Poor Compliance with Community-Acquired Pneumonia Antibiotic Guidelines in a Large Australian Private Hospital Emergency Department

Introduction

It has been estimated that there are 2 cases of community-acquired pneumonia (CAP) per 1,000 of the Australian adult population per year. ³⁵ Hospital admissions for CAP accounted for 4% of all hospital admissions in patients >65 years of age in Victorian hospitals between 2000 and 2002. ³³ When compared with similar patients admitted for other diagnoses, CAP is associated with longer hospital length of stay, increased likelihood of requiring intensive care admission, and increased mortality. ³³ Key components in the management of pneumonia are timely administration and appropriate choice of empiric antibiotics.

Early administration of antibiotics in CAP is associated with reduced mortality and decreased hospital length of stay in patients over the age of 65,^5,16,26 although the same association has not been demonstrated in patients under 65. ¹⁵ The Infectious Diseases Society of America (IDSA) published CAP guidelines in 2003 recommending the administration of antibiotics within 4 hours of arriving in the emergency department (ED). ²⁰ This 4-hour target soon became linked to financial reimbursement in U.S. hospitals. ²⁰ However, it was later found that the effect of the 4-hour target was to increase the misdiagnosis of CAP and increase inappropriate antibiotic administration. ²⁰ The IDSA guidelines were later changed to recommend no specific time target for antibiotic administration but recommend that the first dose be given in the ED.^20,23 Currently there is no guidance in Australia or New Zealand with regards to the timing of antibiotics in CAP from either the Australian College of Emergency Medicine, The Thoracic Society of Australia and New Zealand, or the Australasian Society for Infectious Diseases. While a balance between the timing of antibiotic administration and ensuring the correct diagnosis must be maintained, evidence has demonstrated that delays in antibiotic administration are associated with increased mortality.^16,26

Appropriate choice of empiric antibiotics is a key component in the management of CAP. Guidelines provide similar recommendations on the choice of antibiotics in CAP in Queensland and Australia.^1,8 The vast majority of CAP in Australia can be treated successfully with narrow-spectrum beta-lactam treatment combined with doxycycline or a macrolide. ⁹ The use of broader spectrum antibiotics has the potential to increase rates of antibiotic resistance that is proving a major problem in Australia and internationally.^12,21 Antibiotic-resistant bacteria are associated with increased morbidity, mortality, and increased health care costs. ³⁷ Thus, the Australian Commission for Safety and Quality in Health Care has listed antibiotic stewardship as one of its key priority areas and mandatory for hospital accreditation. ³

Most studies of antibiotic use in patients with CAP in Australia have been in large public teaching hospitals.^9,24,25 These hospitals generally promote good access to local and national guidelines, and often have restrictions on the use of “last-line antibiotics.” While each private hospital is different, some do not have intense promotion of therapeutic guidelines nor restrictions on antibiotic use other than those of the Pharmaceutical Benefits Scheme (PBS; the central government pharmaceutical agency). Given that the private hospital sector makes up a significant proportion of Australian health care, it is important to be aware of antibiotic use in these hospitals. ³⁰

The aim of this study was to evaluate the management of pneumonia in a large private Brisbane hospital ED. Timing of antibiotic administration and choice of antibiotics were key outcome measures of this study.

Patients and Methods

A 500-bed private Brisbane hospital with an ED was the setting for the study. All patients 18 years of age and older whose hospital admission was coded as pneumonia as per ICD-10 codes J10-18 were eligible for the study. Case notes were examined and patients were included if the emergency physician documented a diagnosis of pneumonia or lower respiratory tract infection. At the time of this study, this hospital did not have restrictions on the use of antibacterial agents, nor did it have on-line access to antimicrobial guidelines via the hospital intranet, but staff did have general internet access. No staff were employed for the purpose of promoting antimicrobial stewardship and there was only one full-time infectious disease physician (private practitioner).

Patients were excluded from the study if they were neutropenic, had suspected nosocomial infection, had suspected aspiration pneumonia, or had commenced antibiotic therapy prior to ED presentation. Patients were also excluded if they had a concurrent infection that may have influenced antibiotic prescribing and hospital length of stay. Patients who had treatment initiated at another hospital and were transferred to the study hospital or admitted directly onto the ward from a specialist outpatient clinic were also excluded.

