Does a bundle of evidence-supported treatments (including adjunctive corticosteroids) improve the outcomes of patients with community-acquired pneumonia under conditions of routine care?
In this stepped-wedge, cluster-randomized effectiveness trial of 816 patients hospitalized with community-acquired pneumonia, a bundled intervention aiming to optimize treatment with corticosteroids, early mobilization, early switch to oral antibiotics, and screening for malnutrition had no significant effect on length of hospital stay, readmissions, mortality, or other complications of community-acquired pneumonia and resulted in a higher incidence of gastrointestinal bleeding when compared with usual care.
This process-of-care bundle failed to improve outcomes and may have increased risks of adverse events; it cannot be recommended for patients hospitalized with community-acquired pneumonia.
Community-acquired pneumonia remains a leading cause of hospitalization, mortality, and health care costs worldwide. Randomized clinical trials support the use of adjunctive corticosteroids, early progressive mobilization, antibiotic switching rules, and dietary interventions in improving outcomes. However, it is uncertain whether implementing these interventions will translate into effectiveness under routine health care conditions.
To evaluate the effectiveness of a bundle of evidence-supported treatments under conditions of routine care in a representative population hospitalized for community-acquired pneumonia.
Design, Setting, and Participants
A double-blind, stepped-wedge, cluster-randomized clinical trial with 90-day follow-up was conducted between August 1, 2016, and October 29, 2017, in the general internal medicine service at 2 tertiary hospitals in Melbourne, Australia, among a consecutive sample of patients with community-acquired pneumonia. The primary analysis and preparation of results took place between May 14 and November 25, 2018.
Treating clinical teams were advised to prescribe prednisolone acetate, 50 mg/d, for 7 days (in the absence of any contraindication) and de-escalate from parenteral to oral antibiotics according to standardized criteria. Algorithm-guided early mobilization and malnutrition screening and treatment were also implemented.
Main Outcomes and Measures
Hospital length of stay, mortality, readmission, and intervention-associated adverse events (eg, gastrointestinal bleeding and hyperglycemia).
A total of 917 patients were screened, and 816 (351 women and 465 men; mean [SD] age, 76  years) were included in the intention-to-treat analysis, with 401 patients receiving the intervention and 415 patients in the control group. An unadjusted geometric mean ratio of 0.95 (95% CI, 0.78-1.16) was observed for the difference in length of stay (days) between the intervention and control groups. Similarly, no significant differences were observed for the secondary outcomes of mortality and readmission, and the results remained unchanged after further adjustment for sex and age. The study reported higher proportions of gastrointestinal bleeding in the intervention group (9 [2.2%]) compared with the controls (3 [0.7%]), with an unadjusted estimated difference in mean proportions of 0.008 (95% CI, 0.005-0.010).
Conclusions and Relevance
This bundled intervention including adjunctive corticosteroids demonstrated no evidence of effectiveness and resulted in a higher incidence of gastrointestinal bleeding. Efficacy of individual interventions demonstrated in clinical trials may not necessarily translate into effectiveness when implemented in combination and may even result in net harm.
