend was observed in variable costs (€16.500 and €11.800, P = 0.076), RRT costs (€0 and €0, P = 0.065), and medical microbiology costs (€2.270 and €1.840, P = 0.065). Target attainment was significantly associated with the increased transfusion of blood product costs (€1.050 and €229, P = 0.010).
TABLE 2. -
Total ICU Costs Split by Cost Categories
Categories
Target Attainment (N = 50)Target Nonattainment (N = 29)
P
Median
IQR1–IQR3
Median
IQR1–IQR3
Mann–Whitney U Test
Bootstrapped t test
Total
€44.600
22.100–70.900
€28.200
16.800–48.000
0.10
0.05
Variable
€16.500
8.640–28.900
€11.800
6.610–18.600
0.08
0.04
Fixed admission
€29.400
13.600–52.300
€19.600
8.310–34.600
0.23
0.06
Consultation
€1.340
549–3.220
€1.780
660–2.780
0.79
0.28
RRT
€0
0–0
€0
0–0
0.07
<0.01
Bedside procedures
€747
361–1.220
€722
143–1.110
0.55
0.18
Laboratory diagnostics
€2.600
1670–4.660
€1.830
947–2.870
0.09
0.03
Radiology
€993
476–2.010
€1.130
344–2.610
0.83
0.95
Transfusion
€1.050
459–4.650
€229
0–1.470
0.01
0.06
Microbiology
€2.270
856–5.690
€1.840
525–3.190
0.06
<0.01
Pathology
€0
0–0
€0
0–0
0.68
0.89
Surgery
€450
0–2.290
€346
0–1.260
0.58
0.74
The RRT and pathology costs are described as €0, with an IQR of €0–€0. This can be explained by the fact that less than 25% of patients account for all costs in these categories. Total RRT costs for the patients who incurred these costs were €2500 (1610–5850) for target attainment when compared with €1270 (933–1610) for target nonattainment. Similarly, for total pathology costs, these numbers were €245 (66.6–834) for target attainment and 699 (328–703) for target nonattainment. Fixed ICU costs varied marginally according to the need for extracorporeal circulation and disease severity.
On presenting daily costs (Table 3 ), the aforementioned trends in total, microbiology, and variable costs were no longer observed. Costs associated with the transfusion of blood products were significantly higher in patients with target attainment (€90.60 and €12.70, P = 0.023), with a strong trend toward higher RRT costs in this patient category (0 and 0, P = 0.063). For patients who received RRT, the daily RRT costs were €90.1 (63.6–109) for those with target attainment when compared with €29.3 (24.9–33.8) for those with target nonattainment. The daily costs for pathology were €7.97 (2.85–39.6) for patients with target attainment when compared with €54.1 (39.2–117) for those with target nonattainment, on examining only those patients who incurred costs in this category.
TABLE 3. -
Daily ICU Costs Split by Cost Category
Categories
Target Attainment (N = 50)Target Nonattainment (N = 29)
P
Median
IQR1–IQR3
Median
IQR1–IQR3
Mann–Whitney U Test
Bootstrapped t test
Total
€2.680
2.420–3.290
€2.700
2.930–3.370
0.95
0.95
Variable
€1.080
889–1.630
€1.090
783–1.570
0.76
0.80
Fixed admission
€1.790
1700–1.880
€1.820
1.760–1.890
0.33
0.90
Consultation
€98.3
57–163
€111
68–242
0.42
0.25
RRT
€0
0–0
€0
0–0
0.06
<0.01
Bedside procedures
€48.9
19.1–98.1
€52.9
28.9–83.9
0.85
0.97
Laboratory diagnostics
€181
141–232
€164
134–217
0.38
0.66
Radiology
€93.2
33.4–167
€86.7
49.5–277
0.45
0.17
Transfusion
€90.6
14.4–208
€12.7
0–91.4
0.02
0.16
Microbiology
€149
85.3–223
€110
83.7–154
0.17
0.48
Pathology
€0
0–0
€0
0–0
0.79
0.80
Surgery
€29.2
0–124
€42.7
0–159
0.67
0.31
Bold P values are significant (lower than 0.05).
Cost Determinants
Table 4 describes models 1 and 2. Model 1 shows that target attainment is not a determinant of cost (OR 1.05; CI 0.90–1.23, P = 0.54). However, RRT during ICU admission was a clear predictor of cost (OR, 1.84; CI 1.50–2.27, P = 0.001). On omitting RRT from linear regression (model 2), target attainment was still not a predictor for total costs (OR 1.18 CI 0.98–1.42, P = 0.09). The omission of RRT resulted in a significantly worse prediction of total costs (P < 0.001).
TABLE 4. -
Multivariate Linear Regression With Logarithmic Transformation on Total ICU Costs
Model 1
OR (95% CI)
P
Model 2
OR (95% CI)
P
Target attainment 1.05 (0.90–1.23)
0.54
Target attained
1.18 (0.98–1.42)
0.09
Female sex
0.93 (0.80–1.09)
0.39
Female sex
0.95 (0.79–1.14)
0.59
Age
1.00 (0.99–1.00)
0.46
Age
0.99 (0.99–1.00)
0.06
SOFA score
1.00 (0.98–1.02)
0.98
SOFA score
1.01 (0.98–1.03)
0.47
RRT
1.84 (1.50–2.27)
<0.001
McFadden R2 = 0.10
McFadden R2 = 0.52
Bold P values are significant (lower than 0.05).
