Clinical Radiology (2000) 55, 856860

doi:10.1053/crad.2000.0550, available online at http://www.idealibrary.com on

The Value of the Occipitomental (Waters') View in

Diagnosis of Sinusitis: A Comparative Study with
Computed Tomography
E L I K O N E N *, ME I R F A I B E L * , Y E R O H A M K L E I N B A U M * , M I C H A E L W O L F , A Y A L A L U S K Y ,
C H E N H O F F M A N * , A N A E Y A L * , R I N A T A D MO R *
Departments of *Diagnostic Imaging and Oral Head and Neck Surgery, Chaim Sheba Medical Center
(afliated with the Sackler School of Medicine, Tel-Aviv University), and Health Service Research Unit, Gertner Institute,
Tel Hashomer, 52621, Israel
Received: 23 June 1999 Revised: 31 March 2000

Accepted: 8 April 2000

AIM: Sinus X-rays are still frequently used in the evaluation of paranasal sinusitis. Many radiology
departments nowadays provide the referring doctors with a single Waters' projection. Our purpose
was to evaluate the diagnostic accuracy of a single Waters' view vs high resolution computed
tomography (CT) in the diagnosis of paranasal sinusitis.
SUBJECTS AND METHODS: A total of 134 patients with suspected paranasal sinusitis underwent a
Waters' view X-ray and high resolution CT on the same day. The radiographs were evaluated
independently by nine experienced radiologists, who observed each sinus separately. Sensitivity,
specicity, accuracy, positive and negative predictive values were calculated for each sinus and for
each observer, using the CT ndings as a `gold standard'.
RESULTS: The weighted mean sensitivity for diagnosis of any abnormality in the maxillary sinus
was 67.7%, specicity 87.6%, accuracy 78.6%, positive predictive value 82.5% and negative
predictive value 76.9%. For this sinus the variation between observers was small, however, the
sensitivity for diagnosis of any disease in the frontal and ethmoid sinuses varied widely between
observers (range 1.954.0% and 058.9%, respectively). The sensitivity for the sphenoid sinus was
very low (range 03.8%), even in radiographs which seemed to demonstrate it well.
CONCLUSION: The Waters' view has its limits in the diagnosis of sinusitis of the maxillary sinuses
and its contribution for diagnosing lesions in the remaining sinuses is very poor. Whenever access to
CT is available, a low dose high-resolution CT study of the paranasal sinuses is highly recommended.
Konen, E. et al. (2000). Clinical Radiology 55, 856860.
q 2000 The Royal College of Radiologists
Key words: paranasal sinuses, sinusitis, computed tomography (CT), high-resolution, radiography.

INTRODUCTION

Paranasal sinusitis is a common clinical problem in general

practice. Although radiography is not routinely indicated in
these patients [13], it is still frequently used for the diagnosis
[46]. While some authors still emphasize the value of sinus
radiographs, especially when they are negative [711], others
believe them to be reliable mainly for air-uid level identication [10,12]. Nowadays, many radiologists perform a single
Waters' projection for these patients instead of the previous
three traditional views, i.e., Waters', Caldwell and lateral views
[2,4,13,14]. In our institution we use only the Waters' view for
suspected sinusitis. In order to evaluate its accuracy we
Author for correspondence, and guarantor of study: E. Konen M.D.,
Department of Diagnostic Imaging, Chaim Sheba Medical Center, Tel
Hashomer, 52621, Israel. Fax: 972 3 530 2220; E-mail: konen@inter.
net.il
0009-9260/00/055856+05 $35.00/0

compared the ndings on the sinus X-ray lm with those of

high resolution CT in patients with suspected sinusitis.
SUBJECTS AND METHODS

The study was based on 134 patients with suspected inammatory sinus disease who were referred to our department for
paranasal CT over the period of January 1999 to December
1995. Two groups were included: one group of adult outpatients referred by ENT practitioners for CT who consented
to undergo a sinus X-ray as well (n 110), and a second group
of inpatients, including children, whose sinus radiographs were
inconclusive and were therefore referred for CT for further
evaluation (n 24). Computed tomography and radiography
were performed on the same day in all patients, most of them
within 1 h. All radiographs were taken in the upright position
with a horizontal beam. The patients' ages ranged from 6 to 88
q 2000 The Royal College of Radiologists

