164 / 264
bias was low in two studies (2,3), moderate in five studies
(6,16–20), and high in the remaining six studies (21–27).
In the majority of the included studies, it was not clear
whether a consecutive series of patients was included
(6,18,20–27). In two studies only those patients with
available histopathologic results were included (2,3).
Standardization of the index test (CT) was either not
achieved or not adequately described in six studies
(16,17,21,23–27). Slice thickness varied between 0.6
and 1mm in the study performed by Marx et al./Lagleyre
et al. (16,17) and between 1 and 2mm in the study
performed by de Oliveira Vicente (24,25). Two different
types of scanners were used in the study by Mori et al.
(23). In the study performed by Berretini et al. (26) some
CT scans were assessed on a computer-dedicated work-
station, whereas others were manually read from a
printed version of the CT scan. The diagnostic criteria
and number of assessors were not mentioned in the study
performed by Bozorg Grayeli et al. (21). CT settings were
not mentioned in the study performed by Shin et al. (27).
Standardization of the reference test (middle ear surgery)
was either not achieved or not adequately described in
nine studies (6,19–27). In two studies (6,23), middle ear
surgery was not performed in all of the included patients,
but all patients were included in the analyses nonetheless.
In nine studies (6,19–27), the criteria used for diagnosing
otosclerosis intraoperatively were not mentioned. A sub-
stantial amount of outcome data, 37%, 38%, and 52%
respectively, were missing in three studies (22,26,27). A
complete case analysis was performed in all of these
studies. Seven studies characterized by a moderate-to-
high relevance and moderate-to-low risk of bias were
included for data extraction (2,3,6,16–20).
Data Extraction
The study characteristics are presented in Table 3.
There are major dissimilarities between studies regarding
the study population, index test, and reference test. As
mentioned previously, only two studies included patients
with a clinical suspicion of otosclerosis based on the
presence of conductive hearing loss, a normal tympanic
membrane, and absent stapedial reflexes (16–18). In the
other studies patients with an indication for middle ear
surgery or patients with confirmed otosclerosis were
included (2,3,6,19,20). High-resolution, cone beam, mul-
tislice, and helical CTs were evaluated in the included
studies. In five studies (6,16–20) middle ear surgery was
used as the reference test whereas in two studies (2,3)
histopathology was used. We decided to refrain from
pooling the data because of these dissimilarities. Prev-
alences, sensitivities, specificities, and post-test proba-
bilities are presented in Table 4.
FIG. 1.
Flowchart.
CT IMAGING IN OTOSCLEROSIS
Otology & Neurotology, Vol. 37, No. 1, 2016
143