Potential for Cost-Savings in the Care of Hospitalized Low-Risk Community-Acquired Pneumonia Patients in China

Volume 12 Number 1 2009 VALUE IN HEALTH Potential for Cost-Savings in the Care of Hospitalized Low-Risk Community-Acquired Pneumonia Patients in China Qing-tao Zhou, MD, Bei He, BSMed, Hong Zhu, BSMed Peking University Third Hospital, Beijing, China ABSTRACT Background: The cost of treating community-acquired pneumonia (CAP) in China is a heavy economic burden for the society. Objective: To investigate the costs of hospitalization of low-risk CAP patients and how hospitalization costs can be reduced through proper usage of hospital resources. Methods: Two hundred thirty-six patients with low-risk CAP who were hospitalized between January 2000 and December 2005 in a 1161-bed tertiary care teaching hospital were included in a retrospective cohort study. Their hospitalization costs and antibiotic therapy were analyzed. General linear model was utilized to determine correlative variables associated with total hospital costs. Results: The median length of hospital stay was 12 days and the median duration of intravenous (IV) antibiotic therapy was 10 days, they were correlated significantly (P = 0.000, r = 0.81). The median total hospital cost was $556.50 (mean $705.60), of which 48.9% was for drugs, 21.9% for laboratory tests, 8.6% for radiology, 6.5% for medical staff, 6.3% for hospital beds, and 5.3% for examination. General linear model analysis determined that duration of IV antibiotic therapy, Pneumonia Severity Index class, age, and initial empirical antibiotic therapy failure were correlative factors of total hospital costs. Pathogens were identified in 106 patients (44.9%), Mycoplasma pneumoniae was the most common pathogen (19.9%), followed by Streptococcus pneumoniae (8.5%), and Haemophilus influenza (5.5%). The majority of patients accepted initial empirical b-lactam (37.3%) or fluoroquinolone (30.9%) monotherapy, the empirical treatment failure rates were 20.5% and 5.5%, respectively. Conclusions: Efforts to reduce duration of IV antibiotic therapy will have the most profound effect on reducing total hospital costs of low-risk CAP. The atypical pathogens should be considered for initial empirical antibiotics in low-risk CAP therapy in China. Keywords: antibiotic, community-acquired pneumonia, cost savings, pneumonia severe index. Introduction Community-acquired pneumonia (CAP) is an acute respiratory disease that may occur at any age. The cost of treating CAP is a heavy economic burden for the society. The annual hospitalization cost of CAP in adult was estimated to be over $8 billion in the United States [1] and $136.85 million in Spain [2]. The annual cost of CAP was estimated to be NZ$63 million in New Zealand, including direct medical costs of $29 million, direct nonmedical costs of $1 million, and lost productivity of $33 million [3]. In China, there is no available data about the annual cost of CAP, but the Ministry of Health reported the mean hospitalization costs of CAP was $575.3 in 2005, and it was much higher in tertiary hospitals which was $1137.3 [4]. In 1997, Fine et al. [5] developed the Pneumonia Severity Index (PSI), which divided CAP into five risk classes. The prediction rule identified three distinct risk classes (I, II, and III) of patients who were at sufficiently low risk for death and other adverse medical outcomes who physicians could consider for outpatient treatment, which could reduce the social economic load. The PSI is now applied extensively worldwide. It has been reported that it can be used in Chinese CAP patients and has a good correlation with the disease severity criteria of the Chinese guidelines for diagnosis and treatment of CAP [6]. But in China, there were no strict indications of hospitalization in CAP guidelines [7], and also no definite information in medical insurance or care, so, some low-risk CAP patients were hospitalized, including patients of treatment failure in clinic. In this study, the clinical data and hospitalization costs of low-risk CAP patients were analyzed to determine the key factors in decreasing Address correspondence to: Bei He, Department of Respiratory Medicine, Peking University Third Hospital, Beijing, China 100083. E-mail: puh3_hb@bjmu.edu.cn 10.1111/j.1524-4733.2008.00410.x hospitalization costs. To our knowledge, there is no previous study about low-risk CAP in China. Materials and Methods Materials We retrospectively analyzed CAP patients during the period between 2000 and 2005 in Peking University Third Hospital, which is a nonprofit, tertiary care teaching hospital with 1161 beds in Beijing, China. The