Clinical and economic impact of ventilator-associated pneumonia in intensive care units in Japan

Summary

Ventilator-associated pneumonia (VAP) has a considerable impact on both clinical outcomes and healthcare costs. This study compared 40 patients having VAP with 40 matched controls from a Japanese ICU dataset. Patients with VAP experienced significantly longer ICU and hospital stays, fewer ventilator-free days, and a higher incidence of tracheostomies. VAP cases also required more broad-spectrum antimicrobials, leading to an additional cost of approximately USD 24,410 per case. These results highlight the importance of implementing effective infection control strategies to mitigate VAP's clinical and economic consequences.

Introduction

Hospital-acquired infections (HAIs) affect an estimated 4.5% of hospitalised patients, or approximately 1.7 million per year, resulting in approximately 99,000 deaths in the United States [1]. Infection is independently associated with an increased risk of in-hospital death in intensive care unit (ICU) patients [2]. In 2012, the incidence rate of ventilator-associated pneumonia (VAP) in the United States was 0.9 and 2.0 cases per 1,000 ventilator days in medical and surgical ICUs, respectively [3]. VAP is also important for appropriate antimicrobial use, as up to half of ICU patients receiving antimicrobials have VAP or hospital-acquired pneumonia [2]. Furthermore, VAP requires additional days of ventilation and often requires tracheostomy, leading to extended ICU and hospital stays and increased healthcare costs [4,5].

The additional healthcare costs associated with VAP vary widely, depending on each country's medical insurance and reimbursement systems. However, only one single-centre report has been published on VAP cost in Japan [6]. Moreover, foreign reports on healthcare costs are not generalisable, especially in Japan, which has a universal health insurance system covering all citizens [7]. Therefore, we examined the clinical and economic effects of VAP in Japan using the multicentre Diagnostic Procedure Combination/Per-Diem Payment System (DPC) data to develop efficient VAP infection control strategies.

Methods

Study design and setting

This retrospective case-control study used Japanese DPC data. These data were obtained from 1,730 hospitals that agreed to participate in the Ministry of Health, Labour and Welfare's Policy Science Research Project. Patients were selected using a matching method to compare prognoses and healthcare costs. The DPC database is a classification scheme for acute care hospital inpatients in Japan. DPC data provide inpatient clinical information, including unique identification data, age, sex, dates of admission and discharge, discharge outcome status, major diseases that consumed the most medical resources, comorbidities associated with the International Classification of Diseases (10^th revision) codes, and data linked with a lump-sum payment system [8].

Case definition

Inclusion criteria were cases in which ICU management fees were charged between 1 April 2018 and 31 March 2020. Thus, 829,498 patients were assessed for eligibility. Exclusions included admission prior to 1 April 2018, age <19 years, multiple ICU admissions per hospitalisation, missing Sequential Organ Failure Assessment (SOFA) score at ICU admission, and mechanical ventilation for ≤2 days. Patients diagnosed with VAP were identified in the DPC's lump-sum payment system data. Cases without culture tests on or after the 5th day of ICU admission were considered control candidates with no HAIs. Four variables used to match controls with cases included sex, age grouped in 10-year increments, diagnosis leading to hospitalisation, and SOFA score at ICU admission. The control group was selected based on an exact match of matching items.

Data collection

Information on sex, age, diagnosis, SOFA score at ICU admission, body mass index (BMI), and ambulance transport to the hospital were collected. We also collected information on ICU and hospital length of stay, ventilator-free days (VFD), tracheotomy, in-hospital mortality, discharge route, and Clostridioides difficile infection during hospitalisation for both groups to understand the effect of VAP on prognosis. Finally, we recorded the number of days of antimicrobial therapy before, during, and after ICU admission, focusing on anti-methicillin-resistant Staphylococcus aureus (anti-MRSA), anti-Pseudomonas aeruginosa, and antifungal drugs.

