CAP-START

Published

Postma DF, et al. "Antibiotic treatment strategies for community-acquired pneumonia in adults". The New England Journal of Medicine. 2015. 372(14):1312-1323.
PubMed • Full text • PDF

Clinical Question

In patients with community-acquired pneumonia admitted to non-intensive care unit (ICU) wards, is beta-lactam monotherapy noninferior to quinolone therapy or combination beta-lactam plus macrolide?

Bottom Line

Among patients with community-acquired pneumonia (CAP), beta-lactam monotherapy was noninferior to quinolone monotherapy or combination beta-lactam plus macrolide in terms of 90-day mortality.

Major Points

Current US guidelines for CAP in a non-ICU setting recommend that inpatient treatment include coverage with either a macrolide plus a beta-lactam or a fluoroquinolone for empiric therapy. The evidence supporting this is limited, and in fact, many European countries have guidelines that support beta-lactam monotherapy as the initial choice.

The CAP-START trial was a cluster-randomized, crossover, noninferiority trial conducted at seven hospitals in the Netherlands, and compared empirical CAP treatment with beta-lactam monotherapy (BL) vs. beta-lactam-macrolide combination therapy (BLM) or fluoroquinolone monotherapy (FQ). The antibiotic choices were based on the 2005 Dutch guidelines, and treatment could deviate from these therapy assignments for medical reasons. The trial enrolled patients age 18 years or older with clinically suspected CAP who required hospitalization in a non-ICU setting. The primary outcome was 90-day mortality with noninferiority defined as a less than 3% difference. The unadjusted 90-day mortality for all three groups was 9.0% for BL, 11.1% for BLM, and 8.8% for FQ, indicating noninferiority of the BL group.

This study has some limitations. First, although the protocol randomly assigned patients to treatment groups, it allowed changes to therapy per investigator discretion, and this may have confounded treatment effects. For example, during the course of hospitalization, approximately 40% of patients in the BL group received another antibiotic which provided coverage against atypical bacteria. Second, regional differences limit the study's generalizability. In the US, for example, atypical bacteria such as Mycoplasma and Leigionella are common (5-10%) causes of CAP, whereas atypical bacteria were only identified in 2% of patients in the CAP-START study. Finally, a meta-analysis of BLM vs. BL studies indicated a possible survival benefit with BLM combinations, possibly owing to antiinflammatory properties of macrolides not related to their antibacterial effects. Consequently, CAP-START provides a basis for the rational administration of beta-lactam monotherapy, with modifications based on clinical judgment, to patients with CAP hospitalized in areas with low incidence of atypical infections. US guidelines, jointly created by the Infectious Disease Society of America (IDSA) and American Thoracic Society (ATS), continue to recommend either a beta-lactam plus a macrolide or a respiratory fluoroquinolone as the first-line therapy for inpatient CAP treatment.

Guidelines

IDSA/ATS Guidelines - Inpatient, non-ICU Treatment (2007, adapted)
As of March 2018, these guidelines are currently undergoing revision and the IDSA projects publication in Summer 2018.

1. A respiratory fluoroquinolone (strong recommendation; level I evidence)
2. A beta-lactam plus a macrolide (strong recommendation; level I evidence) with doxycycline (level III evidence) as an alternative to the macrolide.

Design

• Multicenter, cluster-randomized, crossover, non-inferiority trial
• N=2,283 adults with CAP requiring non-ICU hospitalization
  • Beta-lactam (N=656)
  • Beta-lactam-macrolide (n=739)
  • Fluoroquinolone (n=888)
• Setting: 7 hospitals in the Netherlands
• Enrollment: 2011-2013
• Analysis: Intention-to-treat
• Primary outcome: 90-day all-cause mortality
• Secondary outcomes:
  • Time to starting oral antibiotics
  • Length of hospital stay
  • Occurrence of major or minor complications during the hospital stay

Population

Inclusion Criteria

• Age ≥18 years
• Clinically suspected CAP who required antibiotics and non-ICU hospitalization
  • Radiologically confirmed CAP plus 2 other clinical criteria
  • Presence of 2 diagnostic clinical criteria (cough, purulent sputum, temperature >38 degrees Celsius, positive auscultory findings, leukocytosis, CRP >3×ULN, dyspnea/tachypnea/hypoxia) with no obvious non-respiratory source of infection
• One hospital only included patients with a CURB-65 score greater than 2

Exclusion Criteria

• Patients with cystic fibrosis
• Positive Legionella antigen
• Recent hospitalization for >48 hours in the previous 2 weeks
• Patients from long-term care facilities

Baseline Characteristics

Placed in order of BL, BLM, and FQ groups.