diagnosis of pneumonia was made according to the guidelines of CAP diagnosis and treatment made by the Respiratory Society of the Chinese Medical Association [7]. This included the presence of a new infiltration on chest radiographs and at least one of the following: documentation of a new cough with or without sputum production or exacerbation of chronic respiratory diseases with purulent sputum production with or without chest pain; documented fever ( 37.3 C); auscultatory findings on pulmonary examination and/or evidence of pulmonary consolidation; and white blood cells (WBC) count >10 10 9 /L or leucopenia (WBC < 4 10 9 /L). Patients were classified according to the PSI. Classes I, II, and III were considered low-risk groups. Low-risk CAP patients who were at least 18 years of age were included in the study. Patients were excluded if they had a positive titer of antibodies to the human immunodeficiency virus, had been hospitalized within 7 days before the current admission, had been transferred from another acute care hospital, had known or suspected tuberculosis or other infection at baseline caused by viruses or fungi, or their medical records were not included all variables of PSI score calculation. PSI class IV and class V patients were also excluded. Methods This study was retrospective in design. The following clinical data were collected: sex, age, source of payment for medical care 40 2008, International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 1098-3015/09/40 40 46

Cost Savings in Low-Risk CAP 41 (public medical care, national health insurance, self-pay medical care), comorbid conditions (neoplastic disease, chronic obstructive pulmonary disease, cerebrovascular disease, diabetes, congestive heart failure, cirrhosis or other chronic liver disease, or chronic renal disease), PSI class, extent of lobar infiltration on chest radiograph, length of hospital stay, duration of intravenous (IV) antibiotic therapy, duration of IV antibiotic therapy after body temperature (<37.3 C) and respiratory symptoms (dyspnea, cough, and chest pain) improved, and whether initial empirical antibiotic therapy failed. Clinical data were collected from medical record after discharge. We also recorded total hospital costs, including pharmacy, radiology, and laboratory costs (including tests of blood, urine and stool, arterial blood gas, and bacterial cultures), examination costs (including lung function tests, electrocardiogram, and ultrasound), medical staff fees (including costs for physicians and nurses), bed or room fees, and other costs such as nebulization and oxygen therapy. All patients that our hospital costs database can supply were included in this study. The factors influencing total hospital costs (total of pharmacy, radiology, laboratory, examination, medical staff, and other costs) were analyzed. The definition of antibiotic therapy failure was no improvement after 72 hours of therapy or improvement followed by relapse [7]. All patients had at least one sputum culture (with presence of <10 squamous epithelial cells and >25 polymorphonuclear cells per 10 magnification field) and a serum test for IgM antibodies to Mycoplasma pneumoniae, Chlamydia pneumoniae, and Legionella pneumophila. Enzyme linked immunosorbent assay method was used in serum test. Sputum culture and serum test for antibodies of atypical pathogens were done within 48 hours of hospital admission. Statistical Analysis Data were extracted into a data file of SPSS (SPSS, Chicago, IL, USA). The initial analysis separated the cohort into three groups based on PSI class (classes I, II, and III) and costs were compared between groups separated by outcomes of initial empirical antibiotic therapy. Total hospital costs were also compared among groups separated by source of payment in each PSI class. Continuous data were compared using one-way analysis of variance for normally distributed data or nonparametric tests including the Mann Whitney U-test and the Kruskal Wallis H-test for nonnormally distributed data. Chi-square tests were used for comparison of count data. Spearman correlation analysis was used to determine correlation between duration of IV antibiotic therapy and length of hospital stay. General linear model was utilized to determine correlative variables of total hospital costs for the entire cohort including sex, age, comorbid conditions, extent of lobar infiltration on chest radiograph, initial empirical antibiotic therapy failure, severity of illness, duration of IV antibiotic therapy, and duration of IV antibiotic therapy after body temperature and respiratory symptoms improved, most of which were supported from