Characterizing and developing strategies for the treatment of community-acquired pneumonia at a community hospital

1Department of Pharmacy, Lions Gate Hospital, North Shore Health Region, North Vancouver; 2Department of Pharmacy and Critical Care, Lions Gate Hospital, North Shore Health Region, North Vancouver, and Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver; 3Department of Respirology, Lions Gate Hospital, North Shore Health Region, North Vancouver; 4Department of Internal Medicine, Lions Gate Hospital, North Shore Health Region, North Vancouver, British Columbia Correspondence: Mr Mark Fok, Vancouver Hospital and Health Sciences Centre – University of British Columbia Hospital Site, Pharmacy Department, 2211 Wesbrook Mall, Vancouver, British Columbia V6T 2B5. Telephone 604-822-7010, fax 604-822-9742, e-mail mfok@vanhosp.bc.ca MC Fok, Z Kanji, R Mainra, M Boldt. Characterizing and developing strategies for the treatment of communityacquired pneumonia at a community hospital. Can Respir J 2002;9(4):247-252.

C ommunity-acquired pneumonia (CAP) is a disease with significant morbidity and mortality.Canadian statistics indicate that pneumonia and influenza are the sixth leading causes of death, while American data indicate that CAP is responsible for 500,000 hospitalizations, 45,000 deaths and US$9.7 billion in hospital costs each year (1,2).Age-adjusted mortality rates show this to be a growing trend (3).
In response to the implications of this disease and its economic burden on resources, Canadian and American CAP guidelines have been published to facilitate its efficient treatment (4-7).These guidelines provide a valuable resource for effective management of this disease and recommend an assessment of 30-day mortality risk using a pneumonia severity index (PSI) score, diagnostic tests to perform, and choices for empirical, targeted and step-down antibiotic therapy.By incorporating these guidelines into management strategies known as critical pathways, several studies have demonstrated a reduction in institutional resources without compromising outcomes in terms of mortality or patient well-being (8,9).
Lions Gate Hospital (LGH) is a community hospital with 260 acute and 325 extended care beds.Data from program management reports indicated that for the period of April 1, 1998 to March 31, 1999, there were 338 patients admitted to the acute care wards with a primary diagnosis of simple pneumonia and pleurisy.While their average length of stay (LOS) was 9.1 days, the average LOS at peer group hospitals (those hospitals with a similar bed capacity to LGH) was 7.9 days.This increase in LOS by 1.2 days amounts to an annual potential savings of 406 bed days.Furthermore, no hospital-based treatment protocols are available to assist physicians with the decision-making process.The objective of the present study was to assess the management of CAP at LGH and provide recommendations for improved management of the disease.

PATIENTS AND METHODS
A retrospective chart review of patients admitted between May 1, 2000 and August 31, 2000 was conducted using a standardized data collection form.The patients included were 18 years or older and had a primary admitting diagnosis of CAP, defined as an acute infiltrate on chest radiograph suggestive of pneumonia, plus the presence of at least two of the following symptoms of acute infection: fever or hypothermia, rigours, sweats, new cough with or without sputum production, change in colour of respiratory secretions, chest discomfort, dyspnea or ausculatory findings consistent with pneumonia (5).Patients were excluded if they were hospitalized in the 14 days before the onset of symptoms, if they resided in a long term care facility before admission or if they had an immune deficiency (ie, having a history or presence of infection with the human immunodeficiency virus, taking more than 10 mg/day of prednisone or other immunosuppressive agents, receiving active treatment for cancer, or having history of organ transplantation, active tuberculosis or cystic fibrosis).A PSI score based on the available laboratory data and physical findings on admission was calculated for each included patient.Data collected included the decision to hospitalize based on the calculated PSI score; whether guideline-recommended diagnostic tests (chest x-ray, sputum and blood cultures) were performed; antibiotic selection for pneumonia in terms of choice, route and cost; timing and appropriateness of step-down from intravenous to oral antibiotics; whether early discharge occurred; and LOS.To evaluate the timeliness and appropriateness of step-down and early discharge, step-down and early discharge criteria from the literature and Canadian CAP guidelines were used (Table 1).

