Social Media

A Pilot Study to Derive Clinical Variables for Selective Chest Radiography in Blunt Trauma Patients

Description
A Pilot Study to Derive Clinical Variables for Selective Chest Radiography in Blunt Trauma Patients
Categories
Published
of 4
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  TRAUMA/BRIEF RESEARCH REPORT  A Pilot Study to Derive Clinical Variables for Selective ChestRadiography in Blunt Trauma Patients Robert M. Rodriguez, MDGregory W. Hendey, MDGillian Marek, MDRobert A. Dery, MDAnna Bjoring, MD From the University of California, San Francisco (Rodriguez, Hendey), Department of Emergency Services, San Francisco General Hospital, San Francisco, CA (Rodriguez); University of California,San Francisco Fresno Medical Education Program, Fresno, CA (Hendey, Dery); and the Departmentof Emergency Medicine, Alameda County Medical Center, Oakland, CA (Marek, Bjoring). Study objective:  The goal of this pilot study was to determine whether clinical criteria can identify blunt trauma patients with significant acute intrathoracic injury on chest radiograph. Methods:  From January 2003 to May 2004, adult blunt trauma patients who received chestradiographs were prospectively enrolled at 2 urban trauma centers. Exclusion criteria were age lessthan 15 years, penetrating trauma, trauma more than 72 hours before presentation, isolated headtrauma, and Glasgow Coma Scale score less than 14. Before chest radiograph viewing, providersrecorded the following data: mechanism of injury, vital signs including oxygen saturation, patientsymptoms, intoxication, distracting injuries, and the presence or finding of visible chest wall injury,chest palpation tenderness, pain on lateral chest compression, crepitus, and abnormal chestauscultation. Significant acute intrathoracic injury was defined as pneumothorax, hemothorax, aorticinjury, 2 or more rib fractures, sternal fracture, or pulmonary contusion by blinded radiologist chestradiograph interpretation. Results:  Of the 507 enrolled patients, 15 patients were excluded because chest radiograph was notperformed. Significant acute intrathoracic injury was confirmed in 31 of 492 (6.3%) patients.Palpation tenderness and chest pain had the highest sensitivity (90%) as individual criteria forsignificant acute intrathoracic injury, and hypoxia had the highest specificity (97%). The combinationof palpation tenderness and hypoxia identified all significant acute intrathoracic injury with thefollowing screening performance with 95% confidence intervals (CIs): sensitivity 100% (95% CI 91%to 100%); specificity 50% (95% CI 45% to 54%); positive predictive value 12% (95% CI 9% to 17%);and negative predictive value 100% (95% CI 99% to 100%). Conclusion:  In this small sample, the combination of palpation tenderness and hypoxia identified allblunt trauma patients with significant acute intrathoracic injury while potentially eliminating the needfor 46% of chest radiographs. [Ann Emerg Med. 2006;47:415-418.] 0196-0644/$-see front matterCopyright  ©  2006 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2005.10.001 SEE RELATED EDITORIALS, P. 419, 420, AND422. INTRODUCTION Clinician decision rules for ordering cervical spine, ankle,and knee radiographs may considerably decrease radiography use. 1-4 Obtaining routine chest radiographs for blunt trauma patients is a low-yield practice, with rates of intrathoracicinjury ranging from 0% to 13%. 5,6  Given this low yield, a clinical decision rule could similarly decrease the use of chestradiography, thereby lowering medical costs, time, andradiation exposure.The objective of this pilot study was to determine whether we could identify clinical criteria that detect blunt trauma patients with significant acute intrathoracic injury on chestradiograph. Identification of such criteria and determination of significant acute intrathoracic injury frequency facilitate samplesize estimation for a large clinical decision rule derivation andvalidation study to safely reduce chest radiography in blunttrauma. Volume  , .  :  May     Annals  of    Emergency Medicine  415  MATERIALS AND METHODS Study Design and Setting  After institutional review board approval, we conducted a prospective, observational study from January 2003 to May 2004 at 2 urban California trauma (Level I and Level II) centers with emergency medicine residency programs. Selection of Participants Patients who received chest radiographs for blunt trauma during study hours (mostly weekdays) were eligible. Weexcluded patients with any of the following: age less than 15years, penetrating trauma, primary trauma that occurred morethan 72 hours before, isolated head trauma, and Glasgow Coma Scale score less than 14. Because we sought to evaluate patientsin whom chest radiographs were not clearly indicated, we alsoexcluded patients who underwent emergency thoracic orpulmonary procedures before chest radiography, specifically,intubation, needle or tube thoracostomy, thoracotomy, andchest compressions. Additionally, we eliminated patients if clinicians viewed their chest radiographs or chest radiographreports before evaluation. Methods of Measurement and Data Collection Investigators and research assistants identified blunt trauma victims who presented through triage or were brought in by paramedics under any trauma designation. Subjects wereenrolled when chest radiographs were ordered or performed. Inaddition to age, sex, and injury mechanism, research assistantsrecorded presenting vital signs, including room air pulseoximetry saturation. Emergency medicine providers (attending physicians, residents, and physician assistants) then assessedpatients for the following criteria:(1) subjective chest pain,(2) shortness of breath,(3) hemoptysis,(4) visible chest wall injury (ecchymosis, seatbelt mark),(5) chest wall tenderness to palpation,(6) pain on lateral chest compression,(7) chest crepitus,(8) abnormal chest auscultation,(9) intoxication as assessed by providers, and(10) distracting injuries: clinically apparent pain that mightdistract from chest injury. 1 Outcome Measures Based on literature review and investigator consensus, 5-7   wedefined the primary endpoint significant acute intrathoracic injury as pneumothorax, hemothorax, aortic injury (chest radiographaortic injury sign(s) confirmed on subsequent computedtomography [CT]), 2 or more rib fractures, sternal fracture orpulmonary contusion. Because chest radiograph is not a screening test for cardiac contusion and hemopericardium, they were notincluded as significant acute intrathoracic injury. Board-certifiedradiologists blinded to the study’s implementation performed allchest radiographic readings. When chest radiographic findings weresuggestive of aortic injury or when additional emergency department chest imaging was performed (chest CT in all cases),the result of this additional imaging determined the true presenceor absence of significant acute intrathoracic injury. We reviewedadmitted patients’ medical records for inhospital significant acuteintrathoracic injury development but did not conduct follow-upbeyond hospital discharge. Primary Data Analysis  We converted vital signs (excluding temperature) fromcontinuous to dichotomous variables (yes/no) using thefollowing cutoffs: hypotension, systolic blood pressure lessthan 90 mm Hg; tachycardia, pulse rate greater than 100beats/min; tachypnea, respiratory rate greater than 20breaths/min; hypoxia, room air pulse oximetry saturation lessthan 95%. Constructing 2-by-2 tables for each criterion withsignificant acute intrathoracic injury as the other axis, weanalyzed these and the other 10 criteria (chest pain, yes/no)as dichotomous variables.Excluding missing data elements (less than 2% overall) fromcalculations, we computed standard screening tests and 95%confidence intervals (CIs) (method of Clopper and Pearson) 8 using the online statistical calculator at http://members.aol.com/  johnp71/confint.html. 9,10 Because we sought criteria that would detect all significantacute intrathoracic injury, we focused on those with highestsensitivity. Using Microsoft Excel 2000 Microsoft, (Seattle, Editor’s Capsule Summary  What is already known on this topic  Radiologic studies are often routinely ordered for trauma patients. The necessity of such a practice has not beenconfirmed. What question this study addressed   Whether a simple decision rule has the potential toidentify a group of patients who are at extremely low risk of having a clinically important abnormality on chestradiograph. What this study adds to our knowledge  The investigators demonstrated that in this cohort of 492patients, a decision rule that mandates a chest radiographin the presence of hypoxia or palpation tendernessidentified all 31 patients with important chestradiographic findings while potentially eliminating theneed for 46% of radiographs. How this might change clinical practice  This study should not change clinical practice. It is a small derivation study without a validation phase. Thestudy does provide evidence that a larger, morecomprehensive study is warranted.Clinical Variables for Chest Radiography   Rodriguez et al  416  Annals  of    Emergency Medicine Volume  , .  :  May      WA) for data entry, management, and descriptive statisticscalculations, we paired individual criteria with highest sensitivity to find combinations that would capture all significant acuteintrathoracic injury. RESULTS During 17 months, we enrolled 507 patients and excluded15 patients because chest radiographs were ordered but notperformed. The mean age was 41.2  17.4 years, and 63% were male patients. The 4 primary injury mechanisms weremotor vehicle crash (43.7%), fall (20.5%), direct blow withblunt force (16.1%), and auto or pedestrian accident (9.3%).Thirty-one of 492 (6.3%) patients had significant acuteintrathoracic injury (Table 1). Four other patients had chest radiograph signs suggestive of aortic injury but had negativechest CT results.Tenderness to palpation and chest pain had the highestsensitivity and negative predictive value. In this sample, thecombination of palpation tenderness and hypoxia identified allsignificant acute intrathoracic injury (Table 2); imaging only  patients with chest tenderness or hypoxia would have eliminated46% of chest radiographs without missing significant acuteintrathoracic injury. LIMITATIONS The most important limitation is the small sample size;although we report 100% sensitivity, the 95% CI lower limitis 91%, insufficient to change practice. Our goal, however, was not to derive a final decision rule but rather to generatepilot data and to determine whether pursuit of such a rule was feasible. We addressed all other prevalidation proposeddecision rule methodologic standards except intraobserverreliability. 11  Given the simplicity of the criteria, however,intraobserver reliability will likely be high.Our convenience sampling method may have biased our study population, but it is unclear how or if it would significantly alterour findings. Both study sites were teaching hospitals and trauma centers, and therefore our findings may not generalize tononteaching, nontrauma centers. Because we evaluated only patients who received chest radiographs, we cannot determine whether some patients who had palpation tenderness or hypoxia  were not imaged or whether patients who were not imaged hadsignificant acute intrathoracic injury. It is possible that imaging allpatients with palpation tenderness or hypoxia could lead to moreradiography.Finally, our significant acute intrathoracic injury definition may be considered too inclusive or not inclusive enough. Some mightargue that 2 rib fractures are clinically insignificant, whereas othersmay propose that even 1 rib fracture should be included assignificant acute intrathoracic injury in certain patients. Threepatients in our study had a single rib fracture. DISCUSSION Validation of decision rules such as the NEXUS Criteria has shown that selective radiography can be implemented without compromising patient safety. 1  Our study establishespilot data and demonstrates the feasibility of derivation of a rule for selective chest radiography in blunt trauma patients. Assuming a 5% frequency of significant acute intrathoracicinjury and similar criteria evaluation and performance, fewerthan 15,000 patients would be needed to establish 100%sensitivity with a 99.5% lower confidence limit. Table 2.  Screening performance of selected individual and paired criteria. Selected Individualand Paired CriteriaSensitivity, TruePositive/31(95% CI)Specificity, TrueNegative/471(95% CI)Positive PredictiveValue (95% CI)Negative PredictiveValue (95% CI) Abnormal auscultation 31 96 35 96Chest pain 90 48 11 99Hypoxia 29 97 43 95Pain on lateral compression 83 71 16 98Shortness of breath 40 88 18 96Palpation tenderness 90 51 11 99Chest pain  hypoxia 97 (83-100) 47 (42-52) 11 (8-16) 100 (97-100)Palpation tenderness  hypoxia 100 (91-100) 50 (45-54) 12 (9-17) 100 (99-100) Table 1.  Frequency of significant acute intrathoracic injury findings (n  31) and other acute traumatic findings (n  10). Significant Acute Intrathoracic Injury Findings Frequency Multiple (  2) rib fractures 10Pulmonary contusion alone 6Pneumothorax alone 3Hemothorax alone 1Multiple rib fractures and pneumothorax 3Multiple rib fractures, pulmonary contusion,and pneumothorax3Multiple rib fractures and pulmonary contusion 2Multiple rib fractures and hemothorax 1Multiple rib fractures, pneumothorax, andaortic dissection1Pulmonary contusion and pneumothorax 1 Other acute traumatic findings Clavicle fracture 7Proximal humerus fracture 2Acromioclavicular joint separation 1 Rodriguez et al   Clinical Variables for Chest Radiography  Volume  , .  :  May     Annals  of    Emergency Medicine  417   We found that the combination of palpation tendernessand hypoxia detected all cases of significant acuteintrathoracic injury. It is possible, if not likely, however, thatadditional criteria will be necessary to achieve 100%sensitivity in a larger, more heterogeneous study population. Although a clinical decision rule has been proposed forpediatric patients, 7  the adult literature for chest radiographsin blunt trauma reveals only 1 study in which investigatorsdemonstrated that auscultation was 100% sensitive and99.8% specific for hemopneumothorax detection. 2   At least 3limitations of this study should be noted:(1) no other intrathoracic injuries were evaluated;(2) only 7 of 523 (1.3%) patients had hemopneumothorax;and(3) CIs, which would have been wide, were not reported. Abnormal auscultation results would have missed 57% of hemopneumothorax detection and 69% of significant acuteintrathoracic injury in our series.In conclusion, the combination of palpation tendernessand hypoxia identified all blunt trauma patients withsignificant acute intrathoracic injury while potentially eliminating the need for 46% of chest radiographs. A largemulticenter prospective derivation and validation study isfeasible and warranted. Supervising editor:   David L. Schriger, MD, MPH Author contributions:   RMR, GWH, GM, RAD, and ABcontributed to the study design and implementation. RMR,GWH, GM, and RAD contributed to the data analysis andwriting of the manuscript. RMR takes responsibility for thepaper as a whole. Funding and support:   The authors report this study did notreceive any outside funding or support. Publication dates:   Received for publication July 26, 2005.Revisions received September 6, 2005, and October 2, 2005.Accepted for publication October 5, 2005. Available onlineDecember 27, 2005.Presented orally at the national Society for Academic Emergency Medicine conference, May 2005, New York, NY.Reprints not available from the authors. Address for correspondence:   Robert M. Rodriguez, MD,Department of Emergency Services, San Francisco GeneralHospital, 1001 Potrero Ave, San Francisco, CA 94110; 415-206-5875; e-mail rrodriguez@sfghed.ucsf.edu. REFERENCES 1. Hoffman JR, Mower WR, Wolfson AB. Validity of a set of criteria torule out injury to the cervical spine in patients with blunt trauma. N Engl J Med  . 2000;343:94-99.2. Stiell I, Wells G, Vandemheem KL, et al. The Canadian C-SpineRule for radiography in alert and stable trauma patients.  JAMA .2001;286;1841-1848.3. Stiell I, Greenberg G, McKnight R, et al. Decision rules for the useof radiography in acute ankle injuries: refinement and prospectivevalidation.  JAMA . 1993;269:1127-1132.4. Stiell I, Greenberg G, Wells G, et al. Prospective validation of adecision rule for the use of radiography in acute knee injuries. JAMA . 1996;275:611-615.5. Rossen B, Laursen NO, Just S. Chest radiography after minorchest trauma.  Acta Radiol  . 1987;28:53-54.6. Bokhari F, Brakenridge S, Nagy K, et al. Prospective evaluation of the sensitivity of physical examination in chest trauma.  J Trauma  .2002;53:1135-1138.7. Holmes JF, Sokolove PE, Brant WE, Kupperman N. A clinicaldecision rule for identifying children with thoracic injury afterblunt torso trauma.  Ann Emerg Med  . 2002;39:492-499.8. Clopper CJ, Pearson ES. The use of confidence or fiduciallimits illustrated in the case of binomial.  Biometrika  . 1934;20:404-413.9. Pezzullo J. Exact binomial and Poisson confidence intervals.Available at: http://members.aol.com/johnp71/confint.html.Accessed June 19, 2004.10. Confidence interval calculator (v4, November 2002). Available at:http://www.pedro.fhs.usyd.edu.au/utilities/CIcalculator.xls.Accessed September 6, 2005.11. Stiell IG, Wells GA. Methodologic standards for the developmentof clinical decision rules in emergency medicine.  Ann Emerg Med  .1999;33:437-447. Did you know? You can save your online searches and get the results by e-mail. Visit www.annemergmed.com today to see what else is new online! Clinical Variables for Chest Radiography   Rodriguez et al  418  Annals  of    Emergency Medicine Volume  , .  :  May   
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks