Skip to main content

Table 1 Model input parameters. Parameters used for intended and typical practice models

From: Cost-effectiveness of using the Cervex-Brush (broom) compared to the elongated spatula for collection of conventional cervical cytology samples within a high-burden HIV setting: a model-based analysis

Parameter Mean CI 95 % Distribution Source
   p5 p95   
Women tested (annual count) 723,829 583,056 869,393 Normal NHLS database
Presence of endocervical cells broom vs spatula, odds ratio 1.57 1.44 1.70 Normal [7]
Probability of detection of atypia/dyskaryosis in smears with enodcervical cells vs those without, odds ratio 1.89 1.80 1.98 Normal [7]
Probability of detection of severe atypia/dyskaryosis in smears with enodcervical cells vs those without, odds ratio 2.46 1.75 3.17 Normal [7]
HIV-infected (%) 51 40 64 Binomial NHLS database
HIV-infected women tested (annual count) 372,340 261,707 493,988 Normal Calculated
Abnormal Pap smear (% of all smears) 73 71 75 Binomial [9]
HSIL (% of all smears) 31 29 33 Binomial [9]
LSIL (% of all smears) 37 35 39 Binomial [9]
ASC-US (% of all smears) 5 4 6 Binomial [9]
Endocervical cells present, spatula (% of all smears) 53 52 54 Binomial NHLS database
Endocervical cells present, broom (% of all smears) 83 76 90 Binomial Calculated
Sensitivity (%, if endocervical cells present) 76 72 80 Binomial [9]
Sensitivity (%, if no endocervical cells present) 40 35 45 Binomial Calculated
Sensitivity HSIL (%, if no endocervical cells present) 32 23 44 Binomial Calculated
Repeat Pap smear done (% of patients) 90 85 95 Binomial Assumption
Smear satisfactory for evaluation (% of all smears) 98 98 99 Binomial NHLS database
Specificity (%) 84 81 87 Binomial [9]
Lost to follow-up (% of patients) 8 6 10 Binomial [10]