Comments
Abstract
Rationale: Clinical decision making relative to community-acquired pneumonia (CAP) diagnosis is difficult. Chest radiograph is key in establishing parenchymal lung involvement. However, radiologic performance may lead to misdiagnosis, rendering questionable the use of chest computed tomography (CT) scan in patients with clinically suspected CAP.
Objectives: To assess whether early multidetector chest CT scan affects diagnosis and management of patients visiting the emergency department with suspected CAP.
Methods: A total of 319 prospectively enrolled patients with clinically suspected CAP underwent multidetector chest CT scan within 4 hours. CAP diagnosis probability (definite, probable, possible, or excluded) and therapeutic plans (antibiotic initiation/discontinuation, hospitalization/discharge) were established by emergency physicians before and after CT scan results. The adjudication committee established the final CAP classification on Day 28.
Measurements and Main Results: Chest radiograph revealed a parenchymal infiltrate in 188 patients. CAP was initially classified as definite in 143 patients (44.8%), probable or possible in 172 (53.8%), and excluded in 4 (1.2%). CT scan revealed a parenchymal infiltrate in 40 (33%) of the patients without infiltrate on chest radiograph and excluded CAP in 56 (29.8%) of the 188 with parenchymal infiltrate on radiograph. CT scan modified classification in 187 (58.6%; 95% confidence interval, 53.2–64.0), leading to 50.8% definite CAP and 28.8% excluded CAP, and 80% of modifications were in accordance with adjudication committee classification. Because of CT scan, antibiotics were initiated in 51 (16%) and discontinued in 29 (9%), and hospitalization was decided in 22 and discharge in 23.
Conclusions: In CAP-suspected patients visiting the emergency unit, early CT scan findings complementary to chest radiograph markedly affect both diagnosis and clinical management.
Clinical trial registered with www.clinicaltrials.gov (NCT 01574066).
Community-acquired pneumonia diagnosis is a daily challenge whose definition relies on clinical signs and radiograph abnormalities. Chest radiograph lacks sensitivity and specificity.
Chest computed tomography scan improves diagnosis and alters management in emergency patients with suspected community-acquired pneumonia.
Community-acquired pneumonia (CAP) is frequently diagnosed in emergency patients (1, 2). CAP mostly occurs in elderly and frail patients (3, 4), and often leads to life-threatening conditions with an overall 30-day mortality rate of 10% in adults (3). Because delayed treatment impairs prognosis, early diagnosis is necessary to administer antimicrobials in a timely manner (5). The gold standard for diagnosis of CAP should be detection of the microorganisms in the lung tissue (6), which is seldom feasible in everyday practice and requires 48 hours for results.
Clinical CAP diagnosis is often uncertain. Misdiagnosis is frequent and leads to delayed antimicrobial therapy (5) or overuse of antibiotics (7). An operational definition has been established to help physicians diagnose CAP (8). Lacking alternative explanations, CAP should be suspected in patients with systemic signs of infection, symptoms of acute lower respiratory tract infection, and new focal chest symptoms on examination (8). However, combining clinical signs and symptoms has limited value (9, 10).
Because of the difficult clinical decision making in CAP diagnosis, the presence of parenchymal lung disease determination, a requirement for pneumonia diagnosis, is based on evidence of parenchymal infiltrate on chest radiograph (8). However, significance of radiograph abnormalities remains debatable because of a considerable risk of missing or overdiagnosing CAP (11, 12). Concordance of interpretation on the presence of parenchymal infiltrate is poor, whatever practitioners’ experience and qualifications (13–16). Furthermore, appearance of infiltrate can be delayed and performance of chest radiograph distorted by coexisting comorbidities (17–19). Therefore, chest radiograph seems an imperfect gold standard for CAP in the context of emergency diagnosis process; nonetheless, it is currently used.
Some authors advocate the use of computed tomography (CT) scan when standard imaging is inconclusive (8, 9, 20). Additional data support the use of CT scan (21) to improve sensitivity of CAP diagnosis (21). Chest CT scan could thus help to better determine diagnosis.
Almost all major decisions regarding CAP management, including diagnostic and treatment issues, rely on the initial assessment (20). Because suspected CAP patients often are seen in the emergency department, developing strategies that improve early management is essential. Here we explore the impact of systematic early chest CT scan on diagnosis in patients visiting the emergency department with clinically suspected CAP, and on their management according to standard of care.
We conducted a multicenter, prospective, interventional study, entitled Early CT-Scan for Community-Acquired Pneumonia at the Emergency Department (ESCAPED), from November 2011 to January 2013, in four emergency units of tertiary teaching hospitals.
The study was supported by grants from the French Ministry of Health (PHRC AOM 10118), sponsored by Assistance Publique–Hôpitaux de Paris, and monitored by the Clinical Research Unit Paris Centre. The French health authorities (ANSM) and the institutional review board for the protection of human subjects (Paris No. 2011-oct-12749) approved the study protocol and patient informed consent procedures. All enrolled patients provided written informed consent before inclusion.
The primary objective was to measure the impact of multidetector chest CT scan on the probability of CAP as estimated by the attending emergency physician. To assess the study’s primary endpoint, we determined how often the clinical judgment of the physician was modified by the results of multidetector chest CT scan.
