Effect of a 3-Step Critical Pathway to Reduce Duration of Intravenous Antibiotic Therapy and Length of Stay in Community-Acquired Pneumonia:  A Randomized Controlled Trial FREE

In the United States, it is estimated that more than 4 000 000 people develop community-acquired pneumonia (CAP) every year.¹ Patients with CAP are primarily treated as outpatients, but patients who require hospitalization consume the greatest proportion of economic resources. Data from the United States showed that there were 1.3 million hospitalizations for pneumonia in 2005² and that the cost of care for patients with CAP, including both direct and indirect costs, has been estimated at more than $40 billion. Pneumonia in Europe is estimated to result in an annual expenditure of [euro]10.1 billion³; of this amount, inpatient care accounts for [euro]5.7 billion.

Length of hospital stay (LOS) is the most important component of the cost of CAP.^4- 5 Moreover, longer stay places patients at risk of complications such as phlebitis, pulmonary embolism, and nosocomial infection.⁶ Nevertheless, investigators have reported considerable variations in LOS for patients with CAP, suggesting that physicians do not use a uniform strategy to decide hospital discharge.^7- 10 The duration of intravenous (IV) antibiotic therapy is a major determinant of LOS. Therefore, switching from IV to oral therapy as soon as patients are clinically stable may help shorten LOS and reduce associated costs.^11- 13 However, patients with CAP often remain hospitalized after becoming clinically stable, and the maintenance of antibiotic IV therapy is a major limitation for discharge.¹⁴

In an era of increasing competition in medical care, institutions have embraced critical pathways as a strategy for decreasing cost and improving health care quality.^15- 16 Nevertheless, evidence from prospective controlled trials to evaluate the effects of critical pathways for CAP is scarce. We designed this randomized trial to test the hypothesis that the use of a 3-step critical pathway would be as safe as, and more effective than, usual care in reducing the duration of IV antibiotic therapy and LOS in hospitalized patients with CAP. The primary end point of the trial was LOS. Secondary end points were the duration of IV antibiotic therapy, adverse drug reactions, need for readmission, overall case-fatality rate, and patients' satisfaction.

This prospective, randomized trial was conducted at 2 tertiary hospitals in Barcelona, Spain, between May 1, 2005, and December 31, 2007: the Bellvitge Institute for Biomedical Research (IDIBELL)–Hospital Universitari de Bellvitge, a 900-bed university public hospital, and the SCIAS–Hospital de Barcelona, a 300-bed private hospital. The study was approved by the ethics committees of both institutions.

All immunocompetent patients 18 years or older who were diagnosed as having CAP in the emergency department were screened for eligibility. Patients with neutropenia (<500/μL) or human immunodeficiency virus infection or who had undergone transplantation or using immunosuppressive drugs were excluded. Community-acquired pneumonia was defined as the presence of an infiltrate on chest radiograph plus 1 or more of the following: fever (temperature, ≥38.0°C) or hypothermia (<35.0°C), new cough with or without sputum production, pleuritic chest pain, dyspnea, and altered breath sounds on auscultation.

Patients with CAP were stratified into risk classes according to the Pneumonia Severity Index.¹⁷ All patients in risk classes IV and V were considered for randomization. Patients in risk classes I, II, and III were also considered for randomization if they met 1 or more of the following: respiratory failure (PaO₂ <60 mm Hg, saturation ≤90% using pulse oximetry, or both), unstable vital signs (temperature >37.8°C, heart rate >100/min, systolic blood pressure <90 mm Hg), lack of response to previous antibiotic therapy (≥48 hours), metastatic infection, or concomitant unstable comorbid conditions necessitating hospitalization for treatment. Patients were excluded if they met 2 or more of the following: intensive care unit admission from the emergency department, imminent death, shock, complicated pleural effusion (empyema or large effusion), pregnancy, aspiration pneumonia, and severe social problems (eg, homeless, drug abuse, severe mental disorders).

An epidemiologist (X.C.) generated the random allocation sequence. Randomization was performed in computer-generated blocks of 10, with the randomization code kept by the clinical epidemiologist in a sealed envelope. The randomization was stratified according to hospital. In the emergency department, patients who met the study criteria and provided written informed consent were randomized by the infectious disease consultant, who opened the sealed, sequentially numbered, opaque envelopes.

Patients were enrolled and randomly assigned by investigators to follow a 3-step critical pathway or to receive usual care. To avoid potential biases due to the habitual practices of individual physicians, patients' attending physicians were divided into 2 groups: physicians who only had to treat patients randomly assigned to follow the 3-step critical pathway and physicians who only had to treat patients assigned to receive usual care. An epidemiologist (X.C.), who was blinded to the identity of the physicians, created 2 groups of 5 physicians with similar LOS. The groups were formed on the basis of the median LOS of patients with CAP attended by these physicians during the 2 years prior to the present study (median, 7.5 days).

The 3-steps of the critical pathway were (1) early mobilization of patients; (2) use of objective criteria for switching to oral antibiotic therapy; and (3) use of predefined criteria for deciding on hospital discharge. Early mobilization was defined as movement out of bed with a change from the horizontal to the upright position for at least 20 minutes during the first 24 hours of hospitalization, with progressive movement each subsequent day during hospitalization, as described elsewhere.¹⁸ Patients were switched from IV to oral therapy when they experienced clinical improvement and met the following objective criteria: ability to maintain oral intake; stable vital signs (considered as temperature ≤37.8°C, respiratory rate ≤24 breaths/min, systolic blood pressure ≥90 mm Hg without vasopressor support for at least 8 hours); and absence of exacerbated major comorbidities (ie, heart failure, chronic obstructive pulmonary disease) and/or septic metastases. Predefined criteria for hospital discharge were meeting criteria for switching to oral antibiotic, baseline mental status, and adequate oxygenation on room air (PaO₂ ≥60 mm Hg or pulse oximetry ≥90%). For patients with chronic hypoxemia or receiving chronic oxygen therapy, PaO₂ or pulse oximetry measurement had to be similar to their baseline values. Criteria for switching to oral antibiotic therapy and hospital discharge could be met simultaneously or sequentially.

A printed checklist detailing the 3-step pathway was added to the medical chart of patients assigned to this study arm to remind attending physicians of the necessity of early mobilization and also to remind them of the criteria for switching to oral antibiotic therapy and for deciding on hospital discharge. Patients randomly assigned to receive usual care were treated according to the standard practices of individual attending physicians.

The primary end point of the trial was LOS. Secondary end points were the duration of IV antibiotic therapy, adverse drug reactions, need for hospital readmission in the 30 days after randomization, death from any cause in the 30 days after randomization, and patients' satisfaction with the care received for pneumonia.

Empirical antibiotic therapy was administered in the emergency department in accordance with the hospital's guidelines, which recommend the administration of a β-lactam agent (ceftriaxone sodium or amoxicillin sodium–clavulanate potassium) with or without a macrolide or fluoroquinolone. Combination therapy was recommended for patients with severe CAP.¹ Levofloxacin monotherapy was indicated for Legionella pneumonia and for selected cases.

Patients were seen daily during their hospital stay by attending physicians and by at least one of the investigators. The investigators assessed and recorded all the primary and secondary outcome measures. Length of hospital stay was measured in days and was calculated as the time from the admission date to the date of discharge. Duration of IV antibiotic therapy was also measured in days and was calculated as the time from the initial dose of antibiotics in the emergency department to the last dose of the IV antibiotics. Patients were followed-up at the outpatient clinic 30 days after hospital discharge. All assessments were made using a standard protocol with a checklist of items. The investigators recorded readmission for any reason within 30 days after pneumonia diagnosis. This information was obtained from a specific search for hospital readmission in the admission databases of both hospitals and checked by asking patients at the final outpatient 30-day visit. Overall case fatality rate was defined as death due to any cause less than 30 days after hospitalization.

Patients' satisfaction with their overall care for pneumonia was evaluated at hospital discharge in response to the question “How would you rate your overall care for this episode of pneumonia?” as previously reported.¹⁹ Responses were recorded on a scale of 1 to 5, from “very unsatisfactory” to “very satisfactory.” Patients were considered satisfied if the response recorded was 4 or 5.

Samples obtained per protocol consisted of 2 sets of blood cultures, a sputum sample when available, urine for detection of antigens, and paired acute and convalescent serum samples.²⁰Streptococcus pneumoniae antigen in urine was detected using a rapid immunochromatographic assay (BinaxNOW; Binax Inc). Legionella pneumophila serogroup 1 antigen in urine was detected using a commercial immunoenzymatic method (enzyme-linked immunosorbent assay [Bartels ELISA]; Trinity Biotech). Serological studies were performed by standard methods to determine antibodies against atypical agents (Mycoplasma pneumoniae, Chlamydophila psittaci, Chlamydophila pneumoniae, and Coxiella burnetii).