previous research. Intensive Care Unit (ICU) admission and mortality also could influence total hospital costs of CAP, but only low-risk CAP patients were included in the current study, there was no ICU admission case and no patient died. All correlative variables were entered into the model as covariates at the same time. A P-value <0.05 was considered significant in all statistical analyses. Normally distributed continuous data (e.g., age) are expressed as mean SD and non-normally distributed continuous data (e.g., cost) are expressed as the median (25th to 75th percentile). SPSS version 12 for Windows software (SPSS, Chicago, IL) was used for statistical analysis. Results Demographic Data In this study, we evaluated 236 low-risk CAP patients, including 130 patients (55.1%) in class I, 71 patients (30.1%) in class II, and 35 patients (14.8%) in class III. In each year, from 2000 to 2005, there were 43, 68, 53, 19, 27, and 26 cases, respectively. The age of patients was 41.0 19.8 years old, with 144 males and 92 females. The age and percentage of patients who had bilateral or multiple lobar infiltration increased with increasing PSI risk class. Payments for medical care included national health insurance in 91 (38.6%), public medical care in 106 (44.9%), and self-pay medical care in 39 (16.5%) patients (Table 1). Clinical Data For all patients, the median length of hospital stay was 12 days, median duration of IV antibiotic therapy was 10 days. Length of hospital stay was significantly correlated with duration of IV antibiotic therapy (P = 0.000, r = 0.81). The median duration of IV antibiotic therapy after body temperature recovered and respiratory symptoms improved was 7 days (Table 1). Pathogens were identified in 106 patients (44.9%) with valid serum samples and sputum cultures. M. pneumoniae was the most common pathogen (47 cases, 19.9%), followed by Streptococcus pneumoniae (20 cases, 8.5%), and Haemophilus influenza (13 cases, 5%). Pathogens of low-risk CAP were listed in Table 2. Of the 54 patients with a bacterial pathogen, M. pneumoniae was identified in three cases whose sputum culture results showed Klebsiella pneumoniae, Citrobacter freundii, and S. pneumoniae, respectively. Table 3 displays the list of antibiotic regimens utilized to empirically treat low-risk CAP in these patients. The majority of patients received b-lactam (88 cases, 37.3%) or fluoroquinolone (73 cases, 30.9%) monotherapy, or a combination of a fluoroquinolone with a macrolide (42 cases, 17.8%). Eleven patients (4.7%) received monotherapy with a macrolide and seven patients (3.0%) received a b-lactam in combination with a macrolide. The remaining 15 patients (6.3%) received alternative regimens; but in all cases, a b-lactam, macrolide, or fluoroquinolone was utilized as at least one of the drugs in the regimen. Antibiotics of b-lactams, macrolides, and fluoroquinolones used in initial empirical antibiotic therapy are listed in Table 4. The initial empirical antibiotic therapy failed in 18 patients who received a b-lactam as monotherapy and 4 patients who received a fluoroquinolone as monotherapy. Three of seven cases receiving penicillin G as monotherapy failed, and two cases changed regimens to monotherapy with other b-lactams and one case changed to a fluoroquinolone monotherapy. The empirical therapy failed in 15 patients who received a b-lactam other than penicillin G as monotherapy; they changed regimens which contained at least one macrolide and fluoroquinolone. Of the 73 patients who received monotherapy with a fluoroquinolone, 4 cases failed and all changed regimens to monotherapy with b-lactams other than penicillin G. The treatment outcome was that 69.9% of patients were cured and 30.1% had remissions. No patients died. Costs of Hospitalization Total and departmental costs are depicted in Table 5. From 2000 to 2005, the median total hospital costs for low-risk CAP in our hospital was $556.50 (25th to 75th percentile, $359.10 to $889.00). The greatest percentage of total cost was allocated to pharmacy costs, particularly antibiotics, at 48.9%, followed