857

THE VALUE OF THE OCCIPITOMENTAL VIEW IN DIAGNOSIS OF SINUSITIS

years (mean age, 36.6). There were 58 (43.3%) women and 76

(56.7%) men. In the same period, more than 15 000 sinus
X-rays were performed in our institution.
All 134 radiographs were reported independently on a
standardized form by nine experienced radiologists blinded to
clinical data. A total of 788 sinuses were evaluated in this study.
Each maxillary sinus and each frontal sinus was evaluated
separately for a total of 268 maxillary and 252 frontal sinuses.
Sixteen frontal sinuses in eight children below the age of 13
were excluded, to avoid misinterpretation due to undeveloped
sinuses. The sphenoidal sinuses were considered to be one sinus
as well as the ethmoidal sinuses for a total of 134 sphenoidal
and 134 ethmoidal sinuses.
A ve point scale was used for grading: 0 normal sinus,
1 mucosal thickening, 2 polypoid thickening, 3 air-uid
level, and 4 full opacication. When the observers could not
commit themselves to a denite nding they were asked to
grade the sinus as 5 (inconclusive).
The CT examinations were performed on an Exel Twin CT
system or an Elscint 2400 Elite (Elscint, Haifa, Israel). Patients
underwent our standard protocol which included high resolution 3-mm-thick coronal slices with 5-mm increments. The
window was 2500 H, with a centre of 250 H. The hard copy CT
images were interpreted independently of the plain radiographs
by a single radiologist who had no knowledge of the ndings on
the sinus lms. The results were graded according to the same
criteria as the plain lms.
STATISTICAL METHODS

The interpretations of the paranasal sinus plain lms by nine

observers and the CT results were recorded in a computer
database and were compared for each sinus independently and
for each observer. To estimate the diagnostic performance of
the sinus lm in patients with suspected paranasal sinusitis we
calculated the sensitivity, specicity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) for
each observer and in each sinus, using the CT as a `gold
standard' [15]. Weighted mean and standard deviation were

evaluated for the nine observers. For these calculations, original scores were dichotomized to form two groups: pathology,
scores 14; and normal group, score 0. In order to evaluate the
sensitivity of the Waters' view in more specic abnormalities
(such as sinus opacity, air-uid levels and polyps), additional
calculations were performed for the maxillary sinus when
scores 24 were combined to present the pathology group.
Therefore, score 1 was excluded from analysis. Further separate
evaluation of scores 2, 3 and 4 for each sinus resulted in low
numbers, and was not performed due to limited clinical
signicance. In order to evaluate the accuracy of the Waters'
view with respect to specic abnormality we calculated an
extension index of the accuracy term, namely concordance,
between each specic diagnosis made on plain lms and CT
for each sinus (e.g., the rate of specic correct interpretations,
0, 1, 2, 3 or 4, made on the radiographs when compared with
CT results). Data analysis was performed using the BMDP
(Bio-Medical Data Processing) statistical software [16].
RESULTS

In 97 patients (72.4%), CT revealed abnormal ndings. The

most commonly affected sinuses were the maxillary sinuses (82
patients; 41 with unilateral and 41 with bilateral disease), then
the ethmoidal sinuses (57 patients), followed by the frontal
sinuses (38 patients; 18 with unilateral and 20 with bilateral
disease) and sphenoidal sinuses (26 patients). In 15 of the 97
patients, CT showed disease in the frontal, ethmoid or sphenoid
sinuses with a normal maxillary sinus.