The secondary objectives were to assess how multidetector chest CT scan influenced the management of the patient (i.e., prescription of antimicrobial therapy and decision on site-of-care [admission or discharge] by the attending emergency physician), to describe multidetector chest CT scan results as compared with chest radiograph, to estimate whether multidetector chest CT scan led emergency physicians to properly classify patients as compared with an adjudication committee judgement, and to determine the factors associated with final adjudicated diagnosis of CAP based on chest CT independently of CAP category.
Consecutive adults (>18 yr) were enrolled if the attending emergency physician clinically suspected CAP. Clinical suspicion of CAP was based on investigator’s judgment in patients that fulfilled the following criteria: new onset of systemic infection (at least one among sweat, chills, aches and pain, temperature ≥38°C or <36°C) and symptoms of an acute lower respiratory tract infection (at least one among cough, sputum production, dyspnea, chest pain, altered breathing sounds at auscultation) (8). Pregnant women, patients in palliative care or with anticipated barriers to completing follow-up data collection, patients classified three or higher according to the CRB65 score (22), and those requiring intensive care for any purpose because of specific management of critically ill CAP patients were not eligible. Because of organizational constraints, patients could only be enrolled from Monday to Friday (8:00 a.m.–6:00 p.m.).
Patient management was based on current recommended practice guidelines. Recorded baseline data consisted of demographic data, coexisting illnesses, clinical findings, and laboratory tests.
Chest radiograph was performed using a standardized protocol. Conversely to most studies on CAP, inclusion criteria were based on clinical features solely; therefore, results of chest radiograph did not preclude inclusion. Characteristics of chest radiographs were recorded by the local radiologist on a dedicated form that specified position (standing, sitting, prone); views (front, profile); technical quality (good, fair, poor); and description of parenchymal, pleural, and mediastinum abnormalities. The radiologist established the radiologic CAP probability solely based on chest radiographs (high, intermediate, low, ruled out).
Multidetector chest CT scan was performed, as soon as possible after chest radiographs and after pre–CT scan evaluation questionnaire, ideally within the 4 hours following inclusion. A low-dose protocol was recommended. Contrast material was injected at the local radiologist’s discretion. CT scan was interpreted by the local radiologist who, in addition to usual description, indicated on a dedicated form the level of radiologic CAP probability according to the CT scan criteria (see the online supplement). The local radiologist was aware of CAP suspicion and patient history but of no other data.
To identify the patients for whom chest CT scan may be most beneficial in diagnosing or excluding pulmonary infiltrate, we compared the characteristics of patients using cross-tabulation according to the presence or absence of infiltrate on chest radiograph and results of chest CT scan.
Immediately before CT scan, the emergency physician filled in a standardized report form including patient history, laboratory data, and both his own interpretation and standardized interpretation of the chest radiograph by the local radiologist. The physician, aware of the interpretation of the chest radiograph by the local radiologist, established pre–CT scan probability of CAP diagnosis according to a four-level Likert scale (definite, probable, possible, excluded) and outlined an antimicrobial therapy plan and the site-of-care for patient management. This diagnosis classification does not correspond to a validated CAP classification but to the practitioner’s global confidence in CAP diagnosis.
Immediately after viewing the CT scan results, the same physician completed the standardized case report form and rated the post–CT scan probability of diagnosis of CAP according to the Likert scale, and outlined an antimicrobial therapy plan and site-of-care (admission or discharge). Based on post–CT scan evaluation, patients were discharged or admitted to the hospital on an appropriate unit and treated according to unit procedures.
The adjudication committee involved three independent experts in infectious diseases, pneumology, and radiology within a panel of nine experts, masked to emergency investigators’ rating. For each patient, the adjudication committee established two CAP probabilities. First, based on data collected in the baseline standardized case report forms, images of radiographs and multidetector CT scan recorded on a dedicated DVD, the adjudication committee retrospectively assigned the probability of CAP diagnosis using the four-level Likert scale (hereafter referred to as “After CT scan adjudication committee CAP probability”). Second, the adjudication committee assigned a final probability of diagnosis of CAP, using all available follow-up data including patients’ discharge summary and a telephone follow-up by assistant investigators with the patient, relatives, or general practitioners at Day 28 (hereafter referred to as “Day-28 adjudication committee CAP probability”). In patients lost to follow-up, post–CT scan CAP adjudication committee classification was carried forward. This Day-28 adjudication was used as the gold standard in the study.
Baseline and follow-up characteristics were described by means (SD) or median (interquartile range) for continuous variables normally distributed or with skewed distribution, respectively, and by percentages for categorical variables.
We performed chi-square or Fisher exact tests as appropriate for qualitative variables, and the Wilcoxon/Mann-Whitney test for continuous variables with skewed distributions to compare baseline patient characteristics and study outcomes between groups.
We considered that each modification of at least one category in the four-level Likert scale was a change in diagnosis, whatever the direction of the change (increase or decrease in the CAP probability level). To estimate whether chest CT scan helped emergency physicians to properly reclassify patients according to the adjudication committee’s final probability of diagnosis for CAP (gold standard), we calculated the net reclassification index (NRI) (23), thus dichotomizing the CAP level of certainty: patients with high probability (definite/probable) and low probability (possible/excluded).
