Community acquired pneumonia (CAP) is a common respiratory infection and is the main cause of ICU admission and death in adults. Because of most patients were treated empirically against suspected causative microorganism, it is important to assess the effectiveness of treatment after 3 days of anti-infective therapy. However, the criteria for treatment failure is lack of a clear-cut and validated definition from the CAP guidelines.

Pneumonia severity scores is a wide-used severity rating system for treatment selection and outcome prediction for CAP. So far, the pneumonia severity scores only used once before the treatment started. Considering the pneumonia severity scores could reflect the severity of pneumonia, it is reasonable to assume that the change of pneumonia severity scores could reflect the patients' condition and the effectiveness of the treatment. This trail will be designed to validate the feasibility of assessing effectiveness of CAP treatment by using continuous pneumonia severity score.

The applicant has established a CAP patients database and found a new effective pneumonia severity score: Expand-CURB. According to the retrospective study results, applicant found the changes of pneumonia severity score could reflect the patients'condition and was the independent risk factor for 30-day mortality. More specifically, if the patients'Expand-CURB score was not improved after 3-5 days of initial treatment, the odds ratio of 30-day mortality was 5.571 and 5.249 (95%CI 1.831-15.051, P=0.002) by univariate analysis and multivariate analysis (the multivariate was adjusted by the initial severity of pneumonia).

In addition, the applicant found the changes of serum CRP(C reactive protein ) before and after initial treatment also associated the patients'outcomes. If the serum CRP decreased less than 40% or not lower than 20mg/dl after the initial treatment, the odds ratio of 30-day mortality will be 3.692 and 3.806 (95%CI 1.867-7.756, P<0.001) by univariate analysis and multivariate analysis.

In conclusion, the applicant established three criteria for assessing effectiveness of CAP treatment:

1. Effective treatment by PSI: the PSI score decreased after 3-5 days the initial treatment.
2. Effective treatment by Expand-CURB: the Expand-CURB score decreased after 3-5 days the initial treatment.
3. Effective treatment by serum CRP: the serum CRP decreased more than 40% or lower than 20mg/dl after 3-5 days the initial treatment.

Furthermore, the applicant found combined pneumonia severity scores with CRP could predict the 30-day mortality more efficiently. The OR for both CRP and PSI treatment failure was 2.377 (95%CI 1.100-5.136, P=0.028). And the OR for both CRP and Expand-CURB treatment failure was 7.332 (95%CI 1.563-34.385, P=0.012). Therefore, it is reasonable to assume that the change of pneumonia severity scores and CRP could reflect the patients'condition and the effectiveness of the treatment.

To validate the feasibility of assessing effectiveness of CAP treatment by using continuous pneumonia severity score and changes of CRP, applicant intend to randomly allocate the patients into three arms: Comprehensive evaluation group, PSI evaluation group and Expand-CURB evaluation group. Patients who did not agree to participate the randomized controlled study will be asked whether they agree to provide their clinical data for prospective observational studies.

The primary outcomes will be 30-day mortality and ICU admission. The second outcome is length of hospital stay.

Inclusion Criteria:

• community acquired pneumonia(by CTS or IDSA/ATS guidelines)
• severe community acquired pneumonia(PSI scores >= 90 or Expand-CURB scores >= 4);

Exclusion Criteria:

• HIV patients
• Other immunodeficiency disorders (neutropenia, hematologic and solid tumors undergoing chemoradiotherapy, organ transplantation, and long-term treatment with glucocorticoid and cytokine antagonists).

The present study is a randomized controlled open label trial. Patients and their physicians know the group and the treatment plan, and there are clear criteria for the final outcome (30-day fatality rate, length of stay). The masking is not required.