Maxillary Sinus
When all positive CT ndings (scores 14) were combined
to form the group of patients with sinus disease, the weighted
mean sensitivity of the radiographs was 67.7 6 8.4%, weighted
mean specicity was 87.6%, weighted mean accuracy was
78.6%, weighted mean PPV was 82.5% and weighted mean
NPV was 76.9% (Table 1).
When only CT scores 24 were counted (51 maxillary

Table 1 Comparison between the interpretations of Waters' views and high-resolution CT regarding the maxillary sinuses of 134 patients
Observer

True-positive
(n)

False-negative
(n)

False-positive True-negative
(n)
(n)

Inconclusive
answers
(n/%)

Accuracy
(%)

Sensitivity
(%)

Specicity
(%)

PPV
(%)

NPV
(%)

1
2
3
4
5
6
7
8
9
Weighted
mean (%)

67
81
91
98
73
87
74
70
79

54
42
30
23
50
30
30
47
38

11
21
24
27
9
21
17
8
21

2/0.7
2/0.7
4/1.5
2/0.7
1/0.4
9/3.4
27/10.1
9/3.4
8/3.0
2.6 6 2.9

75.6
76.3
79.5
81.2
77.9
80.3
80.5
78.8
77.3
78.6 6 1.9

55.4
65.9
75.2
81.0
59.3
74.4
71.2
59.8
67.5
67.7 6 8.4

92.4
85.3
83.2
81.4
93.8
85.2
87.6
94.4
85.3
87.6 6 4.7

85.9
79.4
79.1
78.4
89.0
80.6
81.3
89.7
79.0
82.5 6 4.5

71.3
74.4
79.9
83.7
73.0
80.1
80.0
74.0
76.3
76.9 6 4.1

134
122
119
118
135
121
120
134
122

Original scores dichotomized to form two groups: pathology, scores 14; and normal, score 0.
PPV, positive predictive value; NPV, negative predictive value.

858

CLINICAL RADIOLOGY

Table 2 Comparison between the interpretations of Waters' views and high resolution CT regarding the frontal sinuses of 126 patients*
Observer

True-positive
(n)

1
6
2
2
3
27
4
10
5
1
6
2
7
4
8**

9
5
Weighted
mean (%)

False-negative
(n)

False-positive True-negative
(n)
(n)

Inconclusive
answers
(n/%)

Accuracy
(%)

Sensitivity
(%)

Specicity
(%)

PPV
(%)

NPV
(%)

48
52
23
42
53
28
37

43

2
4
54
12
2
2
1

0/0
2/0.8
10/5.6
2/0.8
0/0
53/21.0
30/13.5

10/4.0
5.0 6 6.9

80.2
76.1
60.7
76.6
78.2
66.7
71.8

75.8
78.5 6 4.9

11.1
3.7
54.0
19.2
1.9
6.7
9.8

10.4
14.6 6 16.3

99.0
98.0
71.3
93.9
99.0
98.8
99.4

96.9
94.5 6 9.3

75.0
33.3
33.3
45.5
33.3
50.0
80.0

45.5
49.2 6 18.0

80.3
78.7
85.4
81.6
78.7
85.6
82.9

81.4
81.7 6 2.6

196
192
134
186
196
167
180

188

Original scores dichotomized to form two groups: pathology, scores 14, and normal if score 0.
* Eight children younger than 13 years were eliminated.
** This observer did not agree to ll scores for the frontal sinus and therefore was excluded from the analysis.
PPV, positive predictive value; NPV, negative predictive value.

sinuses), the weighted mean sensitivity was 77.3 6 7.9%. Thus,

the sensitivity of the more specic abnormalities (sinus opacity,
air-uid levels and polyps) was higher than the overall scores.
The concordance rate between the specic interpretations
made on plain lms and CT ranged from 60 to 68% (weighted
mean 63.8 6 2.1). The observers could not commit themselves
to a diagnosis regarding the maxillary sinuses in 2.6% of the
radiographs.