RESULTS
Of the 93 charts reviewed, 51 patients met the inclusion criteria and 42 patients were excluded.Of the excluded patients, six (14%) were younger than 18 years old, seven (17%) did not meet the definition of CAP, 18 (43%) had a suspected nosocomial or long term care facility source for their infection, and 11 (26%) had an immune deficiency.
The average age of the patients was 75±15 years (mean ± SD), and there were more male than female patients (69% versus 31%, respectively).The average LOS for these patients was 9.9 days, with a median of five days and a range of one to 85 days.
The frequency of performing guideline-recommended diagnostic procedures (chest radiograph, and blood and sputum cultures) within 24 h of admission and the breakdown of positive culture results are illustrated in Figure 1.
Patients had an average PSI score of 101±33 (mean ± SD), and these scores are categorized into their respective classes in Table 2.The PSI, which stratifies patients into five mortality risk categories ranging from low to high risk, recommends inpatient treatment for patients in classes IV and V (mortality risk 8.2% to 31.1%) and outpatient treatment for patients in classes I, II and III (mortality risk 0.1% to 2.8%); class III patients could possibly be admitted for a brief observation period (5).As demonstrated in Table 2, eight patients were admitted as class III patients with borderline scores, which may suggest appropriate admission.However, the other eight patients in classes I and II could likely have been treated on an outpatient basis without affecting their mortality from a CAP perspective.The five deaths that occurred in classes IV and V were attributed to pneumonia, while the death of one person from class IV was attributed to a myocardial infarction.
Empirical antibiotic choices were consistent with the Canadian CAP guidelines in 27 patients (53%).These consisted of at least a second-generation cephalosporin plus a macrolide (Figure 2).Therapy was inconsistent with the guidelines in 24 patients (47%), with the prescription of cephalosporin monotherapy, macrolide monotherapy and regimens for suspected aspiration (Figure 3).
Twenty patients (39%) underwent step-down to oral antibiotics (Figure 4).Step-down could have occurred earlier in 11 of these patients according to the predetermined step-down criteria.An additional 12 patients (24%) met the criteria for step-down, but remained on intravenous antibiotics for the duration of their course.Step-down was not applicable in 19 patients (37%), with the reasons pro-

