MAJOR ARTICLE

Patterns of Resolution of Chest Radiograph Abnormalities in Adults Hospitalized with Severe Community-Acquired Pneumonia
Anke H. W. Bruns,1 Jan Jelrik Oosterheert,1 Mathias Prokop,2 Jan-Willem J. Lammers,3 Eelko Hak,4 and Andy I. M. Hoepelman1,5
1

Division of Medicine, Department of Internal Medicine and Infectious Diseases, Departments of 2Radiology and 3Pulmonary Diseases, 4Julius Center for Health Sciences and Primary Care, and 5Eijkman-Winkler Institute for Microbiology, Infectious Diseases and Inammation, University Medical Center Utrecht, Utrecht, The Netherlands

Background. Timing of follow-up chest radiographs for patients with severe community-acquired pneumonia (CAP) is difcult, because little is known about the time to resolution of chest radiograph abnormalities and its correlation with clinical ndings. To provide recommendations for short-term, in-hospital chest radiograph followup, we studied the rate of resolution of chest radiograph abnormalities in relation to clinical cure, evaluated predictors for delayed resolution, and determined the inuence of deterioration of radiographic ndings during follow-up on prognosis. Methods. A total of 288 patients who were hospitalized because of severe CAP were followed up for 28 days in a prospective multicenter study. Clinical data and scores for clinical improvement at day 7 and clinical cure at day 28 were obtained. Chest radiographs were obtained at hospital admission and at days 7 and 28. Resolution and deterioration of chest radiograph ndings were determined. Results. At day 7, 57 (25%) of the patients had resolution of chest radiograph abnormalities, whereas 127 (56%) had clinical improvement (mean difference, 31%; 95% condence interval, 25%37%). At day 28, 103 (53%) of the patients had resolution of chest radiograph abnormalities, and 152 (78%) had clinical cure (mean difference, 25%; 95% condence interval, 19%31%). Delayed resolution of radiograph abnormalities was independently associated with multilobar disease (odds ratio, 2.87; P .01 ); dullness to percussion at physical examination (odds ratio, 6.94; P .01 ); high C-reactive protein level, dened as 1200 mg/L (odds ratio, 4.24; P .001); and high respiratory rate at admission, dened as 125 breaths/min (odds ratio, 2.42; P .03). There were no signicant differences in outcome at day 28 between patients with and patients without deterioration of chest radiograph ndings during the follow-up period (P 1 .09). Conclusions. Routine short-term follow-up chest radiographs (obtained !28 days after hospital admission) of hospitalized patients with severe CAP seem to provide no additional clinical value. Community-acquired pneumonia (CAP) is one of the most common infectious diseases worldwide. Despite advances in diagnosis and treatment, pulmonary infections remain a major cause of morbidity and mortality in adult patients [1, 2]. In Europe, mortality varies from 10 to 40 cases per 100,000 population per year [3]. Chest radiography is an important initial step in conrming or excluding a diagnosis of pneumonia in patients with suspected pulmonary infection. It is also commonly used during follow-up once pneumonia has been diagnosed, although little is known regarding the correlation of time to the resolution of chest radiograph abnormalities with clinical ndings. In our University Medical Center (University Medical Center Utrecht, Utrecht, The Netherlands), for example, chest radiographs were obtained, on average, every third day during CAP-related hospitalizations. Surprisingly, despite their frequent use, the clinical value of routine followup chest radiographs in patients with CAP has not been clearly established, and in routine clinical practice, chest radiographs are probably often used to conrm clinical

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Received 2 May 2007; accepted 20 June 2007; electronically published 12 September 2007. Reprints or correspondence: Prof. Dr. Andy I. M. Hoepelman, University Medical Center Utrecht, Dept. of Internal Medicine and Infectious Diseases, PO Box 85500, 3508 GA Utrecht, The Netherlands (I.M.Hoepelman@umcutrecht.nl). Clinical Infectious Diseases 2007; 45:98391 2007 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2007/4508-0007$15.00 DOI: 10.1086/521893