For each patient the following data were collected: age; gender; date of birth; time of ED triage; triage allocation as per the Australasian Triage Score ⁴ (range 1–5, 1—most acute, 5—least acute); the highest pneumonia severity confusion, oxygen, respiratory rate, and blood pressure (CORB) score while in the ED up to 24 hours after arrival (as per CORB score definition, confusion+/−, oxygen saturation <90%, respiratory rate ≥30, and systolic blood pressure [BP] <90 mmHg or diastolic BP ≤60 mmHg; range 0–4, 0—least severe, 4—most severe) ⁷ ; time of first antibiotic administration; time of second antibiotic administration; time of administration of antibiotic with action against Streptococcus pneumoniae (Streptococcus cover); time of administration of antibiotic with action against atypical respiratory organisms, such as Legionella, Mycoplasma, and Chlamydophila (atypical cover); choice of first and second antibiotic; chest radiograph or chest computed tomography (CT) report; mortality; and length of stay.

Patients whose chest radiographs or chest CT scans were subsequently reported as not having consolidation were included in the analysis of time to antibiotics and choice of antibiotics, as the emergency physician had treated them as though they had pneumonia. These patients were not included in analyses evaluating length of stay and mortality.

Antibiotic prescribing was audited against the Queensland Health (State Public Health Provider) guidelines ⁸ and the National Antimicrobial Guidelines. ¹ Both guidelines recommend specific antibiotics depending on the severity of the pneumonia as measured by the CORB score ⁷ and differ only in regard to the recommended macrolide in CORB score 0–1. Prescribing was deemed to be concordant when the guidelines of either resource were met; antibiotics recommended by the guidelines are shown in Table 1. The choice of antibiotic needed to be correct, as did the dose, frequency, and route of administration. If a patient had an allergy to the recommended first-line antibiotic and was prescribed the recommended second-line antibiotic, then prescribing was deemed to be concordant.

Data were analyzed with the statistical program R. Relationships between continuous variables were measured with linear regression and relationships with binary outcomes were examined using logistic regression. p-Values of <0.05 were deemed significant. Ethics approval was granted from the Hospital's Research and Ethics Committee.

Results

One hundred and thirty-two patients met the inclusion criteria. Two patients were subsequently excluded due to lack of documentation, leaving 130 cases in the study. One patient was initially thought to have a pulmonary embolus; however, a CT pulmonary angiogram showed consolidation. This patient was excluded from the evaluation of the time to antibiotic administration but was included in the evaluation of antibiotic prescribing. Sixteen patients were excluded from analyses involving length of stay and mortality. Eleven of these patients had no consolidation seen on either their chest radiograph or CT scan and radiology reports were not available for 5 patients.

Patient demographics are shown in Table 2. A summary of the timing of antibiotic administration is shown in Table 3. All patients had their antibiotics started in the ED. One patient died as an inpatient, giving a mortality rate of 0.9% (1/114).

CORB severity scores and triage allocation are shown in Table 4.

Discussion

Antibiotic stewardship is a global problem. Similar studies overseas have shown low concordance rates with local antibiotic guidelines.^22,34,36 Concordance with CAP antibiotic guidelines in this study was particularly low at only 6.9% of patients. The reason for the low concordance mostly relates to use of third-generation cephalosporins for nonsevere CAP rather than the recommended penicillin. When cephalosporins were the first-line recommendation, as in severe pneumonia, concordance rates were much higher. This suggests that clinicians have a single antibiotic agent that they use for all severity classes of pneumonia. McIntosh et al. also found a high use of broad-spectrum cephalosporins in their study of CAP treatment in 37 EDs in Australia. ²⁵ They highlighted that there is tension between the restrained use of ceftriaxone and the benefits of using an effective agent.

Cephalosporins were recommended first-line inpatient therapy for nonsevere CAP in Australia from 1992 until 1998. ² Guidelines later changed to recommend penicillin-based treatment and cephalosporins only in the case of allergy. ³⁸ Cephalosporins have been criticized on the basis that they lead to resistance and to the selection of multiresistant organisms. ¹⁰ Prior third-generation cephalosporin use has been shown to be a risk factor for infection with methicillin-resistant Staphyloccocus aureus, cephalosporin-resistant Enterobacter species, and nosocomial bacteremia with enterococci.^{14,18,19,27–29} Charles et al. showed that narrow-spectrum beta-lactam treatment was sufficient for the majority of patients with CAP in Australia and that the use of such treatment could potentially reduce the rates of antibiotic resistance. ⁹