ClinicalTrials.gov identifier: NCT02835040
Lower respiratory tract infections, including community-acquired pneumonia (CAP), represent the third leading cause of lost disability-adjusted life years and the leading infectious cause of mortality worldwide.1 In high-income countries, CAP’s burden concentrates predominantly in the elderly,2-4 driving high and rapidly increasing rates of hospitalization and other health care costs.5-7 Challenges to managing CAP now relate to high rates of multimorbidity, which increases the complexity of care and is independently associated with mortality, longer hospitalization, risks of readmission, and poor functional outcomes.8-10
Quiz Ref IDIn recent years, data from randomized clinical trials (RCTs) have emerged demonstrating improved patient and health-system outcomes in CAP for several interventions, including routine early mobilization,11 rules regarding switching from parenteral to oral antibiotics,12,13 and dietary interventions.14 More recently, several RCTs evaluated corticosteroids as adjunctive treatment for CAP,15,16 followed by numerous meta-analyses of RCT data, including a Cochrane review,17-20 showing faster clinical recovery, shorter hospitalization,16,20 and reduced risk of treatment failure15 and death (in severe disease).19 However, we observe a wide gap between evidence and actual clinical practice in which many of these interventions are poorly implemented or have not been incorporated into guidelines despite high-level supportive evidence (corticosteroids).17-20 A reluctance to do so may reflect challenges in applying findings of RCT data to real-world settings because of concerns regarding the representativeness, generalizability, and external validity of existing clinical trial data: will efficacy demonstrated in highly selected research populations translate to effectiveness in routine clinical practice?21
We therefore designed an effectiveness evaluation in a representative population of mostly elderly hospitalized patients with high rates of comorbidity. We chose to do this by using a process-of-care intervention that would improve adherence to a bundle of 4 interventions for which there was existing level 1 or 2 evidence of improved outcomes in patients with CAP. A stepped-wedge, cluster-randomized, controlled design enabled both phased implementation and the use of established statistical approaches to compare control and intervention groups while minimizing the potential for bias and confounding.22
The study rationale and design have been reported in detail previously23 (trial protocol in Supplement 1). In brief, the Improving Evidence-Based Treatment Gaps and Outcomes in Community-Acquired Pneumonia (IMPROVE-GAP) trial was an investigator-initiated, double-blind, stepped-wedge, cluster-randomized, Hybrid 1, effectiveness-implementation quality improvement trial.24 It was approved by the Melbourne Health Human Research Ethics Committee (HREC) (2016.014). Because the intervention was a bundle of treatments, with each having a high level of evidence supporting their efficacy (therefore effectively constituting a “best-care” package), and because the intervention was a process-of-care approach delivered at scale (rather than on an individual basis), a waiver of consent to participate was sought and granted by the overseeing HREC.23 This approach maximized representativeness and generalizability.
Study Participants and Setting
The study was conducted between August 1, 2016, and October 29, 2017, at 2 tertiary, publicly funded metropolitan teaching hospitals in Melbourne, Australia, where approximately 80% of inpatient care of all adults with CAP is provided by 8 generalist internal medical (GIM) units. Criteria for admission to a GIM unit include presence of comorbidities, and patient allocation to a unit is determined by the weekday of admission. The mean age of patients with CAP in GIM units (approximately 75 years) is similar to that of Australia’s overall CAP hospitalization burden.25
All patients with CAP who were admitted to a GIM unit (according to diagnostic criteria16,26; see the trial protocol in Supplement 1) were eligible to be included in the study. To optimize representativeness and generalizability, inclusion criteria were broad and relied on routine diagnosis (including interpretation of chest radiographs) by treating clinicians. However, patients were excluded if palliative care was initiated on admission or if they were enrolled in a concurrent inpatient research study (including prior enrollment in IMPROVE-GAP within <90 days). Patients were eligible even if they were already being prescribed corticosteroids (regardless of the dosage or indication). Participants were withdrawn if they were transferred to an alternate medical specialty or health service within 24 hours of enrollment. A required sample size of 80 patients per GIM unit (640 patients total) was calculated to detect a clinically important decrease in the proportion of patients with a length of stay (LOS) greater than the median LOS from 36% to 20%, assuming an intracluster correlation of 0.01, with 75% power and a 5% significance level (see the statistical analysis plan in the trial protocol in Supplement 1).
The algorithm-based intervention protocol (the “CAP Service”) is described in detail elsewhere23 and was applied in identical fashion 7 days per week. Briefly, it comprised 4 evidence-based interventions designed to align closely with original implementation in seminal studies, including the following: (1) routine corticosteroid prescription (50 mg of prednisolone acetate or equivalent for 7 days)16; (2) switch from parenteral to oral antibiotics according to predefined rules12; (3) early mobilization (sit out of bed >20 minutes on the day of admission and progressive mobilization daily)11; and (4) routine screening for malnutrition and targeted nutritional therapy.27Quiz Ref ID Predefined contraindications to corticosteroid prescription (active use of intravenous drugs, acute burn injury or gastrointestinal bleeding in the past 3 months, known adrenal insufficiency, pregnancy or breastfeeding, or severe immunosuppression)16 and early mobilization28 (physiological instability according to vital signs and oxygenation, acute clinical deterioration since last mobilization session, drowsiness or inability to follow commands, recent fall not yet assessed by medical staff, or new-onset chest pain) were applied. The CAP Service interventions were delivered as recommendations that could be complied with at the discretion of treating clinical teams.
