Disease characteristics and management of hospitalised adolescents and adults with community-acquired pneumonia in China: a retrospective multicentre survey

Objectives To describe the clinical characteristics and management of patients hospitalised with community-acquired pneumonia (CAP) in China. Design This was a multicentre, retrospective, observational study. Setting 13 teaching hospitals in northern, central and southern China from 1 January 2014 to 31 December 2014 Participants Information on hospitalised patients aged ≥14 years with radiographically confirmed pneumonia with illness onset in the community was collected using standard case report forms. Primary and secondary outcome measures Resource use for CAP management. Results Of 14 793 patients screened, 5828 with radiographically confirmed CAP were included in the final analysis. Low mortality risk patients with a CURB-65 score 0–1 and Pneumonia Severity Index risk class I–II accounted for 81.2% (4434/5594) and 56.4% (2034/3609) patients, respectively. 21.7% (1111/5130) patients had already achieved clinical stability on admission. A definite or probable pathogen was identified only in 12.7% (738/5828) patients. 40.9% (1575/3852) patients without pseudomonal infection risk factors received antimicrobial overtreatment regimens. The median duration between clinical stability to discharge was 5.0 days with 30-day mortality of 4.2%. Conclusions These data demonstrated the overuse of health resources in CAP management, indicating that there is potential for improvement and substantial savings to healthcare systems in China. Trial registration number NCT02489578; Results.


Strengths and limitations of this study
This is the largest multi-center study to investigate demographic characteristics, severity and microbiological testing, empirical antimicrobial treatment, duration of hospitalization and 30-day mortality among adults and adolescents hospitalized with CAP in mainland China, including adolescents and adults of all ages admitted to general hospital wards or ICUs from the participating centers， patients who were critically ill, aged >90 years, and/or immunosuppressed.
The participating hospital sites are teaching hospitals in seven cities in three provinces, and may not be representative of CAP in smaller, rural hospitals.
The majority of patientsare adult CAP patients, our findings do not apply to children hospitalized with CAP.

Background
Community acquired pneumonia (CAP) is one of the most common infectious syndromes and is a leading cause of death worldwide [1][2] . In Europe, the reported rate of CAP ranges from 1·6 to 9 cases per 1,000 in the general adult population per year [3][4][5] . Despite advances in medical technology and global economic development, CAP-associated mortality remains high (e.g., 20.9/100,000 in the United States and 12.7/100,000 in Canada) [2,6] . Patients hospitalized in intensive care units for CAP have mortality in excess of 20% for immunocompetent patients and closer to 30% for those immunocompromised [7] . In Japan and Korea, the 30-day mortality of patients hospitalized with CAP is about 4-6% [8][9] .
Although mainland China has nearly 19% of the world's population, there are limited data on CAP management and disease burden in China during the last ten years.

According to a household interview survey published in the China Health and Family
Planning Statistical Yearbook (2013), the two-week prevalence of pneumonia in China was estimated to be 11/1,000, and the direct cost due to bacterial pneumonia was about 320 million RMB (approximately $46.4 million) [10] .

Data Collection:
A total of 786 variables were included in the formatted CRF, including: (1) Demographic data: age, gender, ID number, source of admission, types of medical insurance; (2) Underlying diseases: chronic lung, heart, renal and liver diseases, diabetes, solid organ cancers, immunocompromised status, such as leukemia and lymphoma, chemotherapy or radiation within six months, bone marrow and solid organ transplantation, splenectomy. Definition of underlying diseases is listed in Appendix file 4.
(3) Factors for acquisition or prevention of CAP: pregnancy, postpartum within six months, current smoking history, excessive drinking, exposure to day care center children, bed-ridden longer than two months, chronic receipt of corticosteroids (dosage equivalent prednisolone ≥10mg/d for more than 30 days), statin use, S. pneumoniae or Influenza vaccination within one year.   Continuous renal replacement therapy (CRRT) and extracorporeal membrane oxygenation (ECMO) were also recorded. (7) Clinical stability was defined as satisfying all of the following: temperature ≤37.8 °C more than 24 hours without use of antipyretic medications; resting heart rate ≤100 beats/min; respiratory rate ≤24 breaths/ minute; systolic blood pressure ≥90mmHg; SpO2 ≥ 90% on room air; ability to maintain oral intake; normal mental status [12] . (8) Over-treatment was defined as: (i) use of antipseudomonal β-lactams or β-lactams+ fluoquinolones in patients aged <65 years without risk factors for pseudomonal infection; (ii) use of (antipseudomal or not) β-lactams+ fluoquinolones in patients aged≥65 years without risk factors for pseudomonal infection and not in an ICU [13] . (9) Risk factors for pseudomonal infection was defined as chronic airway disease (bronchiectasis and COPD), immunocompromised status and at least one risk factor for HCAP as defined by the 2008 IDSA/ATS adult CAP guidelines. [11,[13][14][15][16] .