Calculation of costs

Healthcare costs from ICU admission to discharge were recorded for each patient based on Japan's fee-for-service system, established by the Ministry of Health, Labour and Welfare, and DPC data [7]. A detailed breakdown of the medical costs in the DPC system was divided into five categories: ICU fees, basic hospitalisation fees, drugs, medical materials, and other costs. Other costs include everything else billed as medical expenses for insurance claims, excluding the other four categories. Additionally, case-specific daily medical costs were calculated. The exchange rate used was USD 1 to JPY 110 as of 1 April 2019. The additional healthcare costs incurred due to VAP were calculated from the difference between the median costs of VAP and control cases.

Statistical analysis

The χ² and Mann–Whitney U tests were used to analyse categorical and continuous variables, respectively. No adjustments were made to compare both groups because they were matched. All statistical analyses were performed using SPSS version 26 (IBM Corp., Armonk, N.Y., USA).

Results

During the study period, 829,498 patients were admitted to the ICU and charged for management. A total of 810,889 patients were excluded from the analysis: 22,913 admitted before April 1, 2018; 25,667 aged <19 years; 54,335 with multiple ICU admissions; 602,658 without a recorded SOFA score; and 105,316 on mechanical ventilation for ≤2 days. This left 18,609 patients for analysis, including 97 diagnosed with VAP and 6,289 pooled as controls due to no culture test results after the 5th day of ICU admission. Of these controls, 40 patients matched to the VAP cases by age, sex, diagnosis, and SOFA score at ICU admission were selected as the control group, and 40 out of the 97 patients with VAP matched to this control group were selected for the VAP group.

Table I shows the two groups' clinical characteristics. No significant differences were observed in matching criteria between groups. Furthermore, no significant differences were observed in BMI or ambulance transport between groups.

Table I.

Clinical characteristics of the study population by group

	VAP group (n=40)	Control group (n=40)	P value
Characteristics
Age	68 (58–78)	70 (58–76)	0.718
Sex			>0.999
Male	29 (73)	29 (73)
Female	11 (27)	11 (27)
BMI (kg/m2)	23.0 (20.3–24.5)	22.6 (19.2–25.4)	0.898
Diagnosis			>0.999
Arrhythmia	13	13
Acute myocardial infarction	4	4
Dissecting aortic aneurysm	3	3
Angina pectoris	3	3
Respiratory failure	2	2
Head injury	2	2
Sepsis	2	2
Valvular disease of the heart	2	2
Trauma	2	2
Subarachnoid haemorrhage	1	1
Electrolyte disturbance	1	1
Epilepsy	1	1
Intestinal obstruction	1	1
Oesophageal cancer	1	1
Heart failure	1	1
Aortic aneurysm	1	1
Ambulance transport	35 (87)	30 (75)	0.152
SOFA score	9 (7–11)	9 (7–11)	>0.999
Outcomes
Length of ICU stay (days)	13 (8–14)	5 (3–7)	<0.001
Length of hospital stay (days)	50 (29–68)	17 (11–23)	<0.001
Ventilator-free days (days)	7 (0–18)	23 (0–25)	0.005
Tracheostomy	19 (47)	2 (5)	<0.001
Clostridioides difficile infection	3 (0)	0 (0)	0.120
In-hospital mortality	8 (20)	13 (32)	0.204
Discharge route
Transfer to another hospital	21 (52)	10 (25)
Home	11 (27)	17 (42)	0.028
Days of antimicrobial therapy
Before ICU admission	0 (0–0)	0 (0–0)	0.296
Anti-Pseudomonas	0 (0–0)	0 (0–0)	0.317
Anti-MRSA	0 (0–0)	0 (0–0)	>0.999
Antifungal drugs	0 (0–0)	0 (0–0)	>0.999
Other	0 (0–0)	0 (0–0)	0.079
While in the ICU	12.5 (7–16)	4 (2.2–6)	<0.001
Anti-Pseudomonas	3 (0.5–8.0)	0 (0–0.5)	<0.001
Anti-MRSA	0.5 (0–2.5)	0 (0–0)	<0.001
Antifungal drugs	0 (0–0)	0 (0–0)	0.296
Other	5 (3–9)	3 (1–5)	0.005
After ICU discharge	15 (4.5–29.5)	1 (0–5)	<0.001
Anti-Pseudomonas	6 (0–14.5)	0 (0–0)	<0.001
Anti-MRSA	0 (0–2.5)	0 (0–0)	<0.001
Antifungal drugs	0 (0–0)	0 (0–0)	0.042
Other	1 (0–9)	0 (0–3)	0.031

Values are presented as median (IQR), n, or n (%).