We estimated that we would need a total sample size of 380 patients to achieve 82% power at a 5% significance level using paired t tests to detect a 1.5-day difference in LOS between 2 treatment groups. After assuming a priori that up to 5% of patients would not be evaluable, we set the sample size target for randomization at 200 patients per treatment group.

To assess differences in the frequency of outcomes in the 2 groups, descriptive statistics were calculated for all variables. Categorical variables were compared in the 2 groups using the χ² or Fisher exact test, overall and for each hospital, and continuous variables were compared using the Mann-Whitney test. Percentage differences of each outcome and mean differences between the 2 groups, with corresponding 95% confidence intervals, were also computed and presented. Data for the primary and secondary end points were analyzed on an intention-to-treat and per protocol basis. The intention-to-treat analysis included all randomly assigned patients. Because both analyses produced virtually the same results, only the intention-to-treat analysis is presented in detail. Statistical significance was established at the .05 α value.

We assessed 601 consecutive patients for eligibility, of whom 200 were excluded (Figure). A total of 401 patients were randomly assigned and included in an intention-to-treat analysis for the primary and secondary end points. Of these, 200 were assigned to follow the 3-step critical pathway and 201 received usual care. The baseline characteristics of the patients in the 2 treatment groups were similar (Table 1).

A cause was established in 111 of 187 patients (59.4%) in the 3-step critical pathway group and in 109 of 191 patients (57.1%) in the usual care group who had pneumonia. The distribution of causative organisms did not differ between groups. Streptococcus pneumoniae (85 patients in the 3-step critical pathway group vs 79 in the usual care group) and L pneumophila (13 vs 16 respectively) were the most frequently isolated pathogens, followed by Haemophilus influenzae (10 vs 10) and by atypical agents (3 vs 4).

No differences were found regarding the median (range) time to institution of antibiotic therapy between groups (3.3 [1-13] days vs 4.0 [1-20] days; P = .45). Most patients were initially treated with combination antibiotic therapy (112 vs 111 patients). The regimens most frequently prescribed were β-lactam plus levofloxacin (65 vs 61 patients) and β-lactam plus macrolide (47 vs 49 patients). A single antimicrobial agent was given to 88 patients in the 3-step group and 90 patients in the usual care group. The antimicrobials most frequently administered as monotherapy were ceftriaxone (39 vs 32 patients), amoxicillin-clavulanate (18 vs 20 patients), and levofloxacin (24 vs 34 patients).

Table 2 details outcomes for study patients. In the intention-to-treat analysis, the median LOS was 3.9 days in the 3-step group vs 6.0 days in the usual care group (difference, −2.1 days; 95% CI, −2.7 to −1.7; P < .001). The median duration of IV antibiotic therapy was 2.0 days in the 3-step group and 4.0 days in the usual care group (difference, −2.0 days; 95% CI, −2.0 to −1.0; P < .001). Equivalent results regarding the LOS and the median duration of IV antibiotic therapy were obtained in the per-protocol analysis and when analyzing only the cases microbiologically documented.

Early mobilization was not performed in 8 patients in the 3-step critical pathway group. Six of these patients felt too ill to be mobilized, and 2 had severely altered mental status. Adverse drug reactions, mainly phlebitis, were more frequent in the usual care group (Table 2). In the intention-to-treat analysis, there were no differences between groups regarding the development of in-hospital complications and overall mortality. In the per-protocol analysis, the results were equivalent. Causes of death were respiratory failure (3 patients) and multiorgan failure (1 patient) in the 3-step critical pathway group and cancer (1 patient) and sudden death (1 patient) in the usual care group.