The primary outcome was hospital LOS (days). Secondary outcomes were inpatient mortality, 30- and 90-day mortality, requirement for intensive clinical support (intensive care unit [ICU] admission and requirement for mechanical ventilation), hospital readmission within 30 and 90 days, protocol adherence, and intervention-related adverse events (hyperglycemia, gastrointestinal bleeding, fall or acute clinical deterioration during physiotherapy intervention, or adverse drug reaction requiring that the drug is stopped). In addition to these outcomes specified a priori in the statistical analysis plan (trial protocol in Supplement 1), to enable consistency with data presented in previous RCTs,16 we also collected observational data on complications of CAP (empyema, pleural effusion, acute respiratory distress syndrome, and cardiac and neurologic events) in both the control and intervention groups.
The unit of randomization was treating GIM unit. The 8 GIM units (4 per hospital) were randomized (through generation of an electronic random sequence by a biostatistician blinded to the intervention) within the stepped-wedge structure to receive either usual care (control group) or CAP Service–guided interventions (intervention group). After the initial 10-week control period, 2 units were transitioned to the CAP Service intervention model every 10 weeks, with all 8 units receiving the CAP Service interventions for the final 10 weeks of the study. The study schedule and methods of data collection are described in the protocol (trial protocol in Supplement 1). Adverse events were reported quarterly to the steering committee and supervising Human Research Ethics Committee.
Statistical analyses were prespecified as per the statistical analysis plan (trial protocol in Supplement 1). Primary analyses were conducted consistent with the intention-to-treat principle.29 Primary and secondary analyses were undertaken using linear (continuous outcomes) or logistic (binary outcomes) mixed-effects models, with intervention and time period as fixed effects and cluster as a random effect. Further adjustment for age and sex22,30,31 was also performed. We analyzed LOS as continuous (log-transformed) and binary (LOS longer than the health service’s historical mean diagnosis group related [International Statistical Classification of Diseases and Related Health Problems, Tenth Revision code, J18] LOS of 3 days). Patients who died during the index hospitalization were censored from the analyses of the primary outcome and the secondary outcomes corresponding to readmission (leaving 778 patients analyzed for these outcomes, including 399 receiving usual care and 379 receiving the CAP Service initiation). Sensitivity analyses as per the statistical analysis plan (trial protocol in Supplement 1) included adjustment for clinically plausible confounders, repeated enrollment of a single participant, and seasonal variation. Prespecified subgroup analyses were conducted to assess the effect of various comorbid illnesses using measurements for chronic pulmonary disease, diabetes, and the Charlson Comorbidity Index. Subsequent to our initial statistical analysis plan, 2 further subgroup analyses based on level of C-reactive protein at admission15 and CORB (confusion [acute], oxygen saturation ≤90%, respiratory rate ≥30 breaths/min, and blood pressure <90 mm Hg [systolic] or ≤60 mm Hg [diastolic]) pneumonia severity score were also added.32 Consistent with our statistical analysis plan, all subgroup analyses were limited to the primary outcome (LOS) and readmissions. All statistical analyses were conducted in Stata, version 14.2 (StataCorp).
Between August 1, 2016, and July 31, 2017, a total of 917 patients with CAP were admitted (mean [SD] age, 76  years); 85 were excluded and 16 withdrew, leaving 401 patients in the intervention group and 415 patients in the control group included in the intention-to-treat analysis (Figure 1; eTables 1 and 2 in Supplement 2). Baseline characteristics were similar in the control and intervention groups (Table 1).
Compliance With Evidence-Based Recommendations
Compliance with bundled, evidence-based interventions is shown in Table 2. A total of 105 patients in the control group (25.3%) received an initial corticosteroid prescription; however, only 14 patients in the control group (3.4%) had a prescription for the recommended full 7 days of treatment. In the intervention group, 292 individuals (of 385 [75.8%] without predefined contraindications) received an initial dose of prednisolone acetate 50 mg or higher, but 67 (22.9%) ceased treatment prior to the full 7 days. Reasons for clinicians not prescribing corticosteroids, not prescribing the full 7 days of corticosteroids, or modifying the dose of corticosteroids in patients enrolled in the intervention group are outlined in eTable 3 in Supplement 2.
Quiz Ref IDThe proportion of guideline-concordant de-escalation from parenteral antibiotics to oral antibiotics was high for both groups, although marginally higher in the intervention group than the control group (310 [77.3%] vs 287 [69.2%]) (Table 2). Between the control and intervention groups, the proportion of patients with appropriate early and progressive mobilization increased from 19.3% (n = 80) to 71.6% (n = 287), and the proportion of patients with nutritional assessment and intervention increased from 54.9% (n = 228) to 83.0% (n = 333).