Microbiology testing
The conditions that a pathogen was defined as the definite or probable etiology based on were showed in Appendix 6.

Statistical analysis
No formal sample size calculations were performed because of the retrospective descriptive study design. All data were analyzed by descriptive statistics with SPSS19.
Measurement data were tested for normality by Kolmogorov-Smirnov. Measurement data of normal distribution was reported as mean ± standard deviation. Measurement data of non-normal distribution was reported as median. The χ2 test statistics were used for 30-day mortality subgroup analysis. A P-value of <0.05 was considered statistically significant.

Screening Process
A total of 14,793 patients were screened to meet the inclusion and exclusion criteria for CAP and 6,056 patients were included in the final analysis (Appendix Figure 1).
A substantial proportion of admitted patients had relatively mild disease as indicated by the following: i) CURB-65 score [17] 0-1 accounted for 76.3%, ii) PSI risk class [18] (Table 4)

Clinical outcomes
Clinical outcomes are shown in Table 5. Overall, 6.5% of patients were admitted to an ICU, and 2.8% required invasive mechanical ventilation. Vasopressors were administered to 3.5% of patients, and 27.5% received corticosteroids during the hospitalization. The 30-day mortality was 4.2%. The median duration of hospitalization was 10 days. The median duration from admission to clinical stability was 4 days, and from clinical stability to discharge was 5 days. The median duration of ICU hospitalization was 8 days. The top five causes of death were severe pneumonia/multi-organ dysfunction syndrome (MODS) 66.9% (172/257), cardiac failure 3.1% (8/257), stroke 1.9% (5/257), acute myocardial infarction 1.9% (5/257), and gastrointestinal hemorrhage 1.9% (5/257). Appendix 7 shows the results of sub-group analysis of 30-day mortality. Fatality increased with age and there was a jump up at 16.0% among those aged ≥90 years.
Mortality was similar between male and female patients (4.9% vs 3.5%). Mortality in patients was >10% in patients with organ/bone marrow transplantation, immunosuppressive therapy, long-term oral corticosteroids use, chemotherapy/radiology within 6 months and splenectomy. Mortality in patients admitted to an ICU was 25.3%.