Antimicrobial stewardship is like many interventions in infection control, contingent on clinician behavior. Doctors focus on the outcome of their individual patient at the time. This is not surprising and probably reflects the ethos and training of physicians. It is of significance that the mortality rate in this study was so low at 0.9% when similar studies of patients 65 years and older with CAP found inpatient mortality rates of 7% and 16.3%. Forty patients in our study were under the age of 65 and a large proportion of patients had a CORB score of 0 (36%) giving them a lower expected mortality rate. ⁷ It is possible that this hospital admitted patients who would have been declined admission to the public hospitals where most studies have previously been conducted, on the basis that they were not unwell enough to warrant admission. This may be a finding of private hospitals in general. The excellent outcomes in this study pose a barrier to convincing clinicians to change their antimicrobial prescribing.

The use of cephalosporins for the treatment of CAP is a universal phenomenon. ²⁵ This is not altogether surprising when third-generation cephalosporins have an appropriate spectrum of activity for community-acquired pathogens; they are nontoxic and easy to administer. ¹⁰ One could be forgiven for asking whether the use of cephalosporins as primary treatment for CAP in Australian EDs poses such a risk that clinicians should be dissuaded from using them. This is a difficult question to answer and one that probably deserves debate from the medical community.

Concordance with antibiotic guidelines in this Australian private hospital was particularly low when compared with studies that evaluate antibiotic prescribing in CAP in predominantly public Australian hospitals. Two similar studies undertaken in primarily public Australian hospitals found concordance rates of 18% and 20%.^24,25 In 2007/8, private hospitals in Australia treated 40% of all hospital inpatients ³⁰ and given the likely further increase in the number of patients treated in the private health system; antimicrobial stewardship within the private sector is of upmost importance. The primary restriction to antibiotic prescribing in private hospitals is the PBS. The PBS limits the use of third-generation cephalosporins to the treatment of meningitis or where septicemia with a sensitive organism is suspected. About 4–25% of CAP caused by classical pathogens will present with bacteremia so the use of cephalosporins in this circumstance is not a breach of PBS restrictions.^11,31 A subtle change to the PBS clause to exclude the setting of pneumonia would resolve the issue. Education interventions, mostly conducted in public hospitals, have also been found to increase rates of adherence to CAP guidelines in Australia. ²⁵ If such education interventions were utilized by private hospitals, then this could potentially further improve guideline adherence.

Studies of concordance with antibiotic guidelines internationally have shown a high inappropriate use of fluoroquinolones.^17,34,36 In our study fluoroquinolones were administered only in the case of immediate penicillin hypersensitivity as per the guidelines. This suggests that clinicians are aware of the need to reduce their use of fluoroquinolones.

Increasing age was associated with increased time to antibiotic delivery. Older patients do not always present with classical respiratory symptoms.^6,13 This has the potential to lead to a delay in diagnosis with a subsequent delay in the time to antibiotic administration. It is difficult to explain why the triage score did not correlate with either the CORB score or the time to antibiotic administration. The Australasian Triage Scale stipulates how quickly patients with specific triage scores should receive medical assessment and treatment. The triage score that a patient receives is based on the severity of their observations and presenting complaint when first assessed by the triage nurse in the ED waiting room. One would have expected patients with higher CORB scores to have been given lower triage scores and to have consequently received their antibiotics earlier. The lack of association between the CORB score and the triage score may be explained by the fact that the CORB score is by definition the worst score while the patient is in the ED, whereas the triage score is calculated at the time of arrival to the ED.

Eighty-one percent of patients received their first antibiotic within 4 hours of arriving in the ED. This result is better than previous studies from the United States where 54–66% of patients received their first antibiotic within 4 hours.^16,20,32 This hospital therefore compares favorably with hospitals internationally.

The study is limited in that it only looks at ED-initiated antibiotics and not ward-based prescribing. It was also not possible to determine whether the ED physician or the patient's attending doctor decided on the choice of antibiotic. The study also assumes that all patients admitted with pneumonia were coded appropriately. Patients initially diagnosed with pneumonia in the ED, who subsequently had their diagnosis changed, will have been missed by this study. Further, as the study was retrospective, it was not possible to determine whether antibiotics that were not concordant with guidelines had been chosen for a specific reason unless this was documented. However, it is highly unlikely that these limitations were sufficient to explain the low concordance rate that we found.

Further studies of other private Australian hospitals are required to determine whether the low concordance rate that we found is common in the private sector. If it is common, then there needs to be debate about increasing PBS restrictions on antibiotic prescribing.