Академический Документы
Профессиональный Документы
Культура Документы
1–7, 2013
Copyright Ó 2013 Elsevier Inc.
Printed in the USA. All rights reserved
0736-4679/$ - see front matter
http://dx.doi.org/10.1016/j.jemermed.2013.08.016
Original
Contributions
Katherine E. Kroll, MD,* Marc A. Camacho, MD, MS,† Shiva Gautam, PHD,‡ Robin B. Levenson, MD,† and
Jonathan A. Edlow, MD*
*Department of Emergency Medicine, †Department of Radiology, and ‡Department of Medicine, Beth Israel Deaconess Medical Center,
Boston, Massachusetts
Reprint Address: Jonathan A. Edlow, MD, Department of Emergency Medicine, Beth Israel Deaconess Medical Center, One Deaconess Road,
West Clinical Center 2, Boston, MA 02215
1
2 K. E. Kroll et al.
Thus, one potential cause of misdiagnosis is premature matic headache. The control group was patients with a di-
diagnostic closure after a patient is found to have chronic agnosis of minor head injury. We enrolled only patients
sinusitis findings on CT, then assuming that this is the who were discharged home from the ED. We specifically
cause of the patient’s headache. The diagnosis of ‘‘sinus- chose discharged patients as our subjects for two reasons:
itis’’ is commonly reported in multiple studies of misdiag- 1) patients with headaches thought to be due to sinusitis
nosed subarachnoid hemorrhage; however, no causality would very rarely be admitted to the hospital and 2) to
with respect to incidental CT findings was reported in avoid the potential confounder of inadvertently including
these studies (9–11). One specific cause of misdiagnosis patients with more significant head trauma whose CT
of a serious headache is incorrectly diagnosing sinusitis scans may show mucosal changes or air fluid levels due
in patients with headache whose brain CT scans were to acute bleeding from underlying skull fractures, rather
reported as showing evidence of chronic sinusitis (12). than mucosal changes related to chronic sinusitis.
Although prior studies have recorded the incidence of The raw data were obtained from medical center data
chronic sinusitis findings in asymptomatic individuals, management services, which queried the hospital’s elec-
none has specifically compared these findings in patients tronic medical record and billing database (Casemix TSI,
with acute headaches vs. a control group. Boston, MA) using a structured search designed to detect
all patients who received a CT scan for these diagnoses
Goals of this Investigation during the study period. Patients were identified using
discharge International Classification of Diseases, Ninth
Our hypothesis was that CT findings of chronic sinusitis, Revision (ICD-9) diagnosis codes (Appendix 1). For the
such as mucosal thickening, occur with equal frequency study group, this excluded patients with specific second-
in patients with atraumatic headache as in control pa- ary etiologies of their headache (e.g., subarachnoid hem-
tients. Our objective was to quantify and compare the fre- orrhage, idiopathic intracranial hypertension). For the
quency of radiographic findings indicative of chronic controls, this excluded patients with major head trauma
sinusitis in ED patients presenting with atraumatic head- (e.g., intracranial hemorrhage, skull fractures) in whom
ache to those presenting with minor closed head injury. nonspecific sinus findings may have been due to acute
blood.
MATERIALS AND METHODS
Methods of Measurement and Data Collection
Study Design
We reviewed patients’ records to extract data on their
We performed a retrospective, single-center descriptive noncontrast head CT scan interpretations and specific de-
review of electronic medical records of consecutive pa- mographic information. One trained data collector (K.K.)
tients who received noncontrast head CT scans in a busy abstracted data from the ICD-9 code-based database as
urban, tertiary care ED between October 2007 and April well as from patient charts and recorded the information
2008 for reasons of either minor closed head injury or onto a prespecified spreadsheet. The abstractor was an
atraumatic headache and who were subsequently dis- emergency medicine resident who was not blinded to
charged home. Our hospital’s institutional review board the purpose of the study.
approved the study. Key demographic variables collected about each case
included age, sex, race/ethnicity, date of visit, and trauma
Setting mechanism. Each patient’s head CT radiologic report was
reviewed for findings consistent with sinusitis and classi-
The study institution is an academic, tertiary care, Level I fied into one of four mutually exclusive categories: 1)
trauma center located in Boston, Massachusetts. The chronic sinusitis, 2) indeterminate for sinusitis, 3) air-
annual ED census is approximately 55,000 patients. fluid levels (AFLs), or 4) no findings of sinusitis. Three
investigators discussed and predetermined the sinusitis
Selection of Participants classification criteria prior to data abstraction.