Frontal Sinus
Only 252 frontal sinuses were evaluated (after excluding
eight children aged less than 13 years). Sensitivity rate varied
greatly among observers from 1.9% to 54.0% (Table 2). The
weighted mean specicity rate was 94.5%, the weighted mean
accuracy was 78.5%, the weighted mean PPV was 49.2%, and
weighted mean NPV was 81.7%.
The concordance rate between the specic interpretations
made on plain lms and CT ranged between 64 and 84%
(weighted mean 76.7 6 5.5). Frontal sinus opacication was
demonstrated on CT in 10 cases, six of which were not
diagnosed on the radiographs by any of the observers, one
case detected by two observers and three cases by one observer.
One observer did not agree to ll scores for the frontal sinus
and was therefore excluded from this analysis. The other
observers could not commit themselves to a diagnosis
regarding the frontal sinuses in 5.2% of the radiographs.

Ethmoid and Sphenoid Sinuses

The observers would not commit themselves to a diagnosis
on the radiographs in 36.0% (range, 0.760.4%) of ethmoidal
sinuses and in 56.2% (range, 20.170.1%) of sphenoidal
sinuses; in those cases where the observers thought the sinus
to be sufciently visualized the sensitivity ranged widely from
0 to 58.9% for the ethmoidal sinuses, and from 0 to 3.8% for the
sphenoidal sinuses. Due to the high rate of inconclusive
answers, statistical analysis was of limited value and was not
included.

Air-uid levels were present in four maxillary, three frontal

and two sphenoidal sinuses on CT. None of the sphenoidal or
frontal air-uid levels were detected on plain radiographs. A
maxillary air-uid level was diagnosed on the plain lm in one
case by four radiologists (out of nine), and in two cases by two
radiologists; the last one was missed by all. On the other hand, a
false-positive diagnosis of air-uid level was made 42 times
amongst all radiologists on plain lms.
Due to the limited number of children in our series, no
statistically signicant difference in the accuracy of the Waters'
view between children and adults could be determined.
DISCUSSION

Guidelines published by the Royal College of Radiologists

[2] suggest that sinus radiography is not routinely indicated in
the management of sinusitis and should be reserved for specialist request. Although sinusitis is a clinical diagnosis, sinus
radiographs are still commonly used [46] and some authors
still believe in its contribution as a screening test for maxillary
sinusitis, especially when negative [79].
For many years, the Waters' view had been considered as
optimal for imaging of the maxillary sinus. We found, however,
that this radiograph has a weighted mean false-negative rate of
32% (weighted mean sensitivity 67.7%), and a mean negative
predictive value of 76.9% when compared to CT (Table 1).
These data suggest that about one third of all abnormal
ndings in the maxillary sinus might be overlooked when
using one single view, and that in almost a quarter of
negative interpretations based on this view, CT would be
positive.
The discrepancy between CT and radiographic ndings is
even greater regarding the frontal sinus (Table 2). Firstly, only
two observers recorded scores for all the lms. The others
recorded only some of the radiographs. Secondly, an average of
more than 85% of the 58 abnormal ndings in the frontal
sinuses were overlooked on the radiographs (weighted mean
sensitivity 14.6%). Therefore, the predictive value of this lm

