Original Contribution
Vital-sign abnormalities as predictors of pneumonia in adults with acute cough illness

https://doi.org/10.1016/j.ajem.2006.11.031Get rights and content

Abstract

Purpose

This study examines the strength of the association between vital-sign abnormalities, advanced age, and the diagnosis of community-acquired pneumonia (CAP) in the evaluation of adults with acute cough illness.

Methods

A random sample of adult visits for acute cough to 15 EDs during the winter period of 2 consecutive years (2003-2005) was selected for medical record abstraction. Visits were initially sampled based on discharge diagnoses for a broad range of acute respiratory tract infection diagnoses. Participating sites were a national sample of EDs in Veterans Administration and non–Veterans Administration hospitals stratified across the US region.

Results

Of 4464 charts reviewed, 421 had a diagnosis of CAP based on physician discharge diagnosis and radiographic findings. Age greater than 50 years and vital-sign abnormality (including fever, hypoxemia, tachycardia, or tachypnea) were the only significant predictors of CAP. Hypoxemia had the strongest association with CAP diagnosis (odds ratio, 3.5; 95% confidence interval, 2.4-5.0). A greater number of abnormalities were associated with a higher prevalence of CAP, from 12% with 1 abnormality to 69% with 4 vital-sign abnormalities (P < .001). Most vital-sign abnormalities were predictive of CAP regardless of age.

Conclusions

Increases in vital-sign abnormalities are associated with a greater probability of CAP, and the strength of the association does not vary substantially by age.

Introduction

In adults presenting to the ED with acute cough illness, a small but clinically important proportion will be diagnosed as having community-acquired pneumonia (CAP). The prevalence of CAP among adults with acute cough varies according to patient population, ranging from 3% to 7% in the ED [1], [2]. As 1.3 million of the 5 million annual cases of pneumonia are treated in that setting [3], the evaluation of these patients represents an important function for ED physicians. With its potential morbidity and mortality, CAP is a diagnosis that must be considered early in patients presenting with cough.

Making this diagnosis is important because it responds to antibiotics, and a delay in treatment is associated with increased mortality [4], [5]. However, CAP must be differentiated from acute respiratory tract infections (ARIs) with viral causes, where antibiotic therapy is not indicated. The challenge is that the differential diagnosis for acute cough is broad, and no finding or combination of findings from history or physical exam is sufficient to differentiate CAP from alternative diagnoses [6], [7].

Previously described prediction rules [1], [8], [9], [10] have shown that vital-sign abnormalities are key indicators of radiographic pneumonia, having higher accuracy and specificity than physician judgment [2]. A detailed analysis of these rules found that an initial exclusion of CAP required normal vital signs [11]. However, even these signs are not specific to pneumonia [12], requiring additional evidence to make the diagnosis. Professional guidelines recommend that CAP be diagnosed when clinical suspicion is supported with radiographic evidence [13], [14], [15]. Therefore, a vital-sign abnormality is an indication for a chest radiograph (CXR) in patients presenting with acute cough [11]. However, in developing these prediction rules, the conventional bounds of normal vital signs have been used and the optimal cutpoints have not been clearly defined.

There are also situations in which the interpretation of vital-sign abnormalities is less clear. Elderly patients are known to have atypical presentations of pneumonia, with fewer symptoms, and more subtle findings [16], an observation that dates back to Osler [17]. The reliability of the prediction rules in elderly patients has not been well described.

The primary aims of this study were to examine the strength of the association between vital-sign abnormalities and CAP as (1) the number and degree of vital-sign abnormality increases; and (2) as a function of patient age. We hypothesized that a higher number and more severe vital-sign abnormalities would be associated with a greater likelihood of CAP. In addition, we hypothesized that the predictive value of vital-sign abnormalities for the diagnosis of CAP would be lower among elderly patients.

Section snippets

Methods

We conducted an analysis of ARI management in EDs as part of the Improving Antibiotic Use in Acute Care Treatment Project, a cluster randomized trial to improve antibiotic prescribing at 8 pairs of VA medical centers and non-VA hospitals. The non-VA sites are members of the Emergency Medicine Network (EMNet, http://www.emnet-usa.org) and the selection of study sites was previously described [18]. One of the 8 non-VA sites dropped out before data collection and is not included in these analyses.

Results

We sampled a total of 4464 visits among patients with ARI and a presenting symptom of cough. The characteristics of the study population are displayed in Table 1. The majority of the patients were male (68%), and most were aged 50 years or less (53%), but 21% were aged 65 years or older.

Of all visits sampled, 557 patients (12%) received a discharge diagnosis of CAP, among whom 525 (94%) had a CXR, and 421 (76%) had a final radiologic interpretation compatible with CAP. Of these 421 patients

Discussion

The diagnosis of CAP poses an important challenge in patients presenting with acute cough. As expected from previous reports [1], [2], [8], [9], [10], [11], our results confirm that older age and vital-sign abnormalities are associated with the diagnosis of CAP. Our results also demonstrate that as the number or degree of abnormal vital signs increases, the association with CAP becomes stronger. Advanced age did not significantly weaken this association.

These results are consistent with

Acknowledgments

We are grateful to the members of our National Advisory Committee for their timely input and feedback: Rich Besser, MD; Kitty Corbett, PhD; Stanley Edinger, PhD; Bradley Frazee, MD; Mathew Samore, MD, PhD; Merle A Sande, MD; Sanford Schwartz, MD.

We also acknowledge the contributions of all of the site investigators within the Improving Antibiotic Use in Acute Care Treatment trial (in alphabetical order): Sherrie Aspinall, PharmD (Pittsburgh VAMC); Brian Catto, MD (Augusta VAMC); Cameron

References (20)

There are more references available in the full text version of this article.

Cited by (0)

Funded by the Translating Research into Practice initiative, jointly sponsored by the Agency for Healthcare Research and Quality (grant no. 1 R01 HS013915) and the Health Services Research and Development Service of the Department of Veterans Affairs (AVA-03-239).

View full text