Digital cervicography using mobile phones with real-time consultation (DCRC) to improve performance of Visual Inspection with Acetic Acid (VIA) in cervical cancer screening of HIV-infected women. A cross-sectional study

Abstract

Introduction: Visual Inspection with Acetic Acid (VIA) has been adopted for cervical cancer screening in Kenya and other Low-Middle Income Countries despite providing suboptimal results among HIV-infected women. It is mostly performed by nurses in health centers. Innovative ways of improving the performance of VIA in HIV- infected women are desired. Objective: To establish the feasibility of screening with VIA and Digital Cervicography with Real-time Consul- tation (VIA-DCRC), and compare its performance to screening with VIA alone among HIV + women. Methods: This was a cross-sectional analytical study of two hundred HIV + women. There were two groups of women who underwent either VIA or VIA/DCRC cervical cancer screening arms. In the VIA/DCRC arm, a trained nurse did the VIA, captured an image of the cervix, uploaded it, and electronically shared it in real-time with three blinded study consultants (gynecologic oncologists) who separately assessed the digital image and clas- sified it as VIA/DCRC positive or negative. Any two opinions of the gynecologic oncologists that concurred were considered as the final diagnosis. All participants who screened positive underwent colposcopy and biopsy prior to treatment. Tissues obtained were subjected to histopathological examination. A fraction (15 %) of those who screened negative for VIA and VIA/ DCRC had random cervical biopsies taken at 12 and 6o’clock positions. We estimated the measures of accuracy using the Bayesian method. Results: The mean age was 39.7 +/- 10.7 years. Average CD4 + count and plasma viral load (log base 10) were 492.2 (SD: 255.3) cells per mm3, and 2.6 (SD: 0.7) copies per ml respectively. None of the women was a smoker. The median (IQR) time taken for at least one gynecologic oncologist to respond to a digital consultation was 2.0 (IQR 1.0, 4.0) minutes, range: 1.0 – 47.0. Overall, 60.5 % were diagnosed with cervical pre-malignancies (VIA: 23.1 % (95 % Credible Bounds (CB): 10.1, 37.5), VIA/DCRC: 37.5 % (95 % CB: 26.6, 50.1)). VIA sensitivity, specificity, positive predictive value and negative predictive value were 28.1 % (95 % CB: 11.2, 6.8), 97.8 % (95 % CB: 93.0, 99.7), 79.8 % (95 % CB: 47.3, 96.8), and 80.4 % (95 % CB: 71.0, 87.5) while that of VIA/DCRC was 69.3 % (95 % CB: 47.8, 89.7), 87.9 % (95 % CB: 76.3, 94.4), 77.6 % (95 % CB: 61.9, 89.3), and 80.3 % (95 % CB: 70.2, 88.9) respectively. Compared to the VIA/DCRC group, there was evidence of better sensitivity, comparable negative predictive value, but poor specificity, RR: 2.46 (95 % CB: 1.06, 6.26), RR: 1.65 (95 % CB: 1.00, 3.50), and RR: 0.90 (95 % CB: 0.78, 0.98) respectively.