Factors associated with final adjudicated diagnosis of CAP based on chest CT were analyzed using multivariable logistic regression. Variables were selected to enter the model if associated with outcome with a P value less than 0.10 in bivariate analysis. A stepwise backward procedure, based on the Akaike information criteria, was used to select the final adjusted model.
All tests were two-sided; P values less than 0.05 were considered to denote statistical significance. All statistical analyses were performed using SAS software V9.3 (SAS Institute, Cary, NC).
In a previous study (21), prevalence of CAP changed from 38.3% before CT scan to 55.3% after CT scan; bilateral infiltrates changed from 12.8% before CT scan to 34% after CT scan. Therefore we hypothesized that multidetector CT scan would modify diagnosis probability level of certainty in 20% of patients. We calculated that 300 patients would allow the estimation of diagnosis change prevalence, with a 95% confidence interval (CI), at 15–25%.
For the study period, 319 patients were available for analysis out of 333 included in the ESCAPED study (Figure 1). Characteristics of the participants appear in Table 1. Sex ratio was approximately one. Over half of the patients (56%) were 65 years of age or older. Significant underlying disorders were recorded in 195 (61%), including 89 (28%) pulmonary disorders. Cough (n = 240; 76%) and dyspnea (n = 229; 72%) were frequent. Unilateral crackles were detected in 105 (33%). Parenchymal infiltrate (unilateral and bilateral) were described on chest radiograph in 188 (61%). In seven patients, chest radiograph was performed within 24 hours preceding emergency department visit. In the 312 remaining patients, local radiologists considered that chest radiograph probability for CAP diagnosis was high, intermediate, and low in 80 (25.6%), 88 (28.2%), and 118 (37.8%) patients, respectively, and ruled out in 26 (8.3%) patients (see Table E1 in the online supplement). Based on the pre–CT scan evaluation, the emergency physician classified CAP diagnosis as definite in 143 patients (44.8%), probable in 118 (37.0%), possible in 54 (17.0%), and excluded in four (1.2%).
Figure 1. Flow chart. CAP = community-acquired pneumonia; CT = computed tomography.
[More] [Minimize]| Characteristics | No. (%) or Mean ± SD (n = 319) |
|---|---|
| General characteristics | |
| Age | |
| Mean, yr | 64.7 ± 20.0 |
| ≥65 yr | 177 (55.5) |
| Sex | |
| Female | 164 (51.4) |
| Male | 155 (48.6) |
| Nursing home resident | 12 (3.8) |
| Background and vaccinations | |
| Comorbidities | |
| At least one comorbidity | 195 (61.1) |
| Chronic respiratory disease | 89 (28.0) |
| COPD | 64 (20.1) |
| Asthma | 46 (14.4) |
| Congestive heart failure | 39 (12.3) |
| Diabetes | 51 (16.0) |
| Kidney disease | 36 (11.3) |
| Neoplasia | 32 (10.0) |
| Liver disease | 15 (4.7) |
| History of stroke | 12 (3.8) |
| Vaccination status | |
| Influenzae vaccination during the past year | 118 (40.0) |
| Pneumococcal vaccination | 45 (16.5) |
| CAP characteristics at inclusion | |
| Previous antibiotic treatment | 111 (34.8) |
| Symptom duration before visiting emergency unit, d | |
| In all patients (n = 319) | 7.4 ± 10.5 |
| In antibiotic treatment-naive patients (n = 208) | 5.5 ± 9.5 |
| In patients with prior antibiotic treatment (n = 111) | 10.8 ± 11.2 |
| Signs and symptoms in the emergency unit | |
| Cough | 240 (75.7) |
| Chest pain | 103 (32.4) |
| Sputum production | 147 (46.2) |
| Dyspnea | 229 (71.8) |
| Respiratory rate > 30/min | 42 (13.2) |
| Crackles | 105 (33.2) |
| Chills | 96 (30.2) |
| Headaches | 51 (16.0) |
| Myalgia | 59 (18.6) |
| Fever | 112 (35.3) |
| Confusion | 12 (3.8) |
| Heart rate > 125/min | 24 (7.5) |
| Systolic blood pressure <90 mm Hg | 4 (1.3) |
| Diastolic blood pressure <60 mm Hg | 26 (8.2) |
| PSI risk class | |
| I | 49 (15.4) |
| II | 83 (26.0) |
| III | 69 (21.6) |
| IV | 90 (28.2) |
| V | 28 (8.3) |
| CRB65 score* | |
| 1 | 149 (46.7) |
| 2 | 47 (14.7) |
| 3 | 5 (1.6) |
| 4 | 0 (0.0) |
| Biologic data | |
| White blood cell, 103/mm3 | 11.5 ± 5.6 |
| Procalcitonin, μg/L | 1.8 ± 5.3 |
| CRP, mg/L | 110.8 ± 107 0 |
| Urea > 11 mmol/L | 41 (12.9) |
| pH < 7.35 | 3 (0.9) |
| PaO2 < 60 mm Hg or SaO2 < 90% | 49 (17.0) |
| Radiologic data | |
| Parenchymal infiltrate | 188 (61.0) |
| Including unilateral finding | 128 (71.9) |
| Including bilateral finding | 50 (28.1) |
| Pleural effusion | 84 (26.4) |
| CAP management | |
| 28-d mortality | 13 (4.1) |
The main findings of the chest CT scan are summarized in Table 2 and Table E2. Radiologists considered that the probability for CAP diagnosis based on CT scan was high, intermediate, and low in 138 (43.2%), 38 (11.9%), and 37 (11.6%) patients, respectively, and ruled out in 105 (32.9%).