Discussion
This study represents the largest, multicenter, retrospective cohort study on the etiologies and outcomes in adolescents and adults with CAP in China. In this study, We identified four major categories of overuse of health care resources in CAP management in China: (1) A large number of low-risk patients were admitted to the hospitals. Guidelines for CAP management in China and the U.S. recommend that decisions for hospitalization should be based on illness severity [11,13] . It was estimated that over $8 billion dollars are spent in CAP treatment every year in the U.S, and the cost for inpatient CAP management is 25-30 times more than for outpatient CAP management [22][23][24]. . Therefore, admission of low mortality risk CAP patients results in major unnecessary cost expenditures. Moreover, outpatients usually return to their baseline activity levels much sooner than inpatients, and enjoyed a higher quality of life [25][26] . Finally, hospitalization is associated with the risk of nosocomial infections, potentially caused by high virulent and multidrug-resistant organisms [27] . Admission of low-risk CAP patients was also observed in a recent large U.S. study [28] , so it may not be unique to China.
(2) Length of stay in hospital was unnecessarily long. CAP guidelines recommended that patients should be discharged as soon as they achieve clinical stability and have no other active medical problems. Keeping patients in hospital and observing them while receiving oral antibiotic therapy, or waiting for normalization of all clinical with in-hospital adverse events [27,[29][30] . We observed that CAP patients were discharged a median of 5 days after achieving clinical stability, and 22% met clinical stability criteria at admission. Given the median LOS of 10 days for all CAP patients, discharging CAP patients once they achieved clinical stability would lead to cost-savings of approximately half of the total hospitalization expenses.
(3) 29.1% patients without risk factors for Pseudomonas infection received over-treatment with empiric antimicrobial regimens. Antipseudomonal β-lactams (17.0%) or β-lactams + quinolones (12.1%) were the most common empiric regimens for over-treatment. This may be due to overestimation of illness severity, clinician unfamiliarity with CAP guidelines, or lack of microbiologic diagnostic testing.
Moreover, we found quinolones use in more than 40% of CAP patients. The U.S.
Food and Drug Administration (FDA) has released warnings of potential adverse effects of fluoroquinolones, such as Q-T prolongation, tendon injury, psychiatric disorder, etc [31][32][33] . As second-line anti-tuberculosis drugs, fluoroquinolones can also affect the diagnosis of tuberculosis and induce drug-resistance [34][35] .
(4) Unnecessary serological testing was performed. We observed that many patients had an acute serum specimen collected for IgG serology testing for atypical bacteria and respiratory viruses without a convalescent serum specimen obtained for paired serological testing. Furthermore, many patients had testing for IgM antibodies for a variety of respiratory pathogens, but elevation of IgM antibodies with a low-normal IgG titer is uncommon during acute illness [36][37][38] .Similarly, although a low IgM antibody level with a high IgG titer would be suggestive of past infection, the performance characteristics of these assays may not be reliable. A follow-up convalescent serum specimen to document changes in IgG and IgM antibody levels is generally required for diagnosis [39][40] . Thus, the value of antibody testing on a single acute serum specimen to determine the etiology of CAP is questionable. The costs of more frequent use of PCR testing on lower respiratory specimens may be partially offset by not performing serological testing in CAP patients.
Although we identified substantial over-use of health-care resources, the outcome of patients hospitalized with CAP in China was not ideal. Although the 30-day mortality was low (4.2%), this should be interpreted in the context that approximately 70% had mild CAP as indicated by pneumonia scoring indices. Mortality for CAP patients with a CURB-65 score of ≥2 (15.8%) or PSI risk ≥Ⅲ risk class (9.1%) were higher than what has been reported in developed countries, especially in critically ill patients with a CURB-65 score of 3-5 and PSI risk Ⅳ-Ⅴ class [41][42] .

References
Setting 13 teaching hospitals in northern, central and southern China from 1 January

to 31 December 2014
Participants Information on hospitalized patients aged ≥14 years with radiographically-confirmed pneumonia with illness onset in the community was collected using standard case report forms.
Primary and secondary outcome measures Resource use for CAP management. The median length of stay in hospital was 11 days. The median duration between clinical stability to discharge was 5.0 days with 30-day mortality of 4.2%.

Results
Conclusions These data demonstrated overuse of health resources in CAP management, indicating that there is the potential for improvement and substantial savings to health-care systems in China. The participating hospital sites are teaching hospitals in seven cities in three provinces, and may not be representative of CAP in smaller, rural hospitals.

Strengths and limitations of this study
The majority of patients are adult CAP patients, our findings do not apply to children hospitalized with CAP.

Background
Community acquired pneumonia (CAP) is one of the most common infectious syndromes and is a leading cause of death worldwide. 1 2 In Europe, the reported rate of CAP ranges from 1.6 to 9 cases per 1,000 in the general adult population per year. 3 4 5 Despite advances in medical technology and global economic development, CAP-associated mortality remains high (e.g., 20.9/100,000 in the United States and 12.7/100,000 in Canada). 2 6 Patients hospitalized in intensive care units for CAP have mortality in excess of 20% for immunocompetent patients and closer to 30% for those immunocompromised. 7 In Japan and Korea, the 30-day mortality of patients hospitalized with CAP is about 4-6%. 8 9 Although mainland China has nearly 19% of the world's population, there are limited data on CAP management and disease burden in China during the last ten