Distributions of LOS were similar in the control and intervention groups (Figure 2A) but with slight differences according to different time periods (Figure 2B) and among the 8 clusters (Figure 2C). We observed an unadjusted geometric mean ratio of 0.95 (95% CI, 0.78-1.16) when LOS was analyzed as a continuous variable and an odds ratio of 0.95 (95% CI, 0.57-1.59) when LOS was analyzed as a binary (LOS>3) variable (Table 3).33 Therefore, no difference in the primary outcome of LOS between the control and intervention groups was observed. This finding remained unchanged after adjustment for age and sex. Subanalyses demonstrated no significant differences in LOS between intervention and control participants for any subgroup whether according to presence or absence of chronic pulmonary disease, diabetes, Charlson Comorbidity Index, disease severity, or C-reactive protein (>150 mg/L [to convert to nanomoles per liter, multiply by 9.524]) (eTables 4-7 and eFigure 1 in Supplement 2).
Minimal non–statistically significant differences between the intervention and control groups were observed for mortality and readmission at all time points (inpatient, 30 days, and 90 days) (Table 3 and eFigure 2 in Supplement 2). Overall mortality was 10.7% (87 of 816) at 30 days. Small proportions of patients required ICU support (12 of 816 [1.5%]) or mechanical ventilation (7 of 816 [0.9%]), with minimal differences between the intervention and control groups.
A 2-fold increase in new insulin prescriptions was observed in patients with diabetes in the intervention group (unadjusted odds ratio, 1.96; 95% CI, 0.73-5.25). The number of gastrointestinal bleeding events, while small, was marginally higher for the intervention group (9 events [2.2%]) compared with the control group (3 events [0.7%]; unadjusted estimated difference in mean proportions, 0.008; 95% CI, 0.005-0.010). No gastrointestinal bleeding events resulted in death. Only 1 patient in the control group (not receiving corticosteroids) required a transfusion, with gastric bleeding confirmed by results of endoscopy. Of 9 patients in the intervention group with bleeding events, 6 had received corticosteroids at the full recommended dose. Three patients received transfusions and underwent endoscopies, which all identified upper gastrointestinal sources of bleeding. Bleeding events are described further in eTable 8 in Supplement 2. Further observational data on pneumonia-specific complications are also presented in eTable 9 in Supplement 2.
Quiz Ref IDThis study found no evidence that this bundled intervention including adjunctive corticosteroids improved duration of hospitalization or other clinical outcomes among patients with CAP. This intervention was also associated with an increased incidence of gastrointestinal bleeding.
Prior to this study, evidence supporting all interventions used in our bundle appeared to be sound. Routine early mobilization has been shown to safely reduce duration of hospitalization11 and, when applied in conjunction with an early switch from parenteral to oral antibiotics, achieved a 2-day reduction in LOS.13 Systematic screening for risk of malnutrition and targeted nutritional therapy has been shown to reduce rates of readmission in malnourished medical inpatients.14 Although the issue of adjunctive corticosteroids has continued to be somewhat contentious, at least 17 RCTs (with >2000 participants) have now been performed and subjected to numerous meta-analyses by separate research groups, yielding consistent results interpreted in a similar fashion.17-19,34-36 The recent Cochrane review states that “people with CAP treated with corticosteroids had lower clinical failure rates (death, worsening of imaging studies, or no clinical improvement), shorter time to cure, shorter hospital stay, and fewer complications.”19 Meta-analyses have suggested that overall rates of serious adverse events are no higher or are even reduced with corticosteroids.18,19,34 However, this finding has not allayed concerns.37-39 Our finding of increased gastrointestinal bleeding reinforces these concerns. Although this finding comes in the context of a bundle with 3 other interventions, it is difficult to see how the noncorticosteroid components of the intervention (early mobilization, antibiotic switch rules, or dietary assessment and intervention) could have contributed to increased gastrointestinal bleeding, especially given that the magnitude of the effect was similar to the increased rates of gastrointestinal bleeding seen with corticosteroid use more broadly.40
