TABLE 5:
Cytologic adjuncts to evaluate clinically evident suspicious lesions.*
| TEST RESULT | DOWNSTREAM CONSEQUENCES | EFFECT PER 100,000 PATIENTS TESTED (95% CONFIDENCE INTERVAL [CI]) | NUMBER OF LESIONS (STUDIES) | QUALITY OF THE EVIDENCE (GRADE)§ | |
|---|---|---|---|---|---|
| Prevalence 0.25%† | Prevalence 2%‡ | ||||
| True Positives (Patients With Need for Biopsy) | Patients will be correctly identified as having a potentially malignant or malignant, and timely referral to a specialist or biopsy will be performed. | 230 (215 to 245) | 1,840 (1,720 to 1,960) | 2,148 (15) | Low¶,#,** |
| False Negatives (Patients Incorrectly Classified as Not Having Need for Biopsy) | Appropriate diagnostic would be missed, worsening the prognosis of the disease. | 20 (5 to 35) | 160 (40 to 280) | ||
| True Negatives (Patients Without Need for Biopsy) | Patients will receive reassurance that they do not have a potentially malignant or malignant disorder. | 93,765 (87,780 to 98,753) | 92,120 (86,240 to 97,020) | 2,148 (15) | Low¶,#,** |
| False Positives (Patients Incorrectly Classified as Having Need for Biopsy) | Patients would be incorrectly identified as having a potentially malignant or malignant disorder and would undergo additional unnecessary testing and biopsy. | 5,985 (997 to 11,970) | 5,880 (980 to 11,760) | ||
Setting: primary care. Pooled sensitivity, 0.92 (95% confidence interval [CI], 0.86 to 0.98). Pooled specificity, 0.94 (95% CI, 0.88 to 0.99). Positive likelihood ratio, 14.18 (95% CI, 5.82 to 34.59). Negative likelihood ratio, 0.08 (95% CI, 0.04 to 0.18). Sources: Delavarian and colleagues,32 Fontes and colleagues,33 Kammerer and colleagues,34 Koch and colleagues,35 Mehrotra and colleagues,37 Nanayakkara and colleagues,38 Navone and colleagues,40 Navone and colleagues,39 Ng and colleagues,41 Rahman and colleagues,42 Scheifele and colleagues,43 Sciubba,44 Seijas-Naya and colleagues,45 Svirsky and colleagues,46 and Trakroo and colleagues.47
We estimated the prevalence by using data from the National Cancer Institute Surveillance, Epidemiology, and End Results Program (300,682 people living with oral cavity and pharynx cancer in the United States in 2013) and the 2010 census data for adults 45 years or older collected by the US Census Bureau.
The panel provided illustrative prevalence as an estimation of the number of histopathologic diagnoses from dysplasia to cancer.
GRADE: Grading of Recommendations Assessment, Development and Evaluation.
Patient selection and exclusion from analysis were inappropriate, index and reference tests were conducted in an unblinded fashion, and in some cases the time between index and reference test was greater than 2 weeks. It was unclear whether all participants received the reference test. Poorquality reporting did not provide sufficient information to judge key risk of bias domains.
Investigators conducted most studies in secondary and tertiary care settings. Most patients had a higher probability of having a malignant or potentially malignant disorder.
The positivity threshold for the reference test included from mild dysplasia to cancer in all studies except for those of Kammerer and colleagues,34 Navone and colleagues,39 and Rahman and colleagues.42 The positivity threshold included atypia for Rahman and colleagues,42 Scheifele and colleagues43 (10/96), Sciubba (52/298), and Svirsky and colleagues.46 Parentheses indicate the number of atypical results out of the total (atypical þ positive results).