Study Design and Population
Data were collected from 13 hospitals in Northern (Beijing), Central (Yantai, Qindao, Weifang, Zibo, Rizhao cities in Shandong Province) and Southern (Kunming City in Yunan Province) China. A listing of participating centers can be found in Appendix 1.
All patients admitted to the 13 hospitals during 1 January 2014 through 31 December 2014 with the relevant disease codes of pneumonia or pulmonary infection in the  Patients were excluded if (1)

Data Collection:
A total of 786 variables were included in the formatted CRF, including: (1) Demographic data: age, gender, ID number, source of admission, types of medical insurance; (2) Underlying diseases: chronic lung, heart, renal and liver diseases, diabetes, hypertension, solid organ cancers. Definition of underlying diseases is listed in Appendix file 4.
(3) Factors for acquisition or prevention of CAP: pregnancy, postpartum within six months, current smoking history, excessive drinking, exposure to day care center

Microbiology testing
The conditions that a pathogen was defined as the definite or probable etiology based on were showed in Appendix 6.

Statistical analysis
No formal sample size calculations were performed because of the retrospective descriptive study design. All data were analyzed by descriptive statistics with SPSS19.
Measurement data were tested for normality by Kolmogorov-Smirnov. Measurement data of normal distribution was reported as mean ± standard deviation. Measurement data of non-normal distribution was reported as median. The χ2 test statistics were

Screening Process
A total of 14,793 patients were screened to meet the inclusion and exclusion criteria for CAP and 5828 patients were included in the final analysis (Appendix Figure 1).

Epidemiological characteristics
The proportions of male and female patients were similar.

Clinical and radiological features
Clinical and radiological features on admission are shown in Table 2

Clinical outcomes
Clinical outcomes are shown in Table 5  Definition of 2 nd and 3 rd level hospital in China: The 2 nd level hospital was defined as a hospital providing medical, prevention, health care and rehabilitation services to multiple communities (with a radius of population more than 100,000 peoples); the 3 rd level hospital was defined as a hospital providing medical service to the whole country beyond cities and provinces, with comprehensive medical, teaching and research ability.     1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45      Design This was a multicenter, retrospective, observational study.

Note：a. Direct Microscopy of sputum is not included
Setting 13 teaching hospitals in northern, central and southern China from 1 January

to 31 December 2014
Participants Information on hospitalized patients aged ≥14 years with radiographically-confirmed pneumonia with illness onset in the community was collected using standard case report forms.

Primary and secondary outcome measures
Resource use for CAP management. Conclusions These data demonstrated overuse of health resources in CAP management, indicating that there is potential for improvement and substantial savings to health-care systems in China.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  This is the largest multi-center study to investigate demographic characteristics, severity and microbiological testing, empirical antimicrobial treatment, duration of hospitalization and 30-day mortality among adults and adolescents hospitalized with CAP in mainland China, including adolescents and adults of all ages admitted to general hospital wards or ICUs from the participating centers， patients who were critically ill and aged >90 years.

Strengths and limitations of this study
The participating hospital sites are teaching hospitals in seven cities in three provinces, and may not be representative of CAP in smaller, rural hospitals.
The majority of patients are adult CAP patients, so our findings do not apply to children hospitalized with CAP.
The study design was approved by the Ethics committee of China-Japan Friendship Hospital (No.2015-86). Given the retrospective nature of the study, the Ethics Committee determined that informed consent was not necessary.

Quality control of the study
Key investigators, including clinicians, statisticians, microbiologists and radiologists worked together to draft the protocol and created a single formatted case report form (CRF) that was utilized by all centers. Before study initiation, all investigators from the thirteen centers received training on the protocol, screening process, definition of underlying diseases and formatted CRF(Appendix file 3). After data were collected, the CRF was reviewed by a trained researcher to ensure its completeness and data quality. A second review was performed independently by a trained team of physicians in each center before being entering in duplicate into a computerized database.

Microbiology testing
The conditions that a pathogen was defined as the definite or probable etiology based on were showed in Appendix 6.

Screening Process
A total of 14,793 patients were screened to meet the inclusion and exclusion criteria for CAP and 5828 patients were included in the final analysis (Appendix Figure 1).

Clinical and radiological features
Clinical and radiological features on admission are shown in Table 2. Cough, sputum, shortness of breath and fever were the most common. 64.8% patients had multi-lobar infiltrates and 20.7% of patients had pleural effusion.