| Physician CAP Probability Level after Chest CT Scan | Total | Changes in Classifications | ||||||
|---|---|---|---|---|---|---|---|---|
| Definite | Probable | Possible | Excluded | Number | Modification Rates (95% CI) | |||
| Physician CAP probability level before chest CT scan | ||||||||
| Definite | 107 | 15 | 10 | 11 | 143 (44.8%) | 36 | 25.2% (18.1–32.3) | |
| Probable | 41 | 16 | 13 | 48 | 118 (36.9%) | 102 | 86.4% (80.3–92.6) | |
| Possible | 12 | 4 | 7 | 31 | 54 (16.9%) | 47 | 87.0% (78.1–96) | |
| Excluded | 2 | 0 | 0 | 2 | 4 (1.25%) | 2 | 50.0% (1.0–99.0) | |
| Total | 162 (50.8%) | 35 (10.9%) | 30 (9.4%) | 92 (28.8%) | 319 | 187 | 58.6% (53.2–64.0) | |
| Adjudication Committee CAP Probability after CT Scan | ||||||||
| 150 (47.0%) | 28 (8.7%) | 36 (11.3%) | 105 (32.9%) | |||||
| Day-28 Adjudication Committee CAP Probability | ||||||||
| 150 (47.0%) | 13 (4.1%) | 34 (10.7%) | 122 (38.2%) | |||||
In the 120 patients out of the 308 patients with both completed interpretation by the local radiologist of chest radiograph and CT scan, and without any parenchymal infiltrate on chest radiograph, CT scan revealed parenchymal infiltrates compatible with CAP in 40 patients (33%; 13% of the 308 patients). As compared with the 80 patients without infiltrates on chest CT scan, these 40 patients tended to be older than 65 years (62.5 vs. 45.0%; P = 0.0707), were more likely to present crackles (48.7 vs. 26.6%; P = 0.0169), to have higher C-reactive protein (CRP) levels (138.1 vs. 59.9 mg/L; P = 0.0037), and higher white blood cell counts (12.3 vs. 10.2 103/mm3; P = 0.0387) (see Table E3).
Among the 188 out of the 308 patients with a parenchymal infiltrate on chest radiograph, CT scan excluded pneumonia in 56 patients (29.8%; 18% of the 308 patients). As compared with the 132 patients with infiltrates on chest CT scan, these 56 patients were older (71.1 vs. 63.2 yr; P = 0.0131), had lower white blood cell counts (10.2 vs. 12.6 103/mm3; P = 0.0283), lower CRP levels (163.3 vs. 78.0 mg/L; P = 0.0074), and were more likely to have urea levels above 11 mM/L (25.0 vs. 11.4%; P = 0.0179) (see Table E4).
In the 85 patients with unifocal parenchymal infiltrate on chest radiograph, CT scan revealed multifocal infiltrates in 44 (51.8%). Table 3 presents the cross-tabulation of chest radiograph and CT scan CAP results.
| Parenchymal Infiltrate on Chest Radiograph | Chest CT Scan Probability of CAP | ||
|---|---|---|---|
| High or Intermediate* | Low or Ruled Out* | Total | |
| Yes | 132 | 56 | 188 (61.1%) |
| No | 40 | 80 | 120 (38.9%) |
| Total | 172 (55.8%) | 136 (44.2%) | 308 (100%)† |
Based on the CT scan evaluation, the emergency physician modified the probability for CAP diagnosis in 187 participants (58.6% [95% CI, 53.2–64.0%]). Classification was upgraded in 59 (18.4%) patients (including two excluded cases before CT scan that were reclassified as definite). Among the 162 post–CT scan definite CAP, 55 (34%) were changed to definite CAP because of CT scan. Classification was downgraded in 128 (40.4%) patients (including 11 out of 36 definite cases before CT scan that were reclassified as excluded). CAP was excluded in 28.8% of participants after chest CT scan (Table 2, Figure 1).
Intermediate (probable-possible) diagnostic categories were more subject to modification (76.7% [95% CI, 70.4–83.1%]) than those with a high degree of certainty (definite-excluded) (17% [95% CI, 10.9–23.1]) (for details, see Table E4).
Ten patients were lost to follow-up and their post–CT scan classification was carried forward for the final classification. The “after CT scan adjudication committee CAP probability” and the “Day-28 adjudication committee CAP probability” are presented in Table 2 and Figure 1.