Abbreviations: BMI, body mass index; IQR, interquartile range; SOFA, Sequential Organ Failure Assessment; ICU, intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus; VAP, ventilator-associated pneumonia.

Compared with the control group, the VAP group had significantly longer ICU stays (13 [8–14] days vs. 5 [[3], [4], [5], [6], [7]] days; P<0.001), longer hospital stays (50 [29–68] days vs. 17 [11–23] days; P<0.001), fewer VFD (7 [0–18] days vs. 23 [0–25] days, P=0.005), and more tracheostomies (19/40 vs. 2/40; P<0.001). However, in-hospital mortality did not differ between groups (8/40 vs. 13/40; P=0.204). VAP and control groups had significantly different numbers of patients transferred to long-term care hospitals (21/40 vs. 10/40, P=0.028).

Few cases of antimicrobial use occurred before ICU admission, with no significant difference between the two groups (P=0.296). A higher proportion of patients in the VAP group received antimicrobials during and after ICU stay (P<0.001), especially anti-MRSA and anti-P. aeruginosa drugs (P<0.001).

Table II shows the median medical costs per person in the hospital for the two groups. The VAP group had significantly higher median total hospital medical costs (USD 51,099) than the control group (USD 26,689), with a difference of USD 24,410, representing additional medical costs associated with VAP in Japan. ICU fees were the highest, followed by basic hospitalisation fees, drugs (including antibiotics for VAP), and medical material costs. The VAP group had lower median daily medical costs than the control group (USD 1,049 vs. USD 1,525; P=0.003).

Table II.

Median medical costs per person in the hospital

	VAP group (n=40)	Control group (n=40)	P value
Total	51,099 (33,741–67,539)	26,689 (15,351–47,126)	<0.001
ICU fees	15,915 (11,166–17,188)	6,937 (4,491–9,281)	<0.001
Basic hospitalisation fees	9,702 (6,348–12,491)	4,107 (2,700–5,284)	<0.001
Drug costs	4,106 (2,242–7,311)	1,474 (694–4,877)	0.001
Medical material costs	2,734 (746–8,002)	1,259 (237–8,990)	0.166
Other costs	24,510 (12,122–38,156)	12,321 (6,557–36,462)	0.014
Daily costs	1,049 (752–1,541)	1,525 (937–2,267)	0.003

Note: Values are presented as median (IQR) USD/case.

Abbreviations: ICU, intensive care unit; IQR, interquartile range; VAP, ventilator-associated pneumonia.

Discussion

Clinical impact due to VAP

This study found that VAP resulted in adverse clinical outcomes, such as shorter VFD and prolonged ICU and hospital stays, as previously reported [4]. Moreover, we assumed a high frequency of post-intensive care syndrome (PICS) in the VAP group due to the high number of tracheostomies and transfers. In contrast, there was no significant difference in in-hospital mortality rates between the two groups. It has been demonstrated that VAP is not associated with an increased risk of mortality in specific patient populations such as those with traumatic brain injury [9]. However, the overall attributable mortality rate of ventilator-associated pneumonia is 13% [4]. Therefore, the number of cases in this study was insufficient to draw conclusions about the impact of VAP on in-hospital mortality rates.

Next, this study compared antimicrobial therapy days before, during, and post-ICU admission to general wards. The use of antimicrobials before ICU admission did not differ between the groups. However, a higher proportion of patients in the VAP group received antimicrobials in the ICU and after ICU discharge, especially anti-MRSA and anti-P. aeruginosa drugs. Additionally, 39 out of 40 VAP cases underwent culture testing before initiating new antibiotic therapy in the ICU. Therefore, the increased use of broad-spectrum antimicrobials in the VAP group in the ICU may stem from empirical treatment or treatment based on culture susceptibility results. As mentioned above, the VAP group was more likely to have implanted medical devices, such as tracheal cannulas. Thus, the VAP group faced a higher risk of HAIs after ICU discharge and was more likely to receive broad-spectrum antimicrobials due to their history of antimicrobial administration in the ICU. The development of VAP has increased the use of broad-spectrum antimicrobial agents, potentially contributing to the spread of multidrug-resistant bacteria and C. difficile infections.