Resolution of Radiograph Abnormalities in Severe CAP CID 2007:45 (15 October) 983

cure of uncomplicated pneumonia [4, 5]. Major clinical guidelines do not provide unequivocal recommendations regarding the use of chest radiographs in the short-term follow-up of hospitalized patients [611]. The Infectious Diseases Society of America advises the use of in-hospital chest radiographs only to determine reasons for treatment failure [7], whereas the American Thoracic Society, in their 1993 guideline, recommended obtaining chest radiographs for all hospitalized patients prior to discharge from the hospital [8]. However, the most recent American Thoracic Society guideline no longer supports this recommendation [9]. The most recent Infectious Diseases Society of America and American Thoracic Society consensus guideline does not address this subject [11]. Furthermore, earlier clinical studies performed to evaluate the use of chest radiographs in CAP have used inaccurate denitions of clearance, lacked microbiological conrmation, or included a small or heterogeneous group of patients [1217]. As a consequence, recommendations provided by current clinical guidelines are only based on grade D recommendations. To further investigate the clinical value of routine follow-up chest radiographs within the rst month after diagnosis of CAP in immunocompetent hospitalized adult patients, we studied the time to resolution of pulmonary inltrates and other chest radiograph abnormalities caused by infection, compared with resolution of clinical parameters, and assessed factors associated with delayed resolution of radiographic abnormalities. METHODS Setting and study population. The data used in this study were derived from a multicenter, prospective randomized trial on the cost-effectiveness of an early switch from parenteral to oral therapy for severe CAP [18]. Severe CAP was dened as a pneumonia severity index (PSI) score of 190 or according to the American Thoracic Society denition [9, 19]. Adult patients (age, 18 years) who were admitted to the hospital because of CAP were eligible for inclusion [18]. Data collection. At hospital admission, demographic data and clinical signs and symptoms were recorded. Routine laboratory tests, arterial blood gas analysis, and a chest radiograph were obtained and the PSI score was calculated for each individual patient [19]. Antibiotic treatment was based on the Dutch guidelines for CAP management [10]. Patients were observed for a maximum of 28 days. During the in-hospital follow-up period, clinical data were recorded. Patients who were still hospitalized on day 7 had follow-up chest radiography performed. After discharge from the hospital, all patients were asked to return to the outpatient clinic for clinical evaluation, blood chemistry analysis, and a chest radiograph at day 28. Scores for clinical improvement on day 7 and for clinical cure on day 28 were calculated for each patient [20, 21]. Microbiological analysis. Details of the microbiological
984 CID 2007:45 (15 October) Bruns et al.

analyses are described elsewhere [18]. In short, sputum and blood samples were collected, cultured, and evaluated according to standard procedures. Microorganisms cultured from blood or sputum samples were recorded. In addition, NOW tests (Binax) were used to detect urinary antigens for Legionella pneumophila and Streptococcus pneumoniae. Acute- and convalescent-phase serum samples were collected and tested for Mycoplasma pneumoniae, L. pneumophila, and Chlamydia pneumoniae. Any noncontaminating microorganism detected by culture, serum test, or urinary antigen test was considered to be a cause of the episode of pneumonia. Interpretation of radiographs. The radiologists on call, who had been provided with information on the patients clinical condition, evaluated the chest radiographs. Standard radiographic criteria were used for evaluation [22, 23]. Chest radiographs were reviewed for the presence of pulmonary inltrates, pleural uid, atelectasis, pulmonary edema, and other ndings. The ndings were recorded in standard case-record forms by members of the study team. During follow-up, clearance of pulmonary inltrates (dened as the absence of any radiographic sign of consolidation) and resolution of chest radiograph abnormalities (dened as the absence of any abnormal chest radiograph nding potentially related to infection, such as inltrates, atelectasis, or pleural uid) were established. Furthermore, chest radiographs were evaluated for deterioration of chest radiograph ndings, which was dened as a new abnormal nding or an increase in pulmonary inltrates, pleural uid, or atelectasis during follow-up.

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Figure 1. Flow chart of patients and chest radiographs throughout the study. CAP, community-acquired pneumonia.

Table 1. Descriptive characteristics of study patients with community-acquired pneumonia.

Patients (n p 288) SD (range) SD (range) 69.7 112.8 13.9 (2095) 98 (34.0) 39.2 (33217) 248 (86.1) 154 (53.5) 34 (11.8) 65 (22.6) 25 (8.7) 27 (9.4) 20 (6.9) 248 88 177 33 SD (range) 38.5 26.6 (86.1) (30.6) (61.5) (11.5)

Variable Age, mean years Female sex

Pneumonia severity index, mean score Pneumonia severity index score 190 Comorbidity Chronic heart failure Neoplasm Cerebrovascular disease Renal disease Death Clinical feature Dyspnea Chest pain Productive cough Hemoptysis Temperature, mean C SD (range) Respiratory rate, mean breaths/min Physical examination nding Crackles Rales Friction rub Dullness to percussion Laboratory nding C-reactive protein level, mean mg/L WBC count, mean 1 109 cells/L Microbial nding Streptococcus pneumoniaea