42 Zhou et al. Table 1 Demographic and clinical data for low-risk CAP patients and results for groups defined by PSI class Characteristics All patients (n = 236) Class I (n = 130) Groups separated by PSI class Class II (n = 71) Class III (n = 35) Age, year 41.0 19.8 29.6 9.2 47.5 18.6 70.2 14.6 0.000 Male gender 144 (61.0) 75 (57.7) 38 (53.5) 31 (88.6) 0.001 Sources of payment for medical care 0.661 Public medical care 106 (44.9) 64 (49.2) 29 (40.8) 13 (37.1) National health insurance 91 (38.6) 47 (36.2) 29 (40.8) 15 (42.9) Self-pay medical care 39 (16.5) 19 (14.6) 13 (18.4) 7 (20.0) Comorbid conditions Neoplasm 3 (1.3) 0 1 (1.4) 2 (5.7) 0.027 Cerebrovascular disease 3 (1.3) 0 1 (1.4) 2 (5.7) 0.027 COPD 3 (1.3) 0 0 3 (8.6) 0.000 Diabetes 2 (0.8) 1 (0.8) 1 (1.4) 0 0.75 Congestive heart failure 0 0 0 0 Renal disease 0 0 0 0 Liver disease 0 0 0 0 Bilateral or multiple lobars infiltration 65 (27.5) 29 (22.3) 17 (23.9) 19 (54.3) 0.001 on chest radiograph Length of hospital stay, d 12.0 (9.0 18.0) 12.0 (9.0 16.0) 13.0 (10.0 18.0) 13.0 (8.0 22.0) 0.136 Duration of IV antibiotic therapy, d 10.0 (8.0 14.0) 10.0 (8.0 13.3) 10.0 (8.0 16.0) 11.0 (7.0 21.0) 0.602 Duration of IV antibiotic therapy 7.0 (5.0 10.0) 7.0 (5.0 10.0) 7.0 (5.0 10.0) 7.0 (3.0 9.0) 0.547 after body temperature and respiratory symptoms improved, d Initial empirical antibiotic therapy failure 22 (9.3) 13 (10.0) 8 (11.3) 1 (2.9) 0.347 Admission year 0.914 2000 43 (18.2) 24 (18.5) 12 (16.9) 7 (20.0) 2001 68 (28.8) 40 (30.8) 17 (23.9) 11 (31.4) 2002 53 (22.5) 29 (22.3) 16 (22.5) 8 (22.9) 2003 19 (8.1) 8 (6.2) 7 (9.9) 4 (11.4) 2004 27 (11.4) 14 (10.8) 11 (15.5) 2 (5.7) 2005 26 (11.0) 15 (11.5) 8 (11.3) 3 (8.6) Note:Age is presented as mean SD. Length of hospital stay, duration of IV antibiotic therapy, and duration of IV antibiotic therapy after body temperature and respiratory symptoms improved are presented as median (25th to 75th percentile). Other data are presented as Number (%). PSI, Pneumonia Severity Index; COPD, chronic obstructive pulmonary disease; IV, intravenous; CAP, community-acquired pneumonia. P-value by laboratory costs (21.9%), radiology (8.6%), medical staff (6.5%), hospital beds (6.3%), examination (5.3%), and others (3.3%). The more severe the CAP, the more it cost to treat: PSI class I, $500.60 (25th to 75th percentile, $318.90 to $732.70); class II, $631.00 (25th to 75th percentile, $399.90 to $1002.60); and class III, $908.60 (25th to 75th percentile, $475.30 to $1471.60). The differences between PSI classes were mainly because of pharmacy costs. Patients with successful initial empirical antibiotic therapy had lower total hospital costs than those who had failed initial therapy (median $525.10 vs. $744.30, P = 0.041). There were no significant differences in total hospital costs among groups separated by source of payment in each PSI class, P-value in PSI 1, PSI 2, and PSI 3 were 0.206, 0.189, and 0.271, respectively. Table 2 Pathogens identified in 236 low-risk CAP patients Pathogens Number of cases Proportion (%) Mycoplasma pneumoniae 47 19.9 Streptococcus pneumoniae 20 8.5 Haemophilus influenza 13 5.5 Acinetobacter 8 3.4 Chlamydia pneumoniae 5 2.1 Moraxella catarrhalis 3 1.3 Legionella pneumophila 3 1.3 Klebsiella pneumoniae 3 1.3 Enterobacter cloacae 2 0.8 Pseudomonas aeruginosa 2 0.8 Serratia marcescens 1 0.4 Staphylococcus epidermidis 1 0.4 Citrobacter freundii 1 0.4 CAP, community-acquired pneumonia. Radiology costs for patients who were in PSI risk class II were significantly higher than other patients, which was mainly caused by chest computed tomography (CT) scans. Patients in PSI class II (0.65 times/patient) had more chest CT scans than those who were in class I (0.35 times/patient) and class III (0.63 times/ patient; P = 0.002). The age of patients in class II was 47.5 18.6 years old, an age range at which lung cancer and other diseases are so common that more attention is required and chest CT scans were utilized more frequently for differential diagnosis. Independent variables included in the general linear model predicting total hospital costs are shown in Table 6. Because length of hospital stay was decided by duration of IV antibiotic therapy (P = 0.000, r = 0.81), so the former could not be an independent variable. Duration of IV antibiotic therapy, PSI class, age, and initial empirical antibiotic therapy failure were significantly associated with higher total costs. Discussion In the current study, we analyzed hospitalization costs and antibiotic therapy for low-risk CAP patients admitted to a large urban nonprofit university hospital from 2000 to 2005. The duration of IV antibiotic therapy, PSI class, age, and initial empirical antibiotic therapy failure were independently associated with total hospital costs. Our finding that duration of IV antibiotic therapy were very important in influencing length of hospital stay and total hospital costs for CAP patients supports other studies [8 10]. IV antibiotic therapy is a very important mode of treatment for CAP patients and has a quick response and good outcome. When the