Economic impact due to VAP

This study estimates VAP-related additional medical costs in Japan. Each VAP case had median and average additional medical costs of USD 24,410 and USD 26,216. This is the first multicentre study to report additional medical costs incurred by VAP in Japan. A previous single-centre study in Japan found that the additional healthcare costs for patients with VAP averaged USD 34,884 [6]. After adjusting for the previous study's exchange rate (USD 1=JPY 100), the cost changed to USD 31,712. Therefore, when adjusted for exchange rates and compared with each other's averages, the difference between this study and the previous single-centre study in Japan was smaller.

Analysing medical expenses by item, ICU and basic hospitalisation fees are the major factors generating additional medical costs rather than direct drug costs (e.g. antibiotic costs for VAP). In the VAP group, both ICU and hospital stays were longer, which may have contributed significantly to the increased medical costs. In addition, the daily medical costs per person in the hospital were lower in the VAP group. This suggests that patients who develop VAP often cannot be discharged to their homes; therefore, despite having a low need for acute medical care, it takes a long time for them to be transferred. These results are consistent with those of a previous single-centre study in Japan [6].

Limitations

Our study had some limitations. First, the diagnostic criteria-selected VAP incidence rate of patients was approximately 0.5% (0.54 cases per 1,000 ventilator days) of the cases included in the analysis, which is low compared to previous reports. Under the DPC system, the focus is on ensuring that the medical expenses claimed by hospitals are reimbursed by insurance organisations. Consequently, VAP is not necessarily registered separately from simple pneumonia. Therefore, the number of VAP cases registered under the DPC system tends to be low. Second, controls were selected if no culture tests were performed on or after the 5th day of ICU admission. This may have increased the risk of early death in the control group. Thus, in the control group, the in-hospital mortality rate may have been overestimated and medical costs underestimated. Third, we could not evaluate the sputum or blood culture results because the Japanese DPC data did not include these results. Therefore, we could not determine whether a validly attributable organism was detected in the VAP cases or whether appropriate antimicrobial agents were administered. Fourth, the Japanese DPC data did not include the diagnosis date, as diagnoses were recorded per hospitalisation unit. Hence, we could not analyse the duration from the start of mechanical ventilation to VAP diagnosis. Fifth, the estimated additional medical costs may include indirect VAP costs, as they include all hospitalisation costs for patients who developed VAP. Other complications may have contributed to the high medical costs in the VAP group. However, assessing the effects of VAP alone is challenging because it can both cause and result from complications, complicating the qualitative assessment of additional medical costs. Moreover, the DPC did not include post-discharge data. Finally, this was a retrospective study.

Conclusion

To our knowledge, this is the first study to examine VAP's clinical and economic effects in Japanese ICUs using multicentre DPC data. VAP shortens VFD and increases the frequency of tracheostomies and PICS. Therefore, patients with VAP cannot be easily discharged. VAP also increased the use of broad-spectrum antimicrobials during ICU admission and after discharge. Additionally, VAP prolongs ICU and hospital stays and incurs additional medical costs (approximately USD 24,410 per case) under a universal healthcare insurance system that covers all citizens. VAP, which is an HAI, worsens patient outcomes and generates significant additional medical costs. Therefore, appropriate infection control strategies should be implemented, particularly for VAP.

Author contributions

Taikan Nanao was involved in the conceptualisation, formal analysis, methodology, and writing of the manuscript’s first draft. Koichi Benjamin Ishikawa was involved in data curation, validation, writing, reviewing, and editing. Shunya Ikeda was involved in the methodology. Tsutomu Yamazaki was involved in supervision. All authors have read and approved the final version of the manuscript.

Ethics approval and informed consent

Institutional approval was obtained from the International Health and Welfare University Graduate School Ethics Review Board (Approval number 20-Ig-132). The requirement for informed consent was waived because this was a retrospective analysis of anonymised data.

Declaration of Generative AI and AI-assisted technologies in the writing process

None.

Funding

This study did not receive any specific grants from funding agencies in the public, commercial, or non-profit sectors.

Conflict of interest statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.