1.23 (5.241.4) 8.7 (10.052.0) 151 (52.4) 120 (41.7) 11 (3.8) 45 (15.6)

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

SD (range) SD (range)

205 16.5

152.2 (0724) 9.2 (1.876.0) 62 (21.5) 19 29 27 28 (9.7) 12 (4.2) 7 9 50 148 (2.4) (3.1) (17.4) (51.4)

S. pneumonia proved by sputum culture S. pneumonia proved by blood culture S. pneumonia proved by antigen test Atypical pathogen Chlamydia pneumoniae Mycoplasma pneumoniae Legionella pneumophila Other pathogenb Unknown etiology Multiple pathogens
NOTE. Data are no. (%) of patients, unless otherwise indicated.
a

11 (3.8)

In some cases, the S. pneumonia infection was identied by multiple tests. Other pathogens included gram-negative enteric species (in 19 [6.3%] of the patients), Haemophilus inuenzea (9 [3.1%]), Staphylococcus aureus (9 [3.1%]), Pseudomonas aeruginosa (2 [0.7%]), Klebsiella pneumonia (4 [1.4%]), Moraxella catarrhalis (5 [1.7%]), and other (2 [0.7%]).
b

Outcome measures. Primary objectives of the study included the evaluation of the time to resolution and factors associated with delayed resolution of chest radiograph abnormalities. Time to resolution of chest radiograph abnormalities was estimated on the basis of the proportion of patients who had radiographic evidence of infection (i.e., inltrates, pleural uid, or atelectasis) at day 0, day 7, and day 28, respectively.

Subgroup analysis was performed for patients with proven pneumococcal pneumonia. Factors associated with delayed resolution of radiograph ndings were assessed in univariate and multivariate analyses, which included patient characteristics, symptoms, initial radiographic abnormalities, laboratory data, and microbial etiology. Furthermore, resolution of radiograph abnormalities at day 7 and day 28 was compared with clinical

Resolution of Radiograph Abnormalities in Severe CAP CID 2007:45 (15 October) 985

Table 2. Pattern of resolution of radiograph abnormalities among patients with severe community-acquired pneumonia.
Chest radiograph ndings Time since hospital admission, days 0 7 28
a

No. of patients 288 227 195

Cumulative dropout ratea 0/288 (0) 61/288 (21.2) 93/288 (32.3)

Normal 0/288 (0) 57/227 (25.1) 103/195 (52.8)

Inltrates 288/288 (100) 152/227 (67.0) 74/195 (37.9)

Multilobar inltrates 97/288 (33.7) 51/227 (22.5) 22/195 (11.3)

Atelectasis 18/288 (6.3) 12/227 (5.3) 13/195 (6.7)

Pleural uid 50/288 (17.4) 66/227 (29.1) 25/195 (12.8)

NOTE. Data are proportion (%) of patients. Cumulative dropout rate reects the no. of patients who died during the acute phase of illness or were lost to follow-up during the study.

improvement at day 7 (dened as a respiratory rate !25 breaths/ min; oxygen saturation 190%, as measured by pulse oximetry; partial pressure of oxygen in arterial blood 155 mmHg; hemodynamic stability; and no acute alteration in the mental state of the patient [20]) and with clinical cure at day 28 (dened as hospital discharge in good health, without signs and symptoms of pneumonia, and the absence of complications or clinical deterioration, including the need for mechanical ventilation, readministration of intravenous antibiotics after a switch to oral therapy, readmission for pulmonary infection after hospital discharge, or an increase in body temperature after an initial improvement during the follow-up period [20, 21]). Clinical failure was dened as no clinical improvement or cure. Finally, the inuence of deterioration of chest radiograph ndings on outcome (i.e., clinical failure, mortality, PSI score, and interventions) for patients with and patients without deterioration of chest radiograph ndings was studied. Statistical methods and analytical approach. Statistical analysis was performed using SPSS for Windows, version 12.0. (SPSS). For comparisons between subgroups of patients with continuous scaled data, the Students t test was used. Categorical variables were compared using the x2 or Fishers exact test. All analyses used a 2-tailed signicance level of P ! .05. Univariate assessment of the association of each parameter and resolution of chest radiograph abnormalities on day 7 and day 28 was performed by estimation of the ORs and corresponding 95% CIs for the selected patients. Multivariate logistic regression modeling with P ! .10 as criterion for entry was applied to select those variables that were associated with resolution of radiographic abnormalities. Forward selection was also per-