Cost Savings in Low-Risk CAP 43 Table 3 Antibiotic regimens and duration of IV antibiotic therapy Empirical antibiotic regimens Number (%) (n = 236) All patients Duration of IV antibiotic therapy (d) Patients with empirical antibiotic therapy failure Number (failure rate) Duration of IV antibioti therapy (d) b-lactam monotherapy 88 (37.3) 11.0 (8.0 15.8) 18 (20.5) 13.0 (10.0 17.8) Penicillin G monotherapy 7 (3.0) 11.0 (10.0 17.0) 3 (42.9%) 17.0 (11.0 20.0) Other b-lactam monotherapy 81 (34.3) 11.0 (8.0 15.0) 15 (18.5%) 13.0 (10.0 17.0) Fluoroquinolone monotherapy 73 (30.9) 10.0 (7.0 12.0) 4 (5.5%) 10.0 (7.5 11.0) Macrolide plus fluoroquinolone 42 (17.8) 11.0 (8.8 16.0) 0 Macrolide monotherapy 11 (4.7) 8.0 (2.0 12.0) 0 b-lactam plus macrolide 7 (3.0) 13.0 (10.0 17.0) 0 b-lactam plus fluoroquinolone 4 (1.7) 9.5 (6.3 21.0) 0 Fluoroquinolone plus clindamycin 4 (1.7) 12.0 (8.25 22.5) 0 b-lactam plus clindamycin 3 (1.3) 13.0 (5.0 13.0) 0 b-lactam plus aminoglycoside 2 (0.8) 20.0 (8.0 32.0) 0 b-lactam plus macrolide plus fluoroquinolone 2 (0.8) 19.0 (10.0 28.0) 0 Note: Duration of IV antibiotic therapy is shown as median (25th to 75th percentile). IV, intravenous. patients body temperature, however, has recovered and respiratory symptoms including dyspnea, cough, and chest pain have improved, is it still essential to continue IV antibiotic therapy? Would the prognosis be affected at this stage if we stopped IV antibiotic therapy? Van der Eerden et al. [11] carried out a prospective study in 180 adult CAP patients in The Netherlands. They demonstrated that once body temperature was <38 C for 72 hours and respiratory symptoms including dyspnea, cough, and chest pain had improved, switching to oral antibiotic therapy was safe. In the same study, 174 patients (97%) were cured while 6 patients worsened after switching to oral antibiotic therapy. Of these, five were cured after intensive therapy and one patient died of non-cap-related cardiac disease 2 weeks after admission. There were two reported cases of failure out of a total of 104 low-risk CAP patients and the duration of IV antibiotic therapy was 4 to 4.5 days. Fernandez Alvarez et al. [8] reported that the mean duration of IV antibiotic therapy was 5.8 days in a study including 125 hospitalized CAP patients. Although the duration influenced mean hospital stay and cost, it did not add any evident therapeutic benefit. This research suggests that switching from IV to oral therapy earlier in CAP patients is safe and effective, Table 4 Antibiotics of b-lactam, macrolide, and fluroquinolones used in initial empirical antibiotic therapy Antibiotics Number of cases Proportions (%) Beta-lactams 106 44.9 Cefuroxime 46 19.5 Ceftriaxone 13 5.5 Penicillin G 10 4.2 Ertapenem 10 4.2 Cefutaxime 7 3.0 Ceftazidime 5 2.1 Cefepime 3 1.3 Amoxicillin-clavulanate 3 1.3 Piperacillin 2 0.8 Ampicillin-sulbactam 2 0.8 Cefoperazone-sulbactam 2 0.8 Cefalexin 2 0.8 Imipenem-cilastatin sodium 1 0.4 Macrolides 69 29.2 Azithromycin 44 18.6 Erythromycin 21 8.9 Roxithromycin 4 1.7 Fluroquinolones 125 53.0 Levofloxacin 98 41.5 Gatifloxacin 23 9.7 Ofloxacin 4 1.7 especially in low-risk CAP patients. According to the Chinese CAP guidelines, switching from IV to oral antibiotic therapy is also recommended if the patient has apparently improved after initial therapy for 48 to 72 hours. Therefore, shortening the duration of IV antibiotic therapy appears feasible. In our study, the median duration of IV antibiotic therapy after the patient s temperature and respiratory symptoms had improved was 7 days, which is much longer than practices of Europe and North America, and also longer than the recommended duration according to Chinese CAP guidelines. If the duration of IV antibiotic therapy in CAP patients is shortened, the length of their hospital stay will also be shortened and hospitalization costs will be lowered. Hospital stays can perhaps be shortened to about 5 to 7 days, which is typical in developed countries [5,12]. Shortening hospital stays would not necessarily affect outcome. A study of 1188 hospitalized CAP patients demonstrated that medical outcomes were similar in patients admitted to the hospital with the shortest length of stay and those admitted with longer mean lengths of stay [13]. There are distinct differences between China and developed countries in the proportions of costs of hospitalizing CAP patients. The median total hospital cost for low-risk CAP patients in our study was $556.50 (mean $705.60), which is far less than that of developed countries such as Singapore (mean $2160) [12] and Germany (median of PSI risk classes I III, $987 $1465) [9]. In developed countries, hospital bed and medical staff fees account for the largest percentage [9,14 16] of total costs, while in our country, pharmacy costs