NRI is presented in Table E5. For 100 patients (31.3%), the emergency physician changed the CAP probability level. Modifications of CAP probability level were adequate with the adjudication committee’s final classification in 80 out of 100 (80.0% of the modifications; 25.1% of the total population; NRI = 0.39). Most modifications (70%) consisted in appropriate downgrading of diagnosis probability from definite/probable CAP to low probability (possible/excluded), whereas 10% consisted in appropriate diagnosis upgrading (from low probability to high probability). In 20 out of 100 patients, reclassification was inadequate (Table 2).
In bivariate analysis, few parameters differed between the 80 participants with adequate reclassification and the remaining 239 (219 without changes, 20 with inadequate reclassification) (see Table E6). According to the multivariate analysis, CAP probability was adequately changed by multidetector chest CT scan results if pre–CT scan diagnosis was “probable” (60% of probable cases being downgraded) and the absence of parenchyma infiltrate on chest radiograph (Table 4).
| OR (95% CI)* | P Value† | |
|---|---|---|
| Sex | ||
| Female | 1.00 | |
| Male | 1.60 (0.87–2.92) | 0.13 |
| Previous antimicrobial therapy | ||
| No | 1.00 | |
| Yes | 0.57 (0.30–1.08) | 0.09 |
| Physicians’ agreement for diagnosis | <0.001 | |
| Definite | 1.00 | |
| Probable | 6.43 (3.21–12.88) | <0.001 |
| Possible | 0.79 (0.28–2.22) | 0.66 |
| Excluded | 3.55 (0.32–38.94) | 0.30 |
| Parenchyma infiltrate on chest radiograph | ||
| Present | 1.00 | |
| Absent | 2.97 (1.60–5.50) | <0.001 |
| Unilateral crackles | ||
| No | 1.00 | |
| Yes | 1.76 (0.90–3.43) | 0.10 |
Before CT scan, antimicrobial agents were initiated in 207 (64.8%) patients. After CT scan, administration of antimicrobial agents was stopped in 29 (14.0%) of these 207 patients. Conversely 51 (45.5%) of 112 patients without initial antimicrobial therapy were given antibiotics after CT scan results. CT scan led to initiation of anticoagulation for pulmonary embolism in three patients, and diuretics for cardiac failure in 11 patients.
CT scan also induced a modification in decisions for site-of-care. A total of 45 (14.1%) changed categories: 22 outpatients were finally admitted, and 23 admissions changed for discharge. Modifications in antimicrobial treatments, including changes in pharmacologic classes, and of site-of-care were observed in 194 (60.8%).
In this prospective study, we assessed the effect of early chest multidetector CT scan on clinical decision in patients with clinically suspected CAP visiting emergency units. CT scan modified CAP probability level by emergency physicians in over half of the patients, 80% of these modifications were in accordance with adjudication committee final CAP classification, and led to modifications of medical decisions in two-thirds. These modifications involved patients with clinically suspected CAP and either parenchymal infiltrate on chest radiograph (for one-third of whom CT scan excluded CAP), or those without infiltrate for whom the discovery of the infiltrate on CT scan, also in one-third, made possible the establishment of CAP diagnosis and the initiation of the adequate therapy. Characteristics of patients that benefit from chest CT scan to confirm or to rule out CAP diagnosis differed from those that did not.
Diagnosing CAP currently relies on the combination of systemic and lower respiratory tract symptoms of infection associated with new infiltrates on chest radiograph (8). For this study, we purposely decided to include patients based on initial clinical features, without results of the radiographs, to measure the positive impact of CT scan even in patients with normal chest radiograph. Thereby we challenged the performance of chest radiograph for CAP diagnosis. Despite this specific point, characteristics of patients included in this study are comparable with those reported in the literature. The exclusion of highest CRB65 categories in the ESCAPED study, limiting inclusion of patients older than 65 (22), may explain that mean age (65 yr) falls within the lower values of those reported elsewhere. Previous history of respiratory disorders, cancer, and congestive heart failure was frequent (2, 3, 10, 17, 19). Therefore, we believe that our results can be extrapolated to most emergency patients with suspected CAP who could benefit from CT scan.
To measure the effect of CT scan on CAP probability, we asked the emergency physicians to rate the probability of CAP based on their own judgement (taking into account history, clinical and biologic data, the standardized interpretations of chest radiograph, without and then with multidetector chest CT scan by the local radiologist), using a Likert scale. This classification did not rely on a validated scale, as in other infectious diseases, such as endocarditis (24). However, we believe this classification based on doctor self-assessment to make sense. Such a hypothesis has been raised for pulmonary embolism pretest probability, especially for experienced physicians (25).
Overall, CT scan led to better practitioner confidence in CAP diagnosis. The upgrading of CAP diagnosis level of certainty in 18.4% reveals the low sensitivity of chest radiograph for diagnosing CAP. This is particularly important in the subgroup of patients with clinically suspected CAP but without parenchymal infiltrate on chest radiograph (stricto sensu, non-CAP patients) for one-third of whom the discovery of infiltrate on CT scan suggests CAP diagnosis. Furthermore, in patients with definite/probable chest radiograph, CT scan allowed better staging of the pulmonary involvement (discovery of multifocal or bilateral localizations), identifying localizations or complications undiagnosed by chest radiograph. Conversely, CT scan also underscored the low performance of chest radiograph in the 40% of patients for whom CT scan results led to downgrading of the CAP level of certainty, including patients with definite CAP who were excluded after CT scan. Overall, one-third of the CAP cases was excluded after CT scan. In all these patients, early CT scan corrected patients’ diagnosis and avoided diagnostic and treatment red herrings. Of note, post–CT scan CAP classification performed at Day 1 was confirmed at Day-28 evaluation in 80%, revealing CT scan reliability in early evaluation.