Our study was carefully designed to enroll a study sample most representative of the overall burden of disease in our population and to deploy the intervention as it might realistically be implemented in real life as part of a guideline package designed to improve routine care. It therefore aimed to measure effectiveness rather than efficacy. We hoped this measurement would resolve the uncertainty that has prevented the large, existing body of clinical trial evidence from being translated into clinical practice. A particular strength was that our study circumvented the ascertainment bias that was problematic in previous RCTs.41 For instance, the largest previous RCT of adjunctive corticosteroids by Blum et al16 (n = 785) recruited 27% of screened patients with CAP, requiring a 4.5-year enrollment period at 7 hospitals. By comparison, 89.0% of patients with CAP (816 of 917) were recruited to our study over 12 months. The stepped-wedge, cluster-randomized design was an additional strength that enabled robust analytical methods, minimizing potential for bias or confounding while enabling a pragmatic deployment of the intervention. It is especially well suited for experimental evaluations of health system interventions’ effectiveness.33
Previous studies of corticosteroid use, early mobilization, and antibiotic switching rules are notable for consistently demonstrating reductions in LOS.11,13,19 This difference from our findings could be related to previous studies’ aforementioned issues of ascertainment bias but also to other health system factors influencing LOS. For instance, the 3-day median LOS in our study compares with 6 to 7 days in the adjunctive corticosteroid study by Blum et al.16 Another explanation is that the manner in which we deployed the intervention may have diluted any effect (whether advantageous or disadvantageous). For instance, a small percentage of patients in our control group (mostly those with coexisting chronic lung disease) received some corticosteroids, albeit generally at much lower doses and shorter courses than recommended in our intervention group. Adherence to corticosteroid prescribing in the intervention group was also incomplete. However, our design was valid as a means of measuring effectiveness (rather than efficacy) as it reflects a real-life scenario whereby guideline adherence is likely to be incomplete and subject to a variety of factors (eTable 3 in Supplement 2).
Our study has some limitations. Quiz Ref IDBecause we used 4 separate interventions in our bundle, our overall null finding now raises questions about the real-world usefulness of each component of our intervention. For instance, significant process improvements related to early mobilization did not appear to translate into improved LOS, and process improvements related to dietary interventions did not appear to translate into improved readmission rates, as we might expect from the literature.11,14 However, the bundled nature of our process intervention makes it difficult to comment on the effectiveness of each of the 4 individual interventions deployed. Although we would presume these interventions all act through very different pathways and are directed at separate outcomes (eg, reducing LOS as opposed to preventing readmission), it is possible, for example, that overall null findings reflected a combination of positive and negative effects of some of the 4 interventions cancelling one another out. However, when the 4 interventions were used in combination as part of the health system implementation bundle, the resulting process improvements did not translate to improved patient outcomes.
A significant problem with existing CAP studies relates to the validity and appropriateness of outcome measures used. In particular, time to clinical stability (a composite end point derived from clinical signs, including body temperature) is an outcome commonly used in previous studies but has been widely criticized when used in studies of corticosteroids because their antipyretic properties mean that patients are inherently likely to reach this end point sooner. It is therefore argued that this physiological end point does not necessarily mean an improvement in underlying disease or a patient’s overall well-being.42 Mortality is the most objective and commonly used end point to study but is likely to be poorly discriminating when most hospitalized patients with CAP have moderate disease.43 We therefore chose LOS as our primary predefined end point as it is objective, unambiguous, and valid as a measure of both patient outcome and health system performance.44 We specified mortality as a secondary end point, but even with its substantial sample size, our study was powered to detect only a relatively large mortality difference, so it is not surprising and probably not very informative that we failed to see any difference here. Existing clinical trial evidence suggests a mortality benefit from corticosteroids in severe CAP (ie, among patients in the ICU setting). Although our study included patients with a broad and representative range of disease severity and a large proportion with severe CAP (as evidenced by the 10.7% mortality at 30 days), it was not within the scope of our study to address the role of corticosteroids for CAP in the ICU setting given the bundled nature of the intervention and the low proportion of screened patients (5.5% in total) receiving ICU care. Ongoing studies will generate further data specifically in this setting (NCT01743755 and NCT01283009). Therefore, despite our findings of an overall lack of effectiveness, individual components of our bundled intervention may still have merit in selected populations and settings.