Only RRT therapy during admission proved to be an independent cost determinant for high total ICU costs. Our most important factor, target attainment , was not a factor in both multivariate analyses.
DISCUSSION
Contrary to our hypothesis, we observed a trend toward higher total ICU costs when antibiotic target attainment was achieved. Costs for blood product transfusions were significantly higher in patients with target attainment . We observed that these trends disappeared when correcting for the LOS, except for the aforementioned costs for transfusion and RRT. The latter is the most influential cost determinant, which mainly explains the differences in costs between the 2 groups.
To explain this paradoxical finding of higher ICU costs in patients with target attainment , we further analyzed both patient groups. There were some major patient differences between both groups, with the most relevant being ICU LOS, disease severity scores, RRT use, and the use of blood products. Target attainment seems to be a surrogate marker for patient illness because it relates to a decrease in renal function, greater ICU LOS, a higher illness severity score, and therefore an increased prevalence of multiple organ failure. These risk factors of target attainment have also been confirmed in previous research.16–18 t tests revealed distorted significance: a small number of outliers can easily lead to a low P value, such as in RRT and pathology costs.
Target attainment was not a significant cost driver for the total ICU costs in our analysis. On examining a model excluding RRT, target attainment was still not a predictor. In our model, the most important determinant of costs was RRT. This variable seemed to be solely responsible for explaining high costs, considering that most patients receiving RRT achieved target attainment . Moreover, on visually inspecting the data in Figure 1 , no patient receiving RRT had a LOS of less than 20 days. In Supplemental Digital Content 1 (see Table , https://links.lww.com/TDM/A490 P = 0.126) of target attainment in the univariate analysis did not translate in the multivariate analysis. This can be explained by the aforementioned relationship between RRT and target attainment .
FIGURE 1.: Correlation plot of ICU LOS and total ICU costs split by target attainment and use of renal replacement therapy. Both axes are log10 transformed.
The ICU contributes to a major part of total hospital costs, with relatively high costs per admission day when compared with other clinical wards.19,20 17,21 22 15,23 20,24,25 18,26,27 Table 1 , Supplemental Digital Content 2 , https://links.lww.com/TDM/A491
TDM is considered to be cost-effective for some antibiotics, including glycopeptides and aminoglycosides, because it might prevent costly adverse events.12 28 target attainment after TDM results in a difference in ICU costs and therefore assessing whether TDM is cost-effective for beta-lactam antibiotics.29
This study has a few limitations, mostly attributed to the observational nature of our data. First, we only examined the costs of ICU admittance. No costs could be analyzed from subsequent hospital wards. Second, as this study was performed in a tertiary university hospital, external validity should be considered when interpreting these results. The total or variable costs might not well translate to a different case-mix or less academic patient care facility. Finally, we were unable to extract all costs related to the patients; for example, costs of all medications included in the standard daily admission costs. An inquiry showed that these account for 7% of the standard daily admission costs. However, most beta-lactam antibiotic patents have expired and are therefore relatively inexpensive when compared with other ICU costs.
This is the first study to describe the relationship between ICU costs and beta-lactam antibiotic target attainment . Although we did not determine a significant difference in total costs, interesting trends toward significance were identified. Further research in an ICU population at risk of target nonattainment is needed to assess the efficacy and cost-effectiveness of TDM of beta-lactam antibiotics.
CONCLUSIONS
Target attainment for beta-lactam antibiotics shows a trend toward higher total costs in ICU patients. As RRT is the major cost determinant for total ICU costs, differences in these costs are mainly explained by the fact that patients achieving beta-lactam target attainment are more likely to receive RRT.
ACKNOWLEDGMENTS
The authors thank Dr. Wim Rietdijk for his help with the statistical analysis. The authors also want to acknowledge Ruben Goedhart and Richard Hobbel for their work in extracting cost-related data. The authors would also like to thank Stanley Hau for his help in matching the cost-related data.
REFERENCES
1. Leache L, Aquerreta I, Aldaz A, et al. Clinical and economic impact of clinical pharmacist interventions regarding antimicrobials on critically ill patients. Res Social Adm Pharm. 2020;16:1285–1289.
2. Lipman J, Brett SJ, De Waele JJ, et al. A protocol for a phase 3 multicentre randomised controlled trial of continuous versus intermittent β-lactam antibiotic infusion in critically ill patients with sepsis: Bling iii. Crit Care Resusc. 2019;21:63–68.
3. Vincent JL, Rello J, Marshall J, et al. International study of the prevalence and outcomes of infection in intensive care units. JAMA. 2009;302:2323–2329.
4. Muller AE, Huttner B, Huttner A. Therapeutic drug monitoring of beta-lactams and other antibiotics in the intensive care unit: which agents, which patients and which infections? Drugs. 2018;78:439–451.