Furthermore, CT scan results not only induced diagnosis probability changes but also led to immediate adjustment of patients’ care. Alternative diagnoses to CAP were mainly exacerbation of chronic obstructive pulmonary disease (16%) and acute heart failure (16%) as reported in the current literature (26). We acknowledge that the appropriateness of treatment \changes based on CT scan results may be debatable. Recommendations for daily practice are developed from studies using radiographs for standard diagnosis (8). Therefore, whereas CT scan improved diagnosis of CAP, it is unclear whether it also results in better outcome. Indeed, ESCAPED was not designed to assess the impact on outcome of performing a multidetector chest CT scan; however, it is probable that the initiation of antibiotic in patients for whom CT scan established CAP diagnosis had a positive impact on their outcome.
Despite systematic CT scan, the experts of the adjudication committee were unable to provide firm diagnosis (definite or excluded) at Day 28 in some patients. This underscores how difficult the diagnosis of CAP is even using better quality imaging. Whereas the best diagnosis for infection should be the proof a pathogen in a usually sterile tissue, this is seldom possible in daily practice for CAP patients in whom microbiologic results are frequently negative (27). Whereas new biologic tools may also help (28), current recommendations do not support routine use of biomarkers to assist diagnosis of CAP, and new microbiologic techniques have seldom been evaluated. We also agree that patients without infiltrate observed on radiograph and with unsure diagnosis greatly benefited from CT scan. However, we also observed significant changes in several patients with radiograph abnormalities.
Our results suggest that many patients with suspected CAP would benefit from CT scan, although this strategy may encounter some barriers. Use of CT scan necessitates exposure to radiation and thus might be harmful. For the study’s purpose, the protocol recommended using a low-dose CT scan, with an estimated radiation dose corresponding to 250 mGy.cm (dose length product; i.e., twice the natural radiation received each year in Western countries). However, improvement in reconstruction methods reduces CT scan radiation to levels of a standard chest radiograph and allows adequate quality for parenchymal study (29). Another barrier is the cost effectiveness of the procedure, which was not addressed in the present study. CAP management-related costs vary with site-of-care (30). We observed that CT scan modified (14%) the site-of-care in few patients. On the one hand, CT scan allowed discharge of these initially admitted patients, limiting cost of treatment. On the other hand, it permitted admission of patients for whom delayed treatment would have had a negative medical impact. However, we cannot ascertain in this study whether CT scan is cost-effective. Finally, whereas availability may vary among hospitals, most emergency departments now have easy access to CT scan (31).
CAP presents an extensive clinical and radiologic spectrum (20). Beyond the difficulties of interpretation and interobserver discrepancies, chest radiograph results seemed, in a large number of cases, to inadequately guide emergency physicians, leading them toward making inappropriate decisions on both diagnosis and antimicrobial therapy for CAP-suspected patients, which raises concerns for a disease with a high 30-day mortality rate. The ESCAPED study suggests that a CT scan performed within the first hours facilitates early and accurate positive or negative diagnosis of CAP. A key question is the population that would most benefit from chest CT scan. We observed that CAP-suspected patients with negative chest radiograph but for whom chest CT scan reveals a parenchymal infiltrate (i.e., false-negative of chest radiograph) were more likely to have crackles or high inflammatory markers (CRP or white blood cells count). This suggested that patients will benefit from CT scan when chest radiograph is normal despite clinical signs and biologic markers evocating CAP. Conversely, those patients with positive chest radiograph and negative chest CT scan (i.e., false-positive of chest radiograph) had lower inflammatory markers (CRP or white blood cell count). In both conditions, older patients benefited from chest CT scan.
Here, early use of CT scan clearly outclassed chest radiograph and affected diagnosis, treatment, and decision for site-of-care in emergency patients with suspected CAP. Therefore, we believe that a strategy favoring CT scan as the preferred first imaging technique in targeted patients would improve diagnosis and may even, in the near future, reduce global radiation exposure by limiting unnecessary radiation cause by multiple procedures. Whether these modifications would improve outcome should be addressed in a randomized controlled trial.
The authors thank URC-CIC Paris Centre (C. Auger) for implementation, monitoring, and data management of the study.
Scientific Committee: Steering Committee: Y.-E. Claessens, M.D., Ph.D. (principal investigator), X. Duval, M.D., Ph.D. (coprincipal investigator), E. Bouvard, M.D., M.-F. Carette, M.D., Ph.D., M.-P. Debray, M.D., Ph.D., C. Mayaud, M.D., Ph.D., C. Leport, M.D., Ph.D., N. Houhou, M.D., Ph.D., and S. Tubiana, Ph.D. Validation Committee: M. Benjoar, M.D., F. X. Blanc, M.D., Ph.D., A.-L. Brun, M.D., L. Epelboin, M.D., C. Ficko, M.D., A. Khalil, M.D., Ph.D., H. Lefloch, M.D., J.-M. Naccache, M.D., Ph.D., and B. Rammaert, M.D., Ph.D.