THE VALUE OF THE OCCIPITOMENTAL VIEW IN DIAGNOSIS OF SINUSITIS

regarding the frontal sinus is negligible. Similarly, only 49.2%

of the lms diagnosed as positive on the Waters' views were
true-positive. The relatively high concordance between specic
answers in the frontal sinus is of limited value, due to a high
number of cases with normal frontal sinus. The same trend was
observed for ethmoid and sphenoid sinuses. The majority of
abnormal ndings in the ethmoid and sphenoid sinuses were
overlooked on the radiographs.
Special attention should be given to the diagnosis of air-uid
levels, since some authors consider plain radiography to be a
sensitive examination for the identication of air-uid levels in
the maxillary sinuses [10,12]. Since this nding is more
prevalent in acute sinusitis, while most of our cases were of a
chronic character, we only had a limited number of air-uid
levels. Nevertheless, most air-uid levels in our series,
including the maxillary ones, were overlooked by all nine
observers. These ndings are in agreement with earlier surveys
showing the plain lms to be unreliable in detecting this
clinically signicant abnormality [7,17]. In addition we found
a marked tendency for overdiagnosis of air-uid levels on plain
radiographs.
Another interesting nding was a group of 14 patients where
CT showed abnormalities in the frontal, ethmoid or sphenoid
sinuses while the maxillary sinus was normal: most of those
cases were missed on the radiographs. This further increases the
overall false-negative rate of plain radiography in our series.
These results correspond to previous reports which indicated
a low sensitivity for these sinuses, even with three views [17
20]. In a survey regarding the use of sinus radiographs by
general practitioners, Houghton et al. found that over 75% still
requested sinus radiographs once in a while, and that frontal
opacity was one of the radiographic ndings most likely to
inuence the management of patients with suspected sinusitis
[4]. During the same period, 50% of the radiology departments
participating in the survey used only a single Waters' view. Our
results suggest that the Waters' view is not sufcient for
diagnosing sinusitis, particularly in those sinuses in which
inammation may cause life threatening complications, e.g.,
the frontal, ethmoid and sphenoid sinuses. Thus, a negative
interpretation of the sinus radiograph by the radiologist may
have serious consequences.
Although previous reports have shown the unquestionable
superiority of CT over the three traditional views in the
evaluation of paranasal sinusitis [1720], this is the rst
study to determine the relatively poor diagnostic performance
of a single Waters' view in adults and emphasize its relatively
low accuracy regarding each sinus separately. Our ndings give
further support to guidelines published by The Royal College of
Radiologists [1], that sinus radiography should not be used
routinely by general practitioners in patients with suspected
sinusitis. When medical treatment has failed, it is for the ENT
practitioner to decide whether the patient should undergo
radiological evaluation (radiography or low dose screening
CT) or endoscopy. Our study also strengthens the prevailing
view that whenever imaging is indicated in the management of
paranasal sinusitis (e.g., recurrent or persistent sinusitis) a
`limited screening CT', which has a much greater accuracy,
should be the preferred imaging investigation [1,12,18,19,21
23]. With our current CT technique, a 3-mm-slice thickness at
60 mAs exposes the patient to a Computed Tomography Dose

859

Index (CTDI) of 7 mGy (CT TWIN: operation manual/appendix c.5), while a single Waters' view may expose the patient to
5 mGy and the three conventional sinus radiographs expose
the patient to 813 mGy [24], depending on the quality of the
X-ray equipment used.
In summary, our data suggest that the single Waters' view
has limited value in the diagnosis of maxillary sinusitis and is
even less sensitive for abnormalities in the other sinuses. A
normal X-ray of the paranasal sinuses cannot exclude potentially signicant pathology particularly in the frontal, ethmoid
and sphenoid sinuses, and should therefore not be relied upon
as a guide to treatment. We therefore conrm recommendations
[1] that imaging of the paranasal sinuses should be obtained by
a low dose, high resolution CT study, rather than the Waters'
projection X-ray lm.
Acknowledgement. We thank Prof. M. Hertz for editorial assistance in
preparation of the manuscript, the technicians Gregory Muzikansky and
Dani Bashi and the secretaries Rachel Elias and Pnina Butlikov for their
cooperation and support.