account for the largest percentage of total hospital costs for CAP treatment [4,17], which was 48.9% in our study. Pharmacy costs was nearly 50% of total costs in our hospital, it is much higher than that of developed countries, which was only 4.6% to 20.8% [9,14 16,18,19]. The antibiotics used in our hospital were almost the same with developed countries, and the prices were not higher than those of developed countries. The high pharmacy costs proportion was caused partly by inappropriate long duration of IV antibiotic therapy; it is expected to be decreased by switching from IV to oral antibiotic therapy earlier [20]. At the same time, costs for beds and medical staff in our hospital were much less and proportionally lower than those of developed countries, so the pharmacy costs proportion appeared even higher. Noticeably, the instruments for examination and laboratory reagents used in our hospital in recent years were almost as advanced as those used in the United States, but patients in China paid far less. Laboratory and examination costs were still pro-

44 Zhou et al. Table 5 Low-risk CAP patient costs separated by PSI class and initial empirical antibiotic therapy Groups separated by PSI class Groups separated by empirical therapy Class I (n = 130) Class II (n = 71) Class III (n = 35) P-value Success (n = 214) Failure (n = 22) P-value Variables ($) All patients (n = 236) Total hospital costs 556.5 (359.1 889.0) 500.6 (318.9 732.7) 631.0 (399.9 1002.6) 908.6 (475.3 1471.6) 0.000 525.1 (349.9 882.0) 744.3 (547.7 1081.3) 0.041 Pharmacy costs 257.6 (97.3 482.6) 208.2 (83.9 356.6) 320.9 (111.4 536.7) 530.7 (256.7 787.8) 0.000 253.2 (92.8 475.6) 279.6 (195.1 565.1) 0.132 Laboratory costs 142.3 (107.6 189.0) 138.5 (99.4 175.4) 145.2 (119.9 217.9) 154.2 (101.0 205.5) 0.097 139.4 (104.6 185.7) 167.9 (133.7 269.1) 0.014 Radiology costs 25.5 (9.8 94.3) 20.6 (9.8 83.4) 79.6 (19.7 113.3) 31.6 (9.8 112.7) 0.016 25.4 (9.8 93.2) 89.4 (23.3 120.5) 0.034 Medical staff costs 36.5 (27.2 58.0) 33.5 (26.7 48.7) 42.0 (27.6 62.0) 42.8 (24.5 90.9) 0.014 36.4 (26.4 55.7) 45.0 (32.1 68.1) 0.036 Bed costs 34.5 (23.9 49.7) 31.8 (23.9 43.1) 37.1 (26.5 50.4) 34.5 (21.2 76.9) 0.075 32.3 (23.9 47.7) 38.4 (29.2 57.7) 0.185 Examination costs 18.8 (12.0 47.7) 15.9 (2.4 42.4) 26.5 (14.5 58.1) 38.6 (13.0 73.8) 0.051 18.8 (10.8 46.8) 18.4 (12.0 67.7) 0.728 Other costs 11.4 (1.1 20.0) 11.0 (0.5 11.8) 12.3 (0.5 17.8) 14.6 (2.8 39.7) 0.275 11.4 (2.5 19.2) 8.9 (0.5 35.2) 0.848 Note: Data are presented as median (25% to 75% percentile). The unit costs: complete blood count $2.4; arterial blood gas $4.8; sputum culture $10.2; posterior anterior chest x-ray $4.8; digital posterior anterior chest x-ray $20.5; chest high resolution computed tomography scan $86.7; lung function tests (including spirometry, pulmonary ventilation, and diffusion tests) $31.3; electrocardiogram $2.4; daily cost of bed: $2.7 for general ward, $12 for VIP ward, and $6 for Intensive Care Unit. CAP, community-acquired pneumonia; PSI, Pneumonia Severity Index. portionally high in our hospital mainly because of low costs for beds and medical service. As a developing country, the total hospital costs of CAP treatment should be lower than that of developed countries, but the proportional costs of medical service and beds are expected to be high. Bertran et al. [21] reported that the clinical effectiveness of the first antibiotic used was the main variable in determining the final average cost per patient treated. For patients with lower respiratory tract infections, the therapeutic option with a better cost-effectiveness ratio must be chosen to minimize the risk of therapeutic failure after first line therapy. In the current study, initial empirical antibiotic therapy failure also was an important factor that could increase total hospital costs for CAP patients. The total hospital cost of patients with initial empirical therapy success were significantly lower than that of patients with failed initial empirical therapy (median $525.10 vs. median $744.30, P = 0.041); there were no significant differences of the initial empirical therapy failure rate among groups separated by PSI class (P = 0.347). Thus, increasing the success rate of initial empirical antibiotic therapy is important for decreasing total hospital costs. According to the Chinese CAP guidelines [7] published in 1999, the pathogens of CAP in hospitalized patients were mainly bacteria, such as S. pneumoniae and H. influenzae. In addition, in China, S. pneumoniae has a 73.9% resistance rate to erythromycin [22] and a 75.4% resistance rate to azithromycin [23], so macrolides were not recommended to be first