Clinical Investigators: A. Abry, M.D., J. C. Allo, M.D., S. Andre, M.D., C. Andreotti, M.D., N. Baarir, M.D., M. Bendahou, M.D., L. Benlafia, M.D., J. Bernard, M.D., A. Berthoumieu, M.D., M. E. Billemont, M.D., J. Bokobza, M.D., A.-L. Brun, M.D., E. Burggraff, M.D., P. Canavaggio, M.D., M.-F. Carette, M.D., Ph.D., E. Casalino, M.D., Ph.D., S. Castro, M.D., C. Choquet, M.D., H. Clément, M.D., L. Colosi, M.D., A. Dabreteau, M.D., S. Damelincourt, M.D., S. Dautheville, M.D., M.-P. Debray, M.D., M. Delay, M.D., S. Delerme, M.D., L. Depierre, M.D., F. Djamouri, M.D., F. Dumas, M.D., M. R. S. Fadel, M.D., A. Feydey, M.D., Y. Freund, M.D., L. Garcia, M.D., H. Goulet, M.D., P. Hausfater, M.D., Ph.D., E. Ilic-Habensus, M.D., M. O. Josse, M.D., J. Kansao, M.D., Y. Kieffer, M.D., F. Lecomte, M.D., K. Lemkarane, M.D., P. Madonna, M.D., O. Meyniard, M.D., L. Mzabi, M.D., D. Pariente, M.D., J. Pernet, M.D., F. Perruche, M.D., J. M. Piquet, M.D., R. Ranerison, M.D., P. Ray, M.D., Ph.D., F. Renai, M.D., E. Rouff, M.D., D. Saget, M.D., K. Saïdi, M.D., G. Sauvin, M.D., E. Trabattoni, M.D., and N. Trimech, M.D.
Monitoring, Data Management, and Statistical Analysis: C. Auger, R.N., B. Pasquet, M.D., S. Tamazirt, R.N., J. M. Treluyer, M.D., F. Tubach, M.D., and J. Wang, R.N.
Sponsor: Assistance Publique-Hôpitaux de Paris, Délégation Interrégionale à la Recherche Clinique d’Ile De France (O. Chassany, M.D., and C. Misse, M.D.).
| 1. | Dorca J, Torres A. Lower respiratory tract infections in the community: towards a more rational approach. Eur Respir J 1996;9:1588–1589. |
| 2. | Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, Coley CM, Marrie TJ, Kapoor WN. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243–250. |
| 3. | Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001;29:1303–1310. |
| 4. | Griffin MR, Zhu Y, Moore MR, Whitney CG, Grijalva CG. U.S. hospitalizations for pneumonia after a decade of pneumococcal vaccination. N Engl J Med 2013;369:155–163. |
| 5. | Houck PM, Bratzler DW, Nsa W, Ma A, Bartlett JG. Timing of antibiotic administration and outcomes for Medicare patients hospitalized with community-acquired pneumonia. Arch Intern Med 2004;164:637–644. |
| 6. | Ruiz-González A, Falguera M, Nogués A, Rubio-Caballero M. Is Streptococcus pneumoniae the leading cause of pneumonia of unknown etiology? A microbiologic study of lung aspirates in consecutive patients with community-acquired pneumonia. Am J Med 1999;106:385–390. |
| 7. | Kanwar M, Brar N, Khatib R, Fakih MG. Misdiagnosis of community-acquired pneumonia and inappropriate utilization of antibiotics: side effects of the 4-h antibiotic administration rule. Chest 2007;131:1865–1869. |
| 8. | Lim WS, Levy ML, Macfarlane JT; British Thoracic Society Community Acquired Pneumonia Guidelines Committee. Community acquired pneumonia. Management in primary care. BMJ 2010;341:c4469. |
| 9. | Metlay JP, Fine MJ. Testing strategies in the initial management of patients with community-acquired pneumonia. Ann Intern Med 2003;138:109–118. |
| 10. | Metlay JP, Schulz R, Li YH, Singer DE, Marrie TJ, Coley CM, Hough LJ, Obrosky DS, Kapoor WN, Fine MJ. Influence of age on symptoms at presentation in patients with community-acquired pneumonia. Arch Intern Med 1997;157:1453–1459. |
| 11. | Vilar J, Domingo ML, Soto C, Cogollos J. Radiology of bacterial pneumonia. Eur J Radiol 2004;51:102–113. |
| 12. | Young M, Marrie TJ. Interobserver variability in the interpretation of chest roentgenograms of patients with possible pneumonia. Arch Intern Med 1994;154:2729–2732. |
| 13. | Albaum MN, Hill LC, Murphy M, Li YH, Fuhrman CR, Britton CA, Kapoor WN, Fine MJ; PORT Investigators. Interobserver reliability of the chest radiograph in community-acquired pneumonia. Chest 1996;110:343–350. |
| 14. | Melbye H, Dale K. Interobserver variability in the radiographic diagnosis of adult outpatient pneumonia. Acta Radiol 1992;33:79–81. |
| 15. | Hopstaken RM, Witbraad T, van Engelshoven JM, Dinant GJ. Inter-observer variation in the interpretation of chest radiographs for pneumonia in community-acquired lower respiratory tract infections. Clin Radiol 2004;59:743–752. |
| 16. | Novack V, Avnon LS, Smolyakov A, Barnea R, Jotkowitz A, Schlaeffer F. Disagreement in the interpretation of chest radiographs among specialists and clinical outcomes of patients hospitalized with suspected pneumonia. Eur J Intern Med 2006;17:43–47. |