REFERENCES
1 Royal College of Radiologists. Making the Best Use of a Department of
Clinical Radiology. Guidelines for Doctors. 4th ed. London: Royal
College of Radiolgoists, 1998.
2 Evans KL. Diagnosis and management of sinusitis. BMJ 1994;309:
14151422.
3 Kronemer KA, McAlister WH. Sinusitis and its imaging in the pediatric
population. Pediatr Radiol 1997;27:837846.
4 Houghton DJ, Aitchison FA, Wilkinson L, Wilson JA. Use of sinus X
ray lms by general practitioners. BMJ 1994;308:16081609.
5 Dobson MJ, Fields J, Woodford T. A comparison of ultrasound and
plain radiography in the diagnosis of maxillary sinusitis. Clin Radiol
1996;51:170172.
6 Low DE, Desrosiers M, McSherry, et al. A practical guide for the
diagnosis and treatment of acute sinusitis. CNAJ 1997;156 (Suppl
6):S114.
7 Peiderer AG, Croft CB, Lloyd GA. Antroscopy: its place in clinical
practice. A comparison of antroscopic ndings with radiographic
appearances of the maxillary sinus. Clin Otolaryngol 1986;11:455
461.
8 De Bock GH, Houwing-Duistermaat JJ, Springer MP, Kievit J, Van
Houwelingen JC. Sensitivity and specicity of diagnostic tests in acute
maxillary sinusitis determined by maximum likelihood in the absence
of an external standard. J Clin Epidemiol 1994;47:13431352.
9 Revonta M, Suonpaa J. Diagnosis of subacute maxillary sinusitis in
children. J Laryngol Otol 1981;95:133140.
10 Willett LR, Carson JL, Williams JW Jr. Current diagnosis and management of sinusitis. J Gen Intern Med 1994;9:3845.
11 Lloyd GAS, Lund VJ, Scadding GK. CT of the paranasal sinuses and
functional endoscopic surgery: a critical analysis of 100 symptomatic
patients. J Laryngol Otol 1991;105:181185.
12 Yousem DM. Imaging of sinonasal inammatory disease. Radiology
1993;188:303314.
13 Hayward MWJ, Lyons K, Ennis WP, Rees J. Radiography of the
paranasal sinuses one or three views? Clin Radiol 1990;41:163164.
14 Williams JW Jr, Roberts L Jr, Distell B, Simel DL. Diagnosing sinusitis
X-ray: is a single Waters' view adequate? J Gen Intern Med 1992;7:
481485.
15 Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed. New
York: John Wiley & Sons, 1981; pp. 221225.
16 Dixon WJ, Brown MB, Engelman L, et al. BMDP Statistical Software
Manual. Berkeley: University of California Press, 1990.
17 Burke TF, Guertler AT, Timmons JH. Comparison of sinus X-ray with
computed tomography scans in acute sinusitis. Acad Emerg Med
1994;1:235239.

860

CLINICAL RADIOLOGY

18 Davidson TM, Brahme FJ, Gallagher ME. Radiographic evaluation for

nasal dysfunction: computed tomography versus plain lms. Head Neck
1989;11:405409.
19 Garcia DP, Corbett ML, Eberly SM, et al. Radiographic imaging
studies in pediatric chronic sinusitis. J Allergy Clin Immunol 1994;
94:523530.
20 McAlister WH, Lusk R, Muntz HR. Comparison of plain radiographs
and coronal CT scans in infants and children with current sinusitis. Am J
Radiol 1989;153:12591264.
21 Witte RJ, Heurter JV, Orton DF, Hahn FJ. Limited axial CT of the

paranasal sinuses in screening for sinusitis. AJR 1996;167:1313

1315.
22 Duvoisin B, Landry M, Chapuis L, Krayenbuhl M, Schnyder P. Low
dose CT and inammatory disease of the paranasal sinuses.
Neuroradiology 1991;33:403406.
23 Dunham ME. Evaluating the limited sinus computed tomography scan
in children. Laryngoscope 1997;107:402404.
24 Dosimetry Working Party of the Institute of Physical Sciences in
Medicine. National protocol for patient dose measurements in radiology.
National Radiological Protection Board, 1992.