antibiotics in CAP treatment. In our study, only 11 cases were treated by macrolides monotherapy because they were diagnosed as possible atypical pneumonia by physician. Therefore, the initial empirical antibiotic regimen to CAP patients in this study was in accordance with the guidelines in China. But in recent years, M. pneumoniae was found more and more important in CAP, including in low-risk patients. M. pneumoniae was the common pathogen of low-risk CAP in our current study, which is consistent with other studies [23,24]. Liu et al. [23] reported the results of a multi-center study of urban CAP pathogens that identified M. pneumoniae in 20.7% of patients, especially in those who were no more than 50 years old (about 30%) and those who had no coexisting illness (24.1%). At the same time, Mycoplasma was found to be the most common pathogen coexisting with bacteria. In our current study, empirical b-lactam monotherapy, which has no effect on Mycoplasma, had a higher failure rate (20.5%) than other regimens containing at least one macrolide or fluoroquinolone, which both work well to treat Mycoplasma infection. New CAP guidelines published in 2006 has paid more attention to atypical pathogens in CAP empirical antibiotics therapy [25]. Therefore, the initial empirical antibiotic therapy for low-risk CAP should cover atypical pathogens (mainly M. pneumoniae) and S. pneumoniae and H. influenza, which are still the top two CAP bacteria [23 26]. New fluoroquinolones such as levofloxacin and moxifloxacin are efficient not only for atypical pathogens but also for common CAP bacteria. They had high success rates as empirical monotherapy inour study (only 4 cases failed among 73 patients). Based on the current study and recent studies about etiology of CAP [23,24], we suggest the empirical antibiotic regime of a b-lactam plus macrolide or new fluroquinolones monotherapy which can cover not only atypical pathogens but also common bacteria for low-risk CAP. Fluoroquinolones, however, have been used so widely both in human beings and animals in China that there is a high rate of resistance in many bacteria, and there is cross-resistance to new fluoroquinolones [27], so we should pay attention to avoid high resistance to new fluoroquinolones.

Cost Savings in Low-Risk CAP 45 Table 6 The correlative factors of total hospital costs Parameter B Standard error t Sig. 95% Confidence interval Lower bound Upper bound Intercept -226.429 70.521-3.211 0.002-365.398-87.460 Male gender -27.536 44.196-0.623 0.534-114.630 59.557 Age 4.396 1.638 2.683 0.008* 1.167 7.624 Neoplasm 201.714 194.877 1.035 0.302-182.313 585.740 Cerebrovascular disease -236.303 198.714-1.189 0.236-627.890 155.284 COPD 346.804 239.586 1.448 0.149-125.326 818.934 Diabetes 129.715 230.982 0.562 0.575-325.461 584.891 Bilateral or multiple lobars infiltration on chest radiograph 86.710 51.063 1.698 0.091-13.915 187.334 Initial empirical antibiotic therapy failure 147.744 74.391 1.986 0.048* 1.148 294.341 PSI class 101.105 44.243 2.285 0.023* 13.919 188.291 Duration of IV antibiotic therapy 51.180 5.334 9.595 0.000* 40.668 61.691 Duration of IV antibiotic therapy after body temperature and respiratory symptoms improved -4.218 6.085-0.693 0.489-16.210 7.774 Note: *Indicates significance (P < 0.05). COPD, chronic obstructive pulmonary disease; PSI, Pneumonia Severity Index; IV, intravenous. There were limitations in our study. First, it was a singlecenter study that just reflected the state of tertiary hospitals in Beijing and can not represent the state of secondary level and primary level hospitals. Second, it was a retrospective analysis and lacked randomization. Third, 12 low-risk CAP patients who had incomplete clinical data were excluded, which may lead to sample-related bias. Lastly, patients who were treated in clinics were not included in this study, so the pathogen limited in hospitalized low-risk CAP patients, and antibiotic therapy before admission would influence the result of sputum culture. In conclusion, duration of IV antibiotic therapy, PSI class, age, and initial empirical antibiotic therapy failure were significantly correlated with total hospital costs for low-risk CAP patients admitted into our tertiary urban hospital. The length of hospital stay can be shortened by switching earlier from IV to oral antibiotic therapy. Focused efforts to reduce duration of IV antibiotic therapy should have the most profound effect on reducing total hospital costs. The atypical pathogens should be considered for initial empirical antibiotics in low-risk CAP therapy in China. Source of financial support: None. References 1 Goss CH, Rubenfeld GD, Park DR, et al. Cost and incidence of social comorbidities in low-risk patients with communityacquired pneumonia admitted to a public hospital. Chest 2003; 124:2148 55. 