This study demonstrates how the efficacy of individual interventions demonstrated in clinical trials conducted in selective patient populations may not necessarily translate into effectiveness when broadly applied in combination under routine health care conditions. In fact, they may even result in net harm. It underscores a groundswell of opinion arguing that interventions showing efficacy in RCTs should also undergo real-world evaluation prior to integration into clinical practice guidelines and pathways.21 Our novel implementation research design using a stepped-wedge RCT method provides a precedent and template for future evaluations that address this need. The risks and lack of effectiveness of the bundle of care demonstrated in this study strongly suggest that it is not useful as a means of improving the care of patients with CAP.
Accepted for Publication: March 29, 2019.
Corresponding Author: Harin Karunajeewa, MBBS, PhD, General Internal Medicine Unit, Western Health, The University of Melbourne, Melbourne, Australia (firstname.lastname@example.org).
Published Online: July 8, 2019. doi:10.1001/jamainternmed.2019.1438
Author Contributions: Ms Lloyd and Dr Karunajeewa had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Lloyd, Karahalios, Janus, Skinner, Haines, Shackell, Karunajeewa.
Acquisition, analysis, or interpretation of data: Lloyd, Karahalios, Janus, Skinner, Haines, De Silva, Lowe, Ko, Desmond, Karunajeewa.
Drafting of the manuscript: Lloyd, Janus, Skinner, Karunajeewa.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Lloyd, Karahalios, Haines, De Silva, Karunajeewa.
Obtained funding: Janus, Skinner, Haines, Karunajeewa.
Administrative, technical, or material support: Lloyd, Janus, Skinner, Lowe, Shackell, Ko, Desmond, Karunajeewa.
Supervision: Lloyd, Janus, Skinner, Karunajeewa.
Conflict of Interest Disclosures: Ms Lloyd reported personal fees from Australian Government Research Training Scheme during the conduct of the study. Ms Lloyd and Drs Janus, Skinner, Shackell, Ko, and Karunajeewa reported receiving grants from HCF Research Foundation during the conduct of the study. No other disclosures were reported.
Funding/Support: This study received funding from grant number EJWH2015163 from the HCF Research Foundation. Western Health, the University of Melbourne, and Monash University provided salary support to the investigators. Dr Karunajeewa was supported by an Australian National Health and Medical Research Council Career Development Fellowship. Ms Lloyd was supported by an Australian Government Research Training Scheme Scholarship.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.
Group Information: The members of the Improving Evidence-Based Treatment Gaps and Outcomes in Community-Acquired Pneumonia (IMPROVE-GAP) Implementation Team at Western Health are as follows: Monica Turczyniak, BPhysio, Department of Physiotherapy, Western Health, Melbourne; Kamya Kameshwar, MBBS, Department of General Internal Medicine, Western Health, Melbourne; Ben Stevenson, MBBS, Department of General Internal Medicine, Western Health, Melbourne; Parul Bali, MBBS, Department of General Internal Medicine, Western Health, Melbourne; Elena Gerstman, BPhysio, Department of Physiotherapy, Western Health, Melbourne; and Kirsty May, BPhysio (Hons), Department of Physiotherapy, Western Health, Melbourne.
Data Sharing Statement: See Supplement 3.
Additional Contributions: The Western Health Physiotherapy Team—Cassandra Raios, BPhysio, Melanie Paykel, DPT, Joshua Warren, Sarah Miller, DPT, Codey Lyon, BPhysio, Kate Ryan, BPhysio, Emily Simek, BPhysio, Diana Truong, MPhysio, and Rachel Whitford, DPT, Department of Physiotherapy, Western Health, Melbourne—delivered the early mobilization interventions. Anne-Maree Kelly, MBBS, Joseph Epstein Centre for Emergency Medicine Research at Western Health, Melbourne, May-Lea Ong, MD, Department of General Internal Medicine, Western Health, Melbourne; and Clarice Tang, DPT, Department of Physiotherapy, Western Health, Melbourne, provided advice on the development of the project. Vanessa Carter, MND, and Allison Lauder, MND, Department of Nutrition, Western Health, Melbourne; provided advice on the design of the nutrition intervention. Julie Simpson, PhD, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, provided advice regarding the statistical analysis plan. Ms Turczyniak, Mrs Gerstman, Ms May, Dr Kameshwar, Dr Bali, Dr Stevenson, Ms Raios, Mr Warren, Dr Paykel, Dr Miller, Ms Lyon, Ms Ryan, Ms Simek, Ms Truong, and Dr Whitford received compensation in the form of clinician salary support for delivery of the CAP Service interventions.
