Original article
Sensitivity and specificity of diagnostic tests in acute maxillary sinusitis determined by maximum likelihood in the absence of an external standard

https://doi.org/10.1016/0895-4356(94)90078-7Get rights and content

Abstract

This study shows how to obtain maximum likelihood estimates of test sensitivities and specificities in case of lack of an external standard, using the Expectation Maximisation (EM) algorithm. This method is used to compare four diagnostic tests in patients suspected of acute maxillary sinusitis. Data were analyzed from published studies. Antral aspiration is the test with the highest diagnostic value. The diagnostic value of a positive clinical examination (according to explicit criteria) and of a positive radiograph or ultrasound are comparable. A negative radiograph is of more diagnostic value than a negative clinical examination or ultrasound. The width of the confidence intervals may be too small, due to model deviations which may give incorrect standard errors. However, the estimated likelihood ratios adequately reflect the relative value of the diagnostic tests considered, even when the assumption of independence is dropped.

References (21)

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

Cited by (41)

  • Problems in detecting misfit of latent class models in diagnostic research without a gold standard were shown

    2016, Journal of Clinical Epidemiology
    Citation Excerpt :

    The latent class model combines the information from multiple, generally three or more, imperfect diagnostic tests to uncover the unobserved disease structure. This approach has, for example, been used to study the diagnostic value of immunohistochemical assays of bladder tumors [10], to evaluate diagnostic tests to detect visceral leishmaniasis [11,12], to estimate diagnostic accuracy of test for acute maxillary sinusitis [13], and the accuracy of surgeons' classifications of bone fracture types [14]. The standard two-class latent class model that accounts for most applications in diagnostic accuracy and disease prevalence studies [15] relies on making two interrelated assumptions: (1) existence of two classes representing groups of true target disease-positive subjects and true target disease-negative subjects and (2) local independence with respect to the imperfect diagnostic test used in the latent class analysis [16]: the outcomes of the diagnostic test are stochastically independent conditional on class membership.

  • Statistical methods for the meta-analysis of diagnostic tests must take into account the use of surrogate standards

    2013, Journal of Clinical Epidemiology
    Citation Excerpt :

    However, all these approaches aimed at corrections either when only results of the test of interest vs. the surrogate were available or in the situation in which each individual received all tests. De Bock et al. [11] suggested a solution when at least two types of reference tests were applied under the assumption that sensitivity and specificity of the candidate test did not change across studies. This assumption often fails for tests, such as the D-dimer, which admit different cutoffs for positivity.

  • Antimicrobial treatment guidelines for acute bacterial rhinosinusitis

    2001, Disease-a-Month
    Citation Excerpt :

    A study correlating CT scan and B-mode ultrasound findings demonstrated a sensitivity for ultrasound of 72.8% for the maxillary sinuses, 23.1% for the frontal sinuses, and 11.3% for the ethmoids.31 Compared with clinical evaluation, the sensitivity of B-mode ultrasound was 36% and the specificity was 90%.32 Because ultrasound is technique-sensitive, there may be marked variations in the reliability of the information provided.33

View all citing articles on Scopus
View full text