2 Monge V, San-Martin VM, Gonzalez A. The burden of community-acquired pneumonia in Spain. Eur J Public Health 2001;11:362 4. 3 Scott G, Scott H, Turley M, Baker M. Economic cost of community-acquired pneumonia in New Zealand adults. N Z Med J 2004;117:U933. 4 Ministry of Health of the People s Republic of China. China health statistical yearbook 2006. Available from: http://www. moh.gov.cn [Accessed December 29, 2006]. 5 Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243 50. 6 Xue XY, Gao ZC. Comparison between the severity criteria of Chinese guideline for community acquired pneumonia and Fine s risk classes. Zhonghua Jie He He Hu Xi Za Zhi 2002;25:719 22. 7 Respiratory Society of Chinese Medical Association. Guidelines for diagnosis and treatment of community acquired pneumonia. Zhonghua Jie He He Hu Xi Za Zhi 1999;22:199 201. 8 Fernandez Alvarez R, Gullon Blanco JA, Rubinos Cuadrado G, et al. Community-acquired pneumonia: influence of the duration of intravenous antibiotic therapy on hospital stay and the cost-benefit ratio. Arch Bronconeumol 2001;37:366 70. 9 Bauer TT, Welte T, Ernen C, et al. Cost analyses of communityacquired pneumonia from the hospital perspective. Chest 2005; 128:2238 46. 10 Fine MJ, Pratt HM, Obrosky DS, et al. Relation between length of hospital stay and costs of care for patients with community acquired pneumonia. Am J Med 2000;109:378 85. 11 van der Eerden MM, de Graaff CS, Vlaspolder F, et al. Evaluation of an algorithm for switching from IV to PO therapy in clinical practice in patients with community-acquired pneumonia. Clin Ther 2004;26:294 303. 12 Lee KH, Chin NK, Tan WC, Lim TK. Hospitalised low-risk community-acquired pneumonia: outcome and potential for costsavings. Ann Acad Med Singapore 1999;28:389 91. 13 McCormick D, Fine MJ, Coley CM, et al. Variation in length of hospital stay in patients with community-acquired pneumonia: are shorter stays associated with worse medical outcomes? Am J Med 1999;107:5 12. 14 Sun HK, Nicolau DP, Kuti JL. Resource utilization of adults admitted to a large urban hospital with communityacquired pneumonia caused by Streptococcus pneumoniae. Chest 2006;130:807 14. 15 Niederman MS, McCombs JS, Unger AN, et al. The cost of treating community-acquired pneumonia. Clin Ther 1998;20: 820 37. 16 Klepser ME, Klepser DG, Emst EJ, et al. Health care resource utilization associated with treatment of penicillin-susceptible and -nonsusceptible isolates of Streptococcus pneumoniae. Pharmacotherapy 2003;23:349 59. 17 Tan XY, He QY, Wang YZ, Zhao XH. Investigation on the hospitalization expenses incurred by 362 cases of community acquired pneumonia. Chin J Hosp Admin 2002;18:413 16. 18 Bartolomé M, Almirall J, Morera J, et al. A population-based study of the costs of care for community-acquired pneumonia. Eur Respir J 2004;23:610 6. 19 Orrick JJ, Segal R, Johns TE, et al. Resource use and cost of care for patients hospitalised with community acquired pneumonia: impact of adherence to Infectious Diseases Society of America guidelines. Pharmacoeconomics 2004;22:751 7. 20 Wawruch M, Bozekova L, Krcmery S, et al. Cost-effectiveness analysis of switching from intraveneous to oral administration of antibiotics in elderly patients. Bratisl Lek Listy 2004;105: 33374 8. 21 Bertran MJ, Trilla A, Codina C, et al. Analysis of the costeffectiveness relationship in the empirical treatment in patients

46 Zhou et al. with infections of the lower respiratory tract acquired in the community. Enferm Infecc Microbiol Clin 2000;18:445 51. 22 Song JH, Jung SI, KO KS, et al. High prevalence of antimicrobial resistance among clinical Streptococcus pneumoniae isolates in Asia (an ANSORP study). Antimicrob Agents Chemother 2004; 48:2101 7. 23 Liu YN, Chen MJ, Zhao TM, et al. A multicentre study on the pathogenic agents in 665 adult patients with community acquired pneumonia in cities of China. Zhonghua Jie He He Hu Xi Za Zhi 2006;29:3 8. 24 Lauderdale TL, Chang FY, Ben RJ, et al. Etiology of community acquired pneumonia among adult patients requiring hospitalization in Taiwan. Respir Med 2005;99:1079 86. 25 Respiratory Society of Chinese Medical Association. Guidelines for diagnosis and treatment of community acquired pneumonia. Zhonghua Jie He He Hu Xi Za Zhi 2006;29:651 5. 26 Apisarnthanarak A, Mundy LM. Etiology of communityacquired pneumonia. Clin Chest Med 2005;26:47 55. 27 Wang J, Li JT, Li Y. Antibacterial activities of fluoroquinolones to 2 554 bacterial strains. Chin J Infect Chemother 2004;4:14 7.