SL. Incidence of community-acquired lower respiratory tract infections and pneumonia among older adults in the United Kingdom: a population-based study. PLoS One
. 2013;8(9):e75131. doi:10.1371/journal.pone.0075131PubMedGoogle ScholarCrossref
DT. The impact of multimorbidity on short-term events in patients with community-acquired pneumonia: prospective cohort study. Clin Microbiol Infect
. 2015;21(3):264.e7-264.e13. doi:10.1016/j.cmi.2014.11.002PubMedGoogle ScholarCrossref
et al. New perspectives on community-acquired pneumonia in 388 406 patients: results from a nationwide mandatory performance measurement programme in healthcare quality. Thorax
. 2009;64(12):1062-1069. doi:10.1136/thx.2008.109785PubMedGoogle ScholarCrossref
et al. Early switch from intravenous to oral antibiotics and early hospital discharge: a prospective observational study of 200 consecutive patients with community-acquired pneumonia. Arch Intern Med
. 1999;159(20):2449-2454. doi:10.1001/archinte.159.20.2449PubMedGoogle ScholarCrossref
et al. 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. Arch Intern Med
. 2012;172(12):922-928. doi:10.1001/archinternmed.2012.1690PubMedGoogle ScholarCrossref
et al. Effect of corticosteroids on treatment failure among hospitalized patients with severe community-acquired pneumonia and high inflammatory response: a randomized clinical trial. JAMA
. 2015;313(7):677-686. doi:10.1001/jama.2015.88PubMedGoogle ScholarCrossref
et al. Corticosteroid therapy for patients hospitalized with community-acquired pneumonia: a systematic review and meta-analysis. Ann Intern Med
. 2015;163(7):519-528. doi:10.7326/M15-0715PubMedGoogle ScholarCrossref
et al; Ovidius Study Group; Capisce Study Group; STEP Study Group. Corticosteroids in patients hospitalized with community-acquired pneumonia: systematic review and individual patient data metaanalysis. Clin Infect Dis
. 2018;66(3):346-354. doi:10.1093/cid/cix801PubMedGoogle ScholarCrossref
et al. The IMPROVE-GAP Trial aiming to improve evidence-based management of community-acquired pneumonia: study protocol for a stepped-wedge randomised controlled trial. Trials
. 2018;19(1):88. doi:10.1186/s13063-017-2407-4PubMedGoogle ScholarCrossref
C. Effectiveness-implementation hybrid designs: combining elements of clinical effectiveness and implementation research to enhance public health impact. Med Care
. 2012;50(3):217-226. doi:10.1097/MLR.0b013e3182408812PubMedGoogle ScholarCrossref
et al; Australian Community-Acquired Pneumonia Study Collaboration. SMART-COP: a tool for predicting the need for intensive respiratory or vasopressor support in community-acquired pneumonia. Clin Infect Dis
. 2008;47(3):375-384. doi:10.1086/589754PubMedGoogle ScholarCrossref
L. Safety and feasibility of an exercise prescription approach to rehabilitation across the continuum of care for survivors of critical illness. Phys Ther
. 2012;92(12):1524-1535. doi:10.2522/ptj.20110406PubMedGoogle ScholarCrossref
et al. Intention-to-treat vs on-treatment analyses of clinical trial data: experience from a study of pyrimethamine in the primary prophylaxis of toxoplasmosis in HIV-infected patients: ANRS 005/ACTG 154 Trial Group. Control Clin Trials
. 1998;19(3):233-248. doi:10.1016/S0197-2456(97)00145-1PubMedGoogle ScholarCrossref
et al. Study protocol for two randomized controlled trials examining the effectiveness and safety of current weekend allied health services and a new stakeholder-driven model for acute medical/surgical patients versus no weekend allied health services. Trials
. 2015;16:133. doi:10.1186/s13063-015-0619-zPubMedGoogle ScholarCrossref
et al. Fall rates in hospital rehabilitation units after individualised patient and staff education programmes: a pragmatic, stepped-wedge, cluster-randomised controlled trial. Lancet
. 2015;385(9987):2592-2599. doi:10.1016/S0140-6736(14)61945-0PubMedGoogle ScholarCrossref
et al. Adjunctive systemic corticosteroids for hospitalized community-acquired pneumonia: systematic review and meta-analysis 2015 update. Sci Rep
. 2015;5:14061. doi:10.1038/srep14061PubMedGoogle ScholarCrossref