| 17. | Corrales-Medina VF, Musher DM, Wells GA, Chirinos JA, Chen L, Fine MJ. Cardiac complications in patients with community-acquired pneumonia: incidence, timing, risk factors, and association with short-term mortality. Circulation 2012;125:773–781. |
| 18. | Basi SK, Marrie TJ, Huang JQ, Majumdar SR. Patients admitted to hospital with suspected pneumonia and normal chest radiographs: epidemiology, microbiology, and outcomes. Am J Med 2004;117:305–311. |
| 19. | Ilo J, Rodriquez R, Jubert P, Alavarez B, and the S.C.A.P. Severe community-acquired pneumonia in the elderly: epidemiology and prognosis. Clin Infect Dis 1996;23:723–728. |
| 20. | Waterer GW, Rello J, Wunderink RG. Management of community-acquired pneumonia in adults. Am J Respir Crit Care Med 2011;183:157–164. |
| 21. | Syrjälä H, Broas M, Suramo I, Ojala A, Lähde S. High-resolution computed tomography for the diagnosis of community-acquired pneumonia. Clin Infect Dis 1998;27:358–363. |
| 22. | Lim WS, van der Eerden MM, Laing R, Boersma WG, Karalus N, Town GI, Lewis SA, Macfarlane JT. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003;58:377–382. |
| 23. | Pencina MJ, D’Agostino RB Sr, Steyerberg EW. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011;30:11–21. |
| 24. | Li JS, Sexton DJ, Mick N, Nettles R, Fowler VG Jr, Ryan T, Bashore T, Corey GR. Proposed modifications to the Duke criteria for the diagnosis of infective endocarditis. Clin Infect Dis 2000;30:633–638. |
| 25. | Kelly J, Hunt BJ. The utility of pretest probability assessment in patients with clinically suspected venous thromboembolism. J Thromb Haemost 2003;1:1888–1896. |
| 26. | Wunderink RG, Waterer GW. Clinical practice: community-acquired pneumonia. N Engl J Med 2014;370:543–551. |
| 27. | Woodhead M. Community-acquired pneumonia in Europe: causative pathogens and resistance patterns. Eur Respir J Suppl 2002;36:20s–27s. |
| 28. | van Vugt SF, Broekhuizen BD, Lammens C, Zuithoff NP, de Jong PA, Coenen S, Ieven M, Butler CC, Goossens H, Little P, et al.; GRACE consortium. Use of serum C reactive protein and procalcitonin concentrations in addition to symptoms and signs to predict pneumonia in patients presenting to primary care with acute cough: diagnostic study. BMJ 2013;346:f2450. |
| 29. | Mathieu KB, Ali H, Fox PS, Godoy MC, Munden RF, De Groot PM, Pan T. For CT lung cancer screening using ASIR and MBIR: a phantom study. Radiation dose reduction. J Appl Clin Med Phys 2014;15:4515. |
| 30. | Sato R, Gomez Rey G, Nelson S, Pinsky B. Community-acquired pneumonia episode costs by age and risk in commercially insured US adults aged ≥50 years. Appl Health Econ Health Policy 2013;11:251–258. |
| 31. | Ginde AA, Foianini A, Renner DM, Valley M, Camargo CA Jr. Availability and quality of computed tomography and magnetic resonance imaging equipment in U.S. emergency departments. Acad Emerg Med 2008;15:780–783. |
Supported by Assistance Publique-Hôpitaux de Paris, Délégation Interrégionale à la Recherche Clinique d’Ile de France (O. Chassany, M.D., and C. Misse, M.D.).
Author Contributions: Conception and design, Y.-E.C., C.L., C.M., M.-F.C., and X.D. conceived the study. Y.-E.C. and X.D. designed the study. Critical review of the design was provided by C.L., C.M., and M.-F.C. Clinical data were obtained by C.C., P.H., P.R., and Y.-E.C. Radiologic data were obtained by M.-F.C., M.-P.D., and A.-L.B. Analysis and interpretation of data, Y.-E.C., F.T., and X.D. Drafting of the manuscript, Y.-E.C. and X.D. were responsible for the manuscript draft. Critical review was provided by J.-M.N., B.R., C.L., C.M., and M.-F.C. Each author read the manuscript and contributed critiques that were included in the manuscript. All the authors accept the final draft of the manuscript. F.T. and B.R. were responsible for the statistical analysis. Y.-E.C. and X.D. obtained institutional funding. All authors have participated sufficiently in the work to take public responsibility for the whole content of the manuscript.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201501-0017OC on July 13, 2015
Author disclosures are available with the text of this article at www.atsjournals.org.