formed to verify whether any previously deleted potentially relevant characteristic had been eliminated incorrectly from the model. The reliability of the multivariate logistic regression model was determined by Hosmer-Lemeshow goodness-of-t statistics [24]. The area under the receiver operating characteristic curve (AUC) was used to assess the models discriminative ability. The AUC represents the probability that the logistic regression model will assign a higher probability of the outcome to a randomly chosen patient with an outcome than to a randomly chosen patient without an outcome. An AUC estimate of 0.5 indicates no discrimination, whereas an AUC estimate of 1.0 indicates perfect discrimination. Models with AUC values of 0.700.79 are generally considered to have moderate discriminative properties; those with AUC values 10.80 are generally considered to have good discriminative properties. RESULTS Study population. From July 2000 through June 2003, 302 patients with severe CAP were enrolled in the study. Fourteen patients were excluded from analysis. In 10 patients, a condition other than CAP was eventually diagnosed, and 4 patients had interstitial pneumonia. Chest radiographs were obtained for 227 patients (78.8%) on day 7 and for 195 patients (67.7%) on day 28. In addition, for 52 (18.1%) of the patients, chest radiographs were only obtained on day 7 and not on day 28; in 20 patients (6.9%), radiographs were only obtained on day 28 and were not obtained on day 7 (gure 1). Compared with study patients, the 93 patients (32.3%) for whom no chest radiograph was obtained on day 7 or on day 28 had higher

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Table 3. Pattern of resolution of radiographic abnormalities among patients from whom Streptococcus pneumonia was isolated.
Chest radiograph ndings Time since hospital admission, days 0 7 28
a

No. of patients 62 55 51

Cumulative dropout ratea 0 (0.0) 7/62 (12.7) 13/62 (26.5)

Normal 0 (0.0) 10/55 (18.2) 19/51 (37.3)

Inltrates 62/62 (100) 38/55 (69.1) 28/51 (54.9)

Multilobar pneumonia 28/62 (45.2) 14/55 (25.5) 10/51 (19.6)

Atelectasis 7/62 (11.3) 3/55 (5.5) 4/51 (7.8)

Pleural uid 11/62 (17.7) 23/55 (41.8) 7/51 (13.7)

NOTE. Data are proportion (%) of patients. Cumulative dropout rate reects the no. of patients who died during the acute phase of illness or were lost to follow-up during the study.

986 CID 2007:45 (15 October) Bruns et al.

Table 4. Resolution of chest radiograph abnormalities in patients with infection due to microbial pathogens, compared with clinical improvement and/or cure.
Resolution of chest radiograph abnormalities, by pathogen Time since hospital admission, days 7 28
a

All 57/227 (25.1) 103/195 (52.8)

Streptococcus pneumoniae 10/55 (18.2) 19/51 (37.3)

Atypical pathogensa 4/24 (16.7) 11/24 (45.8)

Overall clinical improvement and/or cure 127/227 (55.9) 152/195 (77.9)

NOTE. Data are proportion (%) of patients. Including Mycoplasma pneumoniae, Legionella pneumophila, and Chlamydia pneumoniae.

PSI scores (mean PSI score, 117 vs. 111; mean difference, 5.9; P p .07), were signicantly older (mean age, 74 years vs. 68 years; mean difference, 6.1 years; P ! .001), and experienced heart failure more often (heart failure rate, 18.2% vs. 8.7%; OR, 2.3; 95% CI, 1.144.83; P p .02). Study patients had a mean age ( SD) of 69.7 13.9 years, 154 patients (53.5%) had at least 1 comorbid illness, and 248 patients (86.1%) had a PSI score of 190. Causative pathogens could be identied for 140 patients (48.6%). The microbial etiology was S. pneumoniae in 62 (21.5%) of the patients and an atypical pathogen in 28 (9.7%). In 50 (17.4%) of the patients, another causative mi-

croorganism was found. (table 1). b-Lactam monotherapy was the most frequently instituted initial antibiotic therapy, administered to 237 (82.3%) of the patients; 41 (14.2%) of the patients received combination therapy with a b-lactam and/or a macrolide, and b-lactam and uorquinolon combination therapy was initiated in 3 (1.4%) of the patients. Time to resolution of chest radiograph abnormalities and factors associated with delay. At day 7, 75 patients (33.0%) had clearance of pulmonary inltrates, and only 57 (25.1%) demonstrated resolution of chest radiograph abnormalities. At day 28, 121 patients (62.1%) had clearance of inltrates, and

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Table 5. Univariate analysis for delayed resolution of chest radiograph abnormalities.

Hospital day, parameter Day 7 Friction rub Dullness to percussion Respiration pain Pleural uid Multilobar pneumonia C-reactive protein level 1200 mg/L Blood urea nitrogen level 110 mmol/L Day 28 Pneumonia severity index score Streptococcus pneumoniae infection Multilobar pneumonia Partial pressure of carbon dioxide !30 mmHg C-reactive protein level 1200 mg/L Blood urea nitrogen level 110 mmol/L 5.28 4.13 (1.2114.05) 1.96 (1.053.66) 2.70 (1.007.26) 2.26 (1.124.60) 2.16 (1.303.57) 2.18 (1.223.88) 10.08 (2.8717.28)b 1.88 (1.153.09) 1.92 (1.033.55) 1.46 (1.012.10) 1.52 (1.152.00) 2.18 (1.132.14)
a

OR (95% CI)

P .08 .02 .03 .04 .02 .002 .005 .006 .01 .04 .03 .002 .003

NOTE. Parameters investigated in univariate analysis included patient characteristics (age, sex, comorbid illnesses, and preexisting chest radiograph abnormalities), clinical scores (pneumonia severity index, APACHE II, and CURB-65 score), symptoms and signs (cough, sputum production, sore throat, dyspnea, chest pain, hemoptoe, confusion, blood pressure, respiratory rate, pulse, oxygen need, crackles, rhonchi, friction rub, and percussion dullness), laboratory data (C-reactive protein level, WBC count, urea, and arterial blood gas analysis), radiograph ndings at admission (pleural uid, atelectasis, pulmonary edema, and multilobar pneumonia), pneumonia etiology (S. pneumoniae, Haemophilus inuenzae, Chlamydia pneumoniae, Legionella pneumophila, atypical pathogens, multiple pathogens, and pathogens of unknown etiology), clinical failure at day 7 and day 28, and inaccurate antibiotic treatment.
a b

Likelihood ratio is displayed. Mean difference is displayed.

Resolution of Radiograph Abnormalities in Severe CAP CID 2007:45 (15 October) 987

Table 6. Multivariate analysis for delayed resolution of chest radiograph abnormalities.

Hospital day, parameter Day 7 Dullness to percussion Multilobar pneumonia Respiratory rate 125 breaths/min C-reactive protein level 1200 mg/L Day 28 C-reactive protein level 1200 mg/L
a

OR (95% CI) 6.94 (1.5232.26) 2.87 (1.296.37) 2.42 (1.085.45) 4.24 (1.839.80) 3.38 (1.676.85)

P .01 .01 .03 .001 .001

Eighty-two patients (28.5%) had deterioration of chest radiograph ndings during follow-up: deterioration of inltrates was seen in 28 patients (9.7%), development of pleural uid was seen in 51 (17.7%), and development of atelectasis was seen in 20 (6.9%). For patients with deterioration of chest radiograph ndings, there were no signicant differences in clinical failure, PSI score, mortality, or interventions (P 1 .09), compared with patients who did not have deterioration of chest radiograph ndings (table 7). DISCUSSION Resolution of chest radiograph abnormalities associated with CAP in immunocompetent, hospitalized patients occurred in approximately one-quarter of the patients at day 7 and in approximately one-half of the patients at day 28. Patients with pneumococcal pneumonia showed signicantly slower resolution. Clinical parameters, such as dullness to percussion, multilobar disease, high respiratory rate, and high C-reactive protein level, were independent predictors for chest radiograph ndings at day 7, whereas a high C-reactive protein level at admission was also a predictor for delayed resolution of chest radiograph abnormalities at day 28. Persistence of chest radiograph abnormalities did not appear to be correlated with inadequate antibiotic treatment or delayed clinical cure. In addition, no signicant differences in outcome were observed in patients with or patients without deterioration of chest radi-

NOTE. Parameters with P ! .1 in univariate analysis were included in multivariate logistic regression analysis. Determined with a mean ( SD) oxygen supply of 1.1 0.009.90 L).
a

1.90 L (range,

103 (52.8%) had resolution of chest radiograph abnormalities. Resolution of chest radiograph abnormalities was signicantly slower for patients with proven S. pneumoniae pneumonia; at day 28, 37.3% had experienced resolution (OR, 1.88; P p .01). When clinical improvement was compared with resolution of chest radiograph abnormalities, improvement in chest radiograph abnormalities lagged behind improvement in clinical condition: at day 7, clinical improvement was observed in 127 patients (55.9%), whereas resolution of chest radiograph abnormalities was only seen in 57 (25.1%). At day 28, clinical cure was observed in 152 patients (77.9%), and resolution of chest radiograph abnormalities was seen in 103 (52.8%) (tables 24). In univariate analysis, serum urea nitrogen levels 110 mmol/ L (OR, 2.18; P .005), C-reactive protein levels 1200 mmol/L (OR, 1.522.16; P .002), and the presence of multilobar inltrates (OR, 1.922.26; P .04) were associated with a delayed clearance of pulmonary inltrates and delayed resolution of chest radiograph abnormalities at both day 7 and day 28. Dullness to percussion at hospital admission was associated with a delayed resolution of radiograph abnormalities at day 7 (OR, 4.13; P p .02) (table 5). Age and comorbidity were not signicantly associated with delayed resolution of chest radiograph abnormalities. In multivariate analysis, multilobar pneumonia (OR, 2,87; P p .01), dullness to percussion (OR, 6.94; P p .01), respiratory rate 125 breaths/min (OR, 2.42; P p .03), and C-reactive protein level 1200 mg/L (OR, 4.24; P p .001) at admission were independently associated with delayed resolution of chest radiograph abnormalities at day 7 (table 6). Figure 2 displays the discriminative ability of these predictors for delayed clearance of chest radiograph abnormalities on day 7 (AUC, 0.71; 95% CI, 0.63 to 0.73). Remarkably, neither preexisting chest radiograph abnormalities, clinical failure, nor inaccurate antibiotic therapy were predictive for delayed clearance of inltrates or delayed resolution of chest radiograph abnormalities at day 7 or at day 28.

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Figure 2. Receiver operating characteristic curve of clinical parameters at hospital admission for delayed radiograph resolution during follow-up. Area under the receiver operating characteristic curve, 0.71 (95% CI, 0.630.79).

988 CID 2007:45 (15 October) Bruns et al.

Table 7. Clinical outcome and deterioration of chest radiograph ndings.

No deterioration of chest radiograph ndings 206/288 (70.1) 128/206 (62.1) 150/206 (72.8) 84.0 24.3 77.5 25.2 15/206 (7.4) 12/206 (6.0) 13/206 (6.4)

Variable No. (%) of patients Clinical improvement at day 7 Clinical cure at day 28 PSI score at day 7, mean value PSI score at day 28, mean value Death Pleuracentesis Bronchoscopy SD SD

Deterioration of chest radiograph ndings 82/288 (28.5) 44/82 (53.7) 62/82 (75.6) 87.7 25 84.1 25.6 5/82 (6.1) 5/82 (6.1) 3/82 (3.7)

OR (95% CI) 1.42 (0.852.38) 0.86 (0.471.56) 3.75 ( 3.13 to 10.62) 6.52 ( 0.98 to 14.02) 0.86 (0.401.89) 1.05 (0.363.08) 0.56 (0.162.02)

P .19 .63 .29 .09 .71 .93 .37

NOTE. Data are proportion (%) of patients, unless otherwise indicated. PSI, pneumonia severity index.

ograph abnormalities during follow-up. Furthermore, the number of interventions in patients with deterioration of chest radiograph ndings was comparable to the number of interventions in other patients, suggesting that physicians decisions were not made solely on the basis of chest radiograph ndings. Routine follow-up chest radiographs, therefore, do not seem to be appropriate for patients who respond to initial therapy. Our study differs from previous studies in several important respects. The strengths of this study were that it involved prospective data collection from multiple centers, that a relatively large number of patients were included in the study, and that strict criteria were used for inclusion. To date, to our knowledge, no studies have analyzed resolution of chest radiograph abnormalities in such a large group of hospitalized patients. Only a few studies, conducted in the past 50 years, have addressed the resolution of chest radiograph abnormalities in patients with CAP. Direct comparison of these studies is difcult because of the heterogeneous composition of the study populations. In the 1950s, 2 studies reported rates of complete resolution of chest radiograph abnormalities of 83%87% by 4 weeks [12, 13]. Major weaknesses of these studies were the inaccuracy of the denition of clearance and the inclusion of only relatively young patients (70% of the patients were !50 years of age). Jay et al. [14] showed that 12.5% of hospitalized patients with bacteremic pneumococcal pneumonia had complete resolution at 2 weeks, and 41.4% had resolution at 4 weeks. Other studies reported that resolution of chest radiograph abnormalities was fastest in patients with pneumonia due to M. pneumoniae, followed by patients with pneumonia due to Chlamydophila psittaci and patients with pneumonia due to S. pneumonia [17, 25]. These results are in line with the rate of resolution in patients with pneumococcal pneumonia in our study. Mittle et al. [15] conducted a study involving 81 inpatients and outpatients and reported resolution rates of