5. Levy MM, Evans LE, Rhodes A. The surviving sepsis campaign bundle: 2018 update. Intensive Care Med. 2018;44:925–928.
6. Abdul-Aziz MH, Lipman J, Mouton JW, et al. Applying pharmacokinetic/pharmacodynamic principles in critically ill patients: Optimizing efficacy and reducing resistance development. Semin Respir Crit Care Med. 2015;36:136–153.
7. McKinnon PS, Paladino JA, Schentag JJ. Evaluation of area under the inhibitory curve (auic) and time above the minimum inhibitory concentration (t>mic) as predictors of outcome for cefepime and ceftazidime in serious bacterial infections. Int J Antimicrob Agents. 2008;31:345–351.
8. Abdulla A, Dijkstra A, Hunfeld NGM, et al. Failure of
target attainment of beta-lactam antibiotics in critically ill patients and associated risk factors: a two-center prospective study (expat). Crit Care. 2020;24:558.
9. Roberts JA, Paul SK, Akova M, et al. Dali: defining antibiotic levels in intensive care unit patients: are current β-lactam antibiotic doses sufficient for critically ill patients? Clin Infect Dis. 2014;58:1072–1083.
10. Blandy O, Honeyford K, Gharbi M, et al. Factors that impact on the burden of escherichia coli bacteraemia: multivariable regression analysis of 2011–2015 data from west london. J Hosp Infect. 2019;101:120–128.
11. Ye ZK, Tang HL, Zhai SD. Benefits of therapeutic drug monitoring of vancomycin: a systematic review and meta-analysis. PLoS One. 2013;8:e77169.
12. Touw DJ, Neef C, Thomson AH, et al. Cost-effectiveness of therapeutic drug monitoring: a systematic review. Ther Drug Monit. 2005;27:10–17.
14. Zorgautoriteit N. Nza zorgproductapplicatie: nederlandse Zorgautoriteit. Available at:
https://zorgproducten.nza.nl/ . Accessed July 22, 2020.
15. CB Statistiek. Jaarmutatie consumentenprijsindex: centraal Bureau voor statistiek updated August 6, 2020. Available at:
https://opendata.cbs.nl/#/CBS/nl/dataset/70936ned/table?dl=5B52 . Accessed September 4, 2020.
16. Dhaese SAM, Thooft ADJ, Farkas A, et al. Early
target attainment of continuous infusion piperacillin/tazobactam and meropenem in critically ill patients: a prospective observational study. J Crit Care. 2019;52:75–79.
17. Karabatsou D, Tsironi M, Tsigou E, et al. Variable cost of icu care, a micro-costing analysis. Intensive Crit Care Nurs. 2016;35:66–73.
18. Aung YN, Nur AM, Ismail A, et al. Characteristics and outcome of high-cost icu patients. Clinicoecon Outcomes Res. 2019;11:505–513.
19. Cooke CR. Economics of mechanical ventilation and respiratory failure. Crit Care Clin. 2012;28:39–55.
20. Moerer O, Plock E, Mgbor U, et al. A German national prevalence study on the cost of intensive care: an evaluation from 51 intensive care units. Crit Care. 2007;11:R69.
21. Tan SS, Bakker J, Hoogendoorn ME, et al. Direct cost analysis of intensive care unit stay in four european countries: Applying a standardized costing methodology. Value Health. 2012;15:81–86.
22. Tan SS, Hakkaart-Van Roijen L, Al MJ, et al. Review of a large clinical series: a microcosting study of intensive care unit stay in The Netherlands. J Intensive Care Med. 2008;23:250–257.
23. Statistiek CB. Consumentenprijzen: centraal Bureau voor statistiek. Available at:
https://www.cbs.nl/nl-nl/cijfers/detail/83131NED . Accessed July 28, 2020.
24. McLaughlin AM, Hardt J, Canavan JB, et al. Determining the economic cost of icu treatment: a prospective “micro-costing” study. Intensive Care Med. 2009;35:2135–2140.
25. Reardon PM, Fernando SM, Van Katwyk S, et al. Characteristics, outcomes, and cost patterns of high-cost patients in the intensive care unit. Crit Care Res Pract. 2018;2018:5452683.
26. Koster-Brouwer ME, Klein Klouwenberg P, Pasma W, et al. Critical care management of severe sepsis and septic shock: a cost-analysis. Neth J Crit Care. 2016;24:12–18.
27. Rossi C, Simini B, Brazzi L, et al. Variable costs of icu patients: a multicenter prospective study. Intensive Care Med. 2006;32:545–552.
28. Imani S, Buscher H, Day R, et al. An evaluation of risk factors to predict target concentration non-attainment in critically ill patients prior to empiric beta-lactam therapy. Eur J Clin Microbiol Infect Dis. 2018;37:2171–2175.
29. Abdulla A, Ewoldt TMJ, Hunfeld NGM, et al. The effect of therapeutic drug monitoring of beta-lactam and fluoroquinolones on clinical outcome in critically ill patients: the dolphin trial protocol of a multi-centre randomised controlled trial. BMC Infect Dis. 2020;20:57.