We defined chronic sinusitis as CT findings of mucosal
We collected data on patients who had a head CT scan thickening, secretions, or opacification involving the eth-
performed while in the ED. Noncontrast head CT scans moid, frontal, or maxillary sinuses. The indeterminate cat-
were ordered at the discretion of the treating teams during egory included CT scans with findings of polyps, polypoid
each subject’s ED evaluation, which resulted in the final thickening, or mucous retention cysts. Any CT report with
ED diagnosis. The indication for each CT scan was deter- mention of an AFL was included in the third category.
mined by reviewing the provider order entry field. The AFLs are associated with acute sinusitis and thus may re-
study group included patients with a diagnosis of atrau- flect a legitimate etiology for an acute atraumatic headache.
Chronic Sinusitis Not Associated with Headaches 3
In the setting of trauma, an AFL may represent blood in the mates and 95% confidence intervals are also reported.
sinuses rather than true sinusitis. The fourth category in- A p-value of < 0.05 was considered significant. Cohen’s
cluded all CTs with no evidence of sinus disease, as well k statistic was used to determine inter-rater reliability.
as CTs with negligible findings such as concha bullosa, If the k value was below 0.40, inter-rater reliability was
small septae, or sphenoid sinus abnormalities. poor to fair; if k was between 0.41 and 0.60, moderate;
Because radiologists may not comment on sinus find- if k was between 0.61 and 0.80, substantial reliability;
ings considered to be incidental or nonetiologic in the fi- and almost perfect if k was between 0.81 and 1.00.
nal report of a CT scan, which would decrease the An a priori power analysis designed to achieve at least
frequency of sinus disease noted in our study and act as 80% power to detect a 10% difference between the two
a confounder, one investigator (R.L.), a board-certified groups (a = 0.05) was performed, which found a mini-
emergency radiologist, reviewed the CT images for 50 mum sample size of 394 (197 patients per group). We
random patients (25 in the atraumatic headache group, chose a round number greater than this to ensure that
and 25 in the head injury group) for evidence of sinusitis, our study was not underpowered. Statistical calculations
with the aim of determining the degree of agreement be- were performed on a Statistical Analysis Systems pack-
tween the official radiologic interpretation as compared age (SAS version 9.2; SAS Inc, Cary, NC).
to her own interpretation with regard to findings of sinus-
itis. At the time of her review, she was blinded both to the RESULTS
purpose of the study and to the official contemporaneous
interpretation. Consistent with routine practice, she did We enrolled 500 consecutive patients who had noncontrast
know the indication for the CT scan – atraumatic head- head CT scans, 234 in the nontraumatic headache group
ache vs. head trauma. These 50 CT scans were randomly and 266 in the minor head injury group. Table 1 summa-
selected using a random number generator. Inter-rater rizes the baseline characteristics for the two groups.
reliability was tested using Cohen’s k statistic. Mean patients’ ages were similar and a majority of
subjects overall were women (64.6%) and white
Data Analysis (63.2%). More women were enrolled in the atraumatic
headache group (70.5%) than in the minor head injury
Descriptive statistics, including proportions, medians, group (59.4%; p = 0.01). Moreover, fewer white subjects
means, SDs, and confidence intervals, are reported for and more black subjects were enrolled in the atraumatic
the two groups. Continuous data were summarized by headache group than in the minor head injury group. Of
means and SDs; categorical data were summarized by the head injury group, falls were the most common type
frequencies, proportions, and percentages. Point esti- of traumatic injury.
Table 1. Patient Characteristics of Atraumatic Headache and Minor Head Trauma Groups
Demographic Category Total Atraumatic Headache Minor Head Injury Difference (95% CI) p-Value
As hypothesized, we found similar rates of CT find- mention minor findings such as a very thin layer, tiny
ings consistent with chronic sinusitis in the minor head rounded lesion, or very small-volume AFL.