50.6% at 2 weeks and 63.7% at 4 weeks. Factors inuencing the resolution rate were age and the involvement of single versus multiple lobes. A limitation of this study was that a microbiologic diagnosis could be established in only 19 patients [15]. More recently, El Soh et al. [16] investigated the resolution of radiographic ndings of bacterial pneumonia in 74 patients aged 170 years. Of these patients, 35.1% had complete resolution at 3 weeks, 60.2% had complete resolution at 6 weeks, and 84.2% had complete resolution at 12 weeks. According to these studies, age, the presence of comorbid illnesses, and the extent of lobar involvement were factors that inuenced the clearance rate [15, 16]. Compared with our study, only multiple versus single lobe involvement was found to be a predictor for delayed resolution after 1 week, whereas comorbidity and age were not. We could not establish an association between preexisting radiograph ndings and delayed resolution of radiograph abnormalities. On the contrary, in our study, multivariate analysis demonstrated that dullness to percussion at admission and high C-reactive protein level at admission had independent predictive value for a delayed resolution, which, to our knowledge, has never been described previously. The current study also has potential limitations. The study population consisted of patients hospitalized with severe CAP (247 [86.1%] of the patients were PSI class IV or V). However, no radiographic information was available for the most severely ill subgroup of patients. Those patients admitted directly or during follow-up to the intensive care unit were excluded from the original study. In addition, patients for whom no chest radiograph could be obtained at day 28 tended to be severely ill patients. Therefore, because no information was available for the most severe cases, improvement rates found in our study may be overestimates. Moreover, the correlation of clinical ndings, compared with chest radiograph abnormalities, in these severely ill patients remains unclear. On the other hand, underestimation of time to resolution may have occurred, because patients with mild

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Resolution of Radiograph Abnormalities in Severe CAP CID 2007:45 (15 October) 989

pneumonia were not included. However, previous studies have indicated that ndings regarding the resolution of chest radiograph abnormalities among hospitalized patients are also applicable to outpatients if data is corrected for the extent of pulmonary involvement and age [15]. Another potential limitation is that all chest radiographs were reviewed only once by the radiologist on call. However, to mimic real-life clinical practice, and because data from previous studies showed great interobserver and intraobserver variability (k, 0.460.60) [16, 26, 27], we did not have radiographs reviewed by 11 radiologist or reviewed twice by the same radiologist. Obviously, different interpretation of radiographs would have led to different outcomes. In line with this, the lack of a standardized system for reviewing chest radiographs and lack of consensus about the denition for clinical resolution impair studies evaluating chest radiographs. To present reliable and objective results regarding the resolution of chest radiograph abnormalities, there is a need for standardized criteria and unequivocal denitions, which are currently lacking. In conclusion, routine chest radiographs obtained during the short-term follow-up of CAP in patients who respond clinically to the initiated antibiotic therapy, as well as chest radiographs obtained prior to hospital discharge (as advised by the American Thoracic Society in their 1993 guideline [8]), seem to be unnecessary. Observing the clinical course of the patient in the rst stages after diagnosis is far more important than followup with chest radiographs. Exceptions should be made for patients who fail to respond to treatment, who develop complications, or who experience an acute deterioration in their clinical condition of unknown cause. On the basis of our data, follow-up chest radiography to exclude noninfectious abnormalities should not be performed within 4 weeks after the initial diagnosis. A delay of at least 812 weeks after an episode of CAP seems to be reasonable. However, further studies to specically address this issue are needed.
Acknowledgments
Potential conicts of interest. All authors: no conicts.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

References
1. Oosterheert JJ, Bonten MJ, Hak E, Lammers JW, Schneider MM, Hoepelman IM. The increase in pneumonia-related morbidity and mortality among adults in the Netherlands and possible explanations for it [in Dutch]. Ned Tijdschr Geneeskd 2004; 148:17659. 2. Pinner RW, Teutsch SM, Simonsen L, et al. Trends in infectious diseases mortality in the United States. JAMA 1996; 275:18993. 3. Franquet T. Imaging of pneumonia: trends and algorithms. Eur Respir J 2001; 18:196208. 4. Woodhead M, Blasi F, Ewig S, et al. Guidelines for the management of adult lower respiratory tract infections. Eur Respir J 2005; 26: 113880. 5. Granet KM, Wallach SL, Horvath K, Jaeger J. Chest radiographs in

20.

21.

22.

23.

patients with community-acquired pneumonia. N J Med 1996; 93: 3741. British Thoracic Society Standards of Care Committee. Guidelines for the management of community acquired pneumonia in adults. Thorax 2001; 56(Suppl 4):IV164. Bartlett JG, Dowell SF, Mandell LA, File TM Jr, Musher DM, Fine MJ. Practice guidelines for the management of community-acquired pneumonia in adults. Infectious Diseases Society of America. Clin Infect Dis 2000; 31:34782. Niederman MS, Bass JB Jr, Campbell GD, et al. Guidelines for the initial management of adults with community-acquired pneumonia: diagnosis, assessment of severity, and initial antimicrobial therapy. American Thoracic Society. Medical Section of the American Lung Association. Am Rev Respir Dis 1993; 148:141826. Niederman MS, Mandell LA, Anzueto A, et al. Guidelines for the management of adults with community-acquired pneumonia: diagnosis, assessment of severity, antimicrobial therapy, and prevention. Am J Respir Crit Care Med 2001; 163:173054. Schouten JA, Prins JM, Bonten MJ, et al. Revised SWAB guidelines for antimicrobial therapy of community-acquired pneumonia. Neth J Med 2005; 63:32335. Mandell LA, Wunderink RG, Anzueto A, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007; 44(Suppl 2):S2772. Eisenberg GM, Flippen HF, Israel HL, Strandness DE Jr, Weiss W. Delayed resolution of pneumonias. Med Clin North Am 1956; 40: 1291303. Van Metre T. Pneumococcal pneumonia treated with antibiotics: the prognostic signicance of certain clinical ndings. N Engl J Med 1954; 251:104852. Jay SJ, Johanson WG Jr, Pierce AK. The radiographic resolution of Streptococcus pneumoniae pneumonia. N Engl J Med 1975; 293: 798801. Mittl RL Jr, Schwab RJ, Duchin JS, Goin JE, Albeida SM, Miller WT. Radiographic resolution of community-acquired pneumonia. Am J Respir Crit Care Med 1994; 149:6305. El Solh AA, Aquilina AT, Gunen H, Ramadan F. Radiographic resolution of community-acquired bacterial pneumonia in the elderly. J Am Geriatr Soc 2004; 52:2249. Macfarlane JT, Miller AC, Roderick Smith WH, Morris AH, Rose DH. Comparative radiographic features of community acquired Legionnaires disease, pneumococcal pneumonia, mycoplasma pneumonia, and psittacosis. Thorax 1984; 39:2833. Oosterheert JJ, Bonten MJ, Schneider MM, et al. Effectiveness of early switch from intravenous to oral antibiotics in severe community acquired pneumonia: multicentre randomised trial. BMJ 2006; 333: 1193. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997; 336:24350. Chow AW, Hall CB, Klein JO, Kammer RB, Meyer RD, Remington JS. Evaluation of new anti-infective drugs for the treatment of respiratory tract infections. Infectious Diseases Society of America and the Food and Drug Administration. Clin Infect Dis 1992; 15(Suppl 1):S6288. Halm EA, Fine MJ, Marrie TJ, et al. Time to clinical stability in patients hospitalized with community-acquired pneumonia: implications for practice guidelines. JAMA 1998; 279:14527. Fraser RG, Pare JA, Pare PD, Fraser RS, Genereux GP. Diagnosis of diseases of the chest. Vol. 4. Philadelphia: WB Saunders, 1998:461, 494511, 54573, 6637. Virkki R, Juven T, Mertsola J, Ruuskanen O. Radiographic followup of pneumonia in children. Pediatr Pulmonol 2005; 40:2237.

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

990 CID 2007:45 (15 October) Bruns et al.

24. Hosmer DW, Lemeshow S. Applied logistic regression. New York: Wiley, 1998:13575. 25. Vilar J, Domingo ML, Soto C, Cogollos J. Radiology of bacterial pneumonia. Eur J Radiol 2004; 51:10213. 26. Hazir T, Nisar YB, Qazi SA, et al. Chest radiography in children aged 259 months diagnosed with non-severe pneumonia as dened by

World Health Organization: descriptive multicentre study in Pakistan. BMJ 2006; 333:629. 27. Hasley PB, Albaum MN, Li YH, et al. Do pulmonary radiographic ndings at presentation predict mortality in patients with community-acquired pneumonia? Arch Intern Med 1996; 156:220612.

Downloaded from cid.oxfordjournals.org by guest on September 29, 2011

Resolution of Radiograph Abnormalities in Severe CAP CID 2007:45 (15 October) 991