As detailed in Table 2, the numbers of patients required readmission (<30 days) were similar in the 2 groups. In the 3-step critical pathway group, the reasons for readmission were as follows: exacerbation of chronic obstructive pulmonary disease (7 patients), cancer (2 patients), empyema (2 patients), acute asthma (2 patients), cerebrovascular disease (1 patient), retinal detachment (1 patient), cholecystitis (1 patient), uncomplicated pleural effusion (1 patient), and pneumonia (1 patient). In the usual care group the reasons for subsequent hospitalization were as follows: exacerbation of chronic obstructive pulmonary disease (5 patients), exacerbation of cardiac failure (3 patients), empyema (1 patient), ascites (1 patient), hepatic encephalopathy (1 patient), seizures (1 patient), dysphagia (1 patient), pneumonia (1 patient), and abdominal wall hernia (1 patient).

For the analysis of patients' satisfaction, data were available for 186 of 200 patients in the 3-step critical pathway group and for 174 of 201 patients in the usual care group. No differences were found in satisfaction between groups (4 or 5 points of the scale): 3-step critical pathway group, 176 of 186 (94.6%), vs usual care group, 164 of 174 (94.3%); absolute difference, 1.4 percentage points (95% CI, −2.7 to 5.4 percentage points) (P = .60).

In this randomized trial, we found a 3-step critical pathway including early mobilization and use of objective criteria for switching to oral antibiotic therapy and for deciding on hospital discharge to be safe and effective in reducing duration of IV antibiotic therapy and LOS compared with usual care.

Controlled clinical trials to evaluate the efficacy of interventions for decreasing the duration of IV therapy and LOS for patients hospitalized with CAP are scarce, and those published have produced mixed results. Some studies support their efficacy,^13,18,21 but others do not.^22- 23 Our randomized trial differs from previous investigations in that the intervention arm consisted in the application of an easy-to-perform 3-step critical pathway, with early mobilization as the first step.^14,22- 24 The mechanism by which early mobilization contributes to reducing LOS is unknown. It has been hypothesized that in mobilization from horizontal to upright position there may be improvement in aeration and/or blood flow redistribution with optimized drug delivery to the site of infection, reduced risk of aspiration, and maintenance of functional health status.¹⁸

The second step of our critical pathway comprised the use of objective and simple bedside criteria for the early switch from IV to oral antibiotics. Although the duration of IV treatment is a key determinant of LOS, strategies of early switching to oral antibiotic have mainly been evaluated in observational studies^11- 12,25 but less frequently in randomized trials.^13,26- 27 Finally, the third step of our intervention arm was based on the use of objective criteria for clinical stability and to decide appropriateness for hospital discharge. In this regard, other investigators have shown that once stability is achieved in patients with CAP, the risk of serious clinical deterioration is 1% or less, even in the sickest subgroup of patients.²⁸

In recent years, there has been concern that efforts to reduce LOS would increase the proportion of patients being discharged “sicker and quicker,” who may thus experience an increased risk of adverse outcomes.^29- 30 In our study, the 2-day decrease in LOS in the 3-step group was not significantly associated with a greater number of hospital readmissions or a higher overall mortality compared with the usual care group. It should be noted, however, that patients receiving usual care were more likely to experience adverse drug reactions, mainly phlebitis, probably related to the longer duration of IV antibiotic therapy in this group.

In an era of cost containment and resource constraints in health care systems, cost-effective health care delivery is of paramount importance.³¹ The economic burden associated with CAP remains substantial, and LOS is the most important driver of the cost in hospitalized patients.³² In a recent study carried out in the United States, it has been estimated that eliminating a day during the course of a CAP admission is potentially worth $2273 to $2373 in economic benefits.³³ Therefore, our finding that the application of a 3-step critical pathway reduced the LOS by 2 days compared with usual care may have significant economic implications.

Our study has several limitations. First, outcomes were assessed by investigators who were aware of patient treatment assignments. Nevertheless, the LOS and the duration of IV antibiotic therapy were ascertained using objective data. Second, it is possible that practices among physicians treating patients in the usual care group may have been influenced by interactions with physicians applying the 3-step pathway during the course of the study. However, the influence of these interactions would probably have reduced the LOS in the control group rather than in the intervention arm. Third, our study was not powered to detect a survival difference. Fourth, the trial was not designed to evaluate the effectiveness of the separate components of the 3-step critical pathway. Finally, since about a third of the hospitalized patients with CAP were excluded, our conclusions apply only to the selected population analyzed.

In conclusion, in a population of immunocompetent adults with CAP requiring hospitalization, the use of a 3-step critical pathway was safe and effective in reducing the duration of IV antibiotic therapy and LOS and did not adversely affect patient outcomes.