injury group (47/266; 17.7%) as in the atraumatic head- Studies of both adult and pediatric populations have
ache group (52/234 patients; 22.2%; p = 0.20), depicted found the rate of incidental sinus abnormalities on CT
in Table 2. The two groups were also similar in regards scans of asymptomatic patients of approximately 30–
to the frequency of patients falling into the CT categories 40% (3). Other researchers have shown that even when
of indeterminate, negative, and air-fluid levels. a patient has both symptoms presumed to be referable to
Our investigator agreed with the original radiologic the sinuses and a CT scan showing evidence of sinus dis-
read in 92% of cases in the nontraumatic headache group ease, the extent of CT findings does not correlate with the
(23/25; k = 0.83) and in 80% (20/25; k = 0.41) of cases in extent of a patient’s symptoms (13,14). Thus, there is little
the head injury group, showing substantial reliability for evidence that sinus abnormalities seen on CT correlate
the nontraumatic headache but only moderate reliability with symptoms. Our study adds to the current literature
for the head injury group. Our radiologist noted findings on this topic by specifically comparing the frequency of
consistent with chronic sinusitis in one more CT scan CT findings of sinusitis in an ED population of patients
than was commented on the original radiology read in with acute atraumatic headache with a control group.
the head injury group (4 vs. 3), and in two more CT scans We chose patients discharged home with a diagnosis of
than the original radiology read in the headache group minor head injury as the control group, reasoning that
(10 vs. 8). findings consistent with chronic sinusitis on a CT scan
obtained due to head trauma would nearly always be
DISCUSSION incidental findings. Documenting similar rates of such
findings in the both the atraumatic headache and the
Emergency physicians frequently order head CT scans in minor head injury group suggests that such findings are
patients with acute headache with the aim of ruling out also incidental in headache patients.
clinically significant pathology, such as intracranial hem- The observed rate of chronic sinusitis findings on CT
orrhages, masses, or other conditions. In addition to pro- between these two groups was similar. The slightly higher
viding information about these pathologies, CT scan rate of chronic sinusitis in the atraumatic headache group
reports frequently include findings related to chronic than in the head injury group (22.2% vs. 17.7%) could
sinus disease. Chronic sinusitis rarely if ever causes an mean that our study was not powered to detect a difference
acute headache, and clinicians may incorrectly diagnose between these groups. However, our sample size well ex-
‘‘sinusitis’’ as the cause of a patient’s headache, prema- ceeded the number suggested by our power analysis to de-
turely stopping the diagnostic evaluation. tect a 10% difference. An alternative explanation is that
Given that sinus findings are inherently incidental on this variance reflects radiologists’ reporting of certain
scans primarily assessing for abnormalities of the brain, findings depending on the chief complaint, or context,
the reporting of such findings (and their degree–minimal for which the CT scan was performed.
to flagrant) is variable. Therefore, even the thinnest of soft There is inherent subjectivity in radiologists’ final im-
tissue layering over any portion of the inner wall of any of pressions. Our data suggest that radiologists more often
the paranasal sinuses could be construed as being abnor- report findings of chronic sinusitis in headache patients
mal. Flat-layering soft tissue is thought to represent mu- than in head trauma patients. There was better inter-
cosal thickening; an AFL implies an intraluminal exudate rater reliability for atraumatic headache CT scans than
in the context of infection, or blood in the context of for head trauma CT scans. This may reflect the influence
trauma. Rounded soft tissue attenuation structures most of the ‘‘chief complaint’’ on whether or not radiologists
likely represent polyps or mucus-retention cysts. Depend- mention certain findings in their final reports. This phe-
ing on the clinical context, a radiologist’s report may not nomenon would make the true rates of chronic sinusitis
Sinusitis Classification Total Atraumatic Headache Minor Head Injury Difference (95% CI) p-Value
even more similar than we observed in our two patient itis in ED patients with atraumatic headaches and patients
groups. with mild head injury. This suggests that CT findings of
chronic sinusitis in patients with atraumatic headache
Limitations may be incidental, rather than causal.
ARTICLE SUMMARY
1. Why is this topic important?
Headache is a common complaint in emergency depart-
ment (ED) patients. Whereas the majority of these pa-
tients have benign primary headache disorders, a small
percentage have more serious, secondary causes such as
subarachnoid hemorrhage, and to avoid missing these sec-
ondary causes, physicians need to be aware of potential
sources of misdiagnosis.
2. What does this study attempt to show?
This study compares the frequency of radiographic
findings of chronic sinusitis in ED patients presenting
with atraumatic headache to a control group consisting
of patients presenting with minor closed head injury.
3. What are the key findings?
We found a similar prevalence of computed tomogra-
phy (CT) scan findings of chronic sinusitis in ED patients
with atraumatic headaches and patients with mild head
injury.
4. How is patient care impacted?
Our finding suggests that CT findings of chronic sinus-
itis in patients with atraumatic headache are incidental,
rather than causal. When evaluating patients with acute
headache in the ED, physicians should not use findings
of chronic sinusitis in clinical decision-making, particu-
larly to exclude other, more serious etiologies of acute
headache.