• Age: 70 (60–79), 70 (59–80), 71 (59–79)
• Current smoker: 109/627 (17.4%), 154/723 (21.3%), 196/872 (22.5%)
• Antibiotics before admission: 219/637 (34.4%), 227/721 (31.5%), 303/873 (34.7%)
• Received pneumococcal vaccine:
  • PPSV23: 16/594 (2.7%), 18/671 (2.7%), 13/822 (1.6%)
  • PCV13: 19/656 (2.9%), 7/739 (0.9%), 10/888 (1.1%)
• Coexisting Conditions:
  • COPD or asthma: 260 (39.6%), 281 (38.0%), 377 (42.5%)
  • Other chronic pulmonary disease: 64 (9.8%), 97 (13.1%), 61 (6.9%)
  • Diabetes: 118 (18.0%), 101 (13.7%), 161 (18.1%)
• Pneumonia Severity Index Score: 84.6±29.0, 84.8±27.8, 85.4±28.5
• Median CURB 65 Score: 1 (1–2), 1 (1–2), 1 (1–2)
• Radiologically confirmed CAP: 506 (77.1%), 566 (76.6%), 665 (74.9%)

Interventions

• Hospitals randomized to a six-block schedule which had two blocks each of macrolide monotherapy, fluoroquinolone monotherapy, and beta-lactam-macrolide combination therapy.
• Each patient diagnosed with CAP was treated with the antibiotic for which the hospital was currently scheduled.

Outcomes

Primary Outcomes

90-day all-cause mortality

Unadjusted:

Beta-lactam (BL) monotherapy: 9.0%

Fluoroquinolone (FQ) monotherapy: 8.8%

Beta-lactam-macrolide (BLM) combination therapy: 11.1%

Adjusted comparisons:

BL vs. FQ (-0.6% in favor of fluoroquinolone monotherapy, 90% CI, -2.8 to 1.9)

BL vs. BLM (1.9% in favor of beta-lactam monotherapy, 90% CI, -0.6 to 4.4)

Secondary Outcomes

Time to starting oral antibiotics

BL monotherapy: 4 (3-5) days

FQ monotherapy: 3 (0-4) days

BLM combination therapy: 4 (3-5) days

Length of hospital stay

BL monotherapy: 6 (4-8) days

FQ monotherapy: 6 (4-8) days

BLM combination therapy: 6 (4-10) days

Occurrence of major or minor complications during the hospital stay (%)

Major Complications

BL monotherapy: 4.9%

FQ monotherapy: 5.3%

BLM combination therapy: 5.7%

Minor Complications

BL monotherapy: 11%

FQ monotherapy: 12.3%

BLM combination therapy: 13.1%

Subgroup Analysis

Post-hoc subgroup analysis was done for both strategy-adherent and antibiotic-adherent subgroups for all outcomes. 90-day all cause mortality was not reported directly in the paper but was found to show that BL monotherapy was non-inferior to both BLM combination therapy and FQ monotherapy.

Strategy-adherent subgroup analysis

Rate ratio for discharge alive (95% CI)

BL vs. FQ: 1.04 (0.93-1.16)

BL vs. BLM: 0.86 (0.77-0.97)

Rate ratio for time to starting oral treatment (95% CI)

BL vs. FQ: 1.33 (1.17-1.51)

BL vs. BLM: 0.94 (0.83-1.08)

Odds ratio for complications (95% CI)

BL vs. FQ: 1.03 (0.73-1.46)

BL vs. BLM: 1.06 (0.74-1.52)

Antibiotic-adherent subgroup analysis

Rate ratio for discharge alive (95% CI)

BL vs. FQ: 1.03 (0.92-1.17)

BL vs. BLM: 0.86 (0.74-0.95)

Rate ratio for time to starting oral treatment (95% CI)

BL vs. FQ: 1.52 (1.28-1.80)

BL vs. BLM: 0.93 (0.79-1.11)

Odds ratio for complications (95% CI)

BL vs. FQ: 1.03 (0.71-1.51)

BL vs. BLM: 1.20 (0.82-1.77)

Criticisms

• Generalizability limited by low incidence of atypical bacteria and by low resistance of S. pneumoniae to penicillins and macrolides in this population compared to other regions (eg, US).
• Treatment-related differences are confounded by the fact that 38.7% of patients in the BL monotherapy group received an additional antibiotic for atypical coverage.

Funding

The study was funded by the Netherlands Organization for Health Research and Development. There were no reported conflicts of interest.