DISCUSSION
The present study suggests that variability exists in the management of CAP at LGH. Areas for improvement were identified as the admission decision, which should use PSI scores as a tool to predict patients at high risk of mortality; antibiotic selection, which should have more consistency in guideline-recommended antibiotic regimens; and step-down, which should be timely and thus potentially lead to early discharge.
The mean LOS for the 51 patients included in the present study appeared to be consistent with program management reports (9.9 versus 9.1 days, respectively).However, the corresponding median LOS was calculated to be 5.0 days; the differences in the mean and the median were largely attributed to five patients who remained in the hospital for prolonged periods of time (15 to 85 days) for reasons other than pneumonia (for instance, patients may have been awaiting placement into an appropriate care facility or requiring further investigations for another dis-ease process).Thus, the median LOS is more reflective of the actual LOS for our pneumonia patients.Because the data from program management reports did not contain information on the median LOS for pneumonia, no comparisons with peer group hospitals were possible.Nonetheless, the calculated median LOS of 5.0 days for the present 51 patients compares favourably with recent Canadian CAP LOS data in the literature, which report an overall median LOS of 7.0 days (interquartile range of 4.0 to 11.0 days) for both community and teaching hospitals (10).
The PSI score, which is a validated, prognostic indicator of 30-day mortality based on specific risk factors and laboratory values, has increasingly been used as a tool to facilitate the decision to admit patients (4,5).The mortality results from the present study appear to correlate with validated data in the literature, which report 30-day mortality rates of 8.2% to 9.3% for class IV patients and of 27.0% to 31.1% for class V patients.Closer examination of the PSI scores for the eight patients in classes I and II, who may have been admitted inappropriately, revealed that they had no significantly abnormal laboratory results nor any comorbid illnesses requiring a workup on presentation.Thus, by extrapolating the rate of possibly inappropriate admissions (eight of 51 patients) to the average number of CAP admissions per year (340 admissions/year based on inhouse data), 53 admissions could potentially be prevented, with a savings of approximately 265 bed days annually (median LOS of 5.0 days/CAP admission).Although other patients do fill these beds, the amount that could have been reallocated based on these bed day savings is approximately $130,000 annually (cost savings/bed day equals $500/day).
The low yield of the blood and sputum cultures is surprising.The literature documents that blood cultures are positive in 8% to 10% of all admitted patients, while sputum cultures have ranged from 30% to 80% in their diagnostic usefulness (4).However, possible confounding factors such as the administration of antibiotics before admission or obtaining a specimen culture, not looking for Figure 2) Initial empirical antibiotic choices that were consistent with the Canadian community-acquired pnemonia guidelines in 51 patients with community-acquired pneumonia at Lions Gate Hospital, North Vancouver, British Columbia a specific pathogen or poor sampling techniques may have played a role in these results.The pathogens identified appear to be consistent with reports in the literature.However, one patient with asthma grew Enterobacter cloacae, and no atypical pathogens were identified, possibly due to a combination of their historically low incidence at LGH and the lack of specific tests ordered for organisms such as Legionella pneumophilia.
The rate of choosing initial antibiotic therapy at LGH that is consistent with the Canadian guidelines is lower than the estimate for Canadian hospitals reported in the literature (53% and 79.8%, respectively) (10).Furthermore, our data show that once patients reached the ward, therapy was altered in 15 patients (29%) (for example, combination therapy with a second-generation cephalosporin plus macrolide initiated in the emergency department was changed to macrolide monotherapy on the ward and viceversa).The potential reasons for lower compliance with practice guidelines and discrepancies in what entailed adequate coverage between physicians in the present study are multifold.First, not all physicians may be aware of the latest practice guidelines, and a lack of familiarity may exist with those who are aware of them.Second, physicians may believe certain regimens are more suitable based on their success with the regimen, despite the practice guidelines.Finally, hospital trends for CAP pathogens and the prevalence of relatively uncommon pathogens may also play a role in determining what constitutes adequate therapy.Although 'respiratory' quinolones (eg, levofloxacin, moxifloxacin and gatifloxacin) were not on formulary at LGH at the time of this study, other guideline-recommended oral antibiotics were available for step-down.They included oral first-and second-generation cephalosporins (cephalexin, cefaclor, cefuroxime), as well as doxycycline, penicillins, erythromycin, azithromycin and clarithromycin.Four patients were prescribed empirical regimens with anaer-obic coverage to cover for suspected aspiration (Figure 3).Because there was suspicion of aspiration, adding anaerobic coverage to the recommended antibiotic regimes for CAP seemed reasonable.However, because there was no atypical coverage with a macrolide, therapy was considered to be inconsistent with the guidelines.
Appropriately applying the predetermined step-down criteria could have decreased the duration of intravenous antibiotics by 2.5 days for 23 of 51 applicable patients, whose median duration of intravenous antibiotics was 4.3 days.The cost savings associated with this reduction of intravenous antibiotics using the equation: cost of intravenous antibiotics -cost of oral equivalent step-down antibiotic = cost savings) amounted to a total savings of approximately $1,300 or $57/applicable patient in this study.The cost of intravenous antibiotics was based on actual acquisition cost and administration costs (preparation and administration of drug).By extrapolating these cost savings to the average number of CAP admissions per year (340 admissions/year), the projected annual cost savings was determined to be roughly $8,700.The number of patients meeting early discharge criteria was 20 of 51, with 25 potential bed days saved (1.25 bed days/patient).By applying these rates to the average number of CAP admissions per year (340 admissions/year based on in-house data), 167 bed days could potentially be saved annually with an estimated annual cost avoidance of $83,000 (cost savings/bed day equals $500/day).
Due to the retrospective nature of the present study, there are inherent limitations.Incomplete or inadequate documentation might have hindered the results.However, many of the end points used in this study were objective in nature and were always documented (for instance, LOS, antibiotic regimens selected, and criteria such as temperature and complete blood counts).Of the 19 parameters used for calculating the PSI, the only parameters that were not available for certain patients were arterial blood gases (ABGs) to determine arterial pH and/or the partial pressure of oxygen.Although the ABGs could have been abnormal and could have influenced the PSI scores from one class to another, if there was no clinical suspicion of an abnormality and measurements of ABGs were not ordered, it is likely that these parameters would not have been abnormal.Our data reflect all of the laboratory information selected by physicians that was used to make an informed clinical assessment.No points were assigned for ABGs not measured on admission.
The strengths of this study included strict criteria for the definition of CAP, which incorporated patient symptoms and a positive chest radiograph suggestive of pneumonia; objective data parameters; and predetermined criteria for step-down and early discharge based on recent literature.
While the median LOS at LGH compared favourably with other Canadian hospitals, the findings of the present study suggest that improvements in areas such as the admission decision, diagnostic tests and the pharmacotherapeutic management of CAP can be made.This led to the recommendation that a CAP preprinted order sheet be created to provide consistency for the management of this disease.The preprinted order sheets incorporated a PSI scoring table as a resource to facilitate the admission decision, recommendations for diagnostic and laboratory tests, suggestions for empirical antibiotic choices based on the Canadian CAP guidelines, intravenous to oral antibiotic step-down criteria, suggestions for oral antibiotic regimens and early discharge criteria.Overall, the potential cost avoidance was estimated to be $220,000/year, which suggests a significant impact on treatment strategies.It is anticipated that this preprinted order sheet will improve the management of CAP at LGH without adversely affecting patient outcomes.

Figure 1 )
Figure 1) The frequency of performing guideline-recommended diagnostic procedures (chest radiograph, and blood and sputum cultures) within 24 h of admission and the breakdown of positive culture results on 51 patients with community-acquired pneumonia at Lions Gate Hospital, North Vancouver, British Columbia

Figure 3 )Figure 4 )
Figure 3) Initial empirical antibiotic choices that were inconsistent with the Canadian community-acquired pnemonia guidelines in 51 patients with community-acquired pneumonia at Lions Gate Hospital, North Vancouver, British Columbia

Strategies for CAP treatment at a community hospital Can
Respir J Vol 9 No 4 July/August 2002 249

Fok et al Can
Respir J Vol 9 No 4 July/August 2002 250