Table 1.
Studies evaluating the performance of a digital colposcopy using a smartphone for cervical cancer screening in LMIC.
| Study | Population | Intervention and Device | Outcome and Results | Strengths and Weaknesses |
|---|---|---|---|---|
| Mungo et al., 2021 [19] | Western Kenya 25–49 y HIV-positive n = 164 * |
D-VIA images taken by nonphysicians. Samsung J8; three off-site expert colposcopists assessed images. |
Outcome: performance to detect CIN2+ (off-site) and acceptability of D-VIA. Results: Se ranging from 21.4% (95% CI, 0.06 to 0.43) to 35.7% (95% CI, 0.26 to 0.46). Sp between 85.5% (95% CI, 0.81 to 0.90) to 94.9% (95% CI, 0.92 to 0.98). 99.4% of women were comfortable with the use of a smartphone. |
Comment: low sensitivity, very good acceptability. Strengths: histology as reference standard. Limitations: HIV population. |
| Goldstein et al., 2019 [20] | China (rural Yunnan areas) 35–65 y n = 216 * |
VIA and digital images. Samsung Galaxy J5 Pro (mobile ODT system). |
Outcome: performance to detect CIN1 and CIN2+ Results: Se: NR, Sp: NR. |
Comment: accuracy of D-VIA to differentiate between CIN1 and CIN2+ Strengths: histology as reference standard. Limitations: low observed prevalence of HPV (6%), small number of CIN2+ (n = 15). |
| Thay et al., 2019 [21] | Cambodia 30–49 y n= 250 HPV-positive = 56 ** |
VIA and digital images. Samsung Galaxy J5 Pro (mobile ODT system). |
Outcome: differentiation between CIN1 and CIN2+. Results: Se: NR, Sp: NR. |
Comment: accuracy of D-VIA to differentiate between CIN1 and CIN2+. Strengths: histology as reference standard (but only in case of CIN2+ suspicion). Limitations: study setting in an urban hospital, results might not be applicable to rural regions, few CIN2+ lesion (n = 4). |
| Tran et al., 2018 [18] | Madagascar 30–69 y n = 125 * |
Forty-five gynecologists (different levels of expertise) assessed D-VIA images. Smartphone Galaxy S4/S5. |
Outcome: performance to detect CIN2+. Results: Se 71.3% (95% CI 67–75.7); Sp 62.4% (95% CI 57.5–67.4) |
Comment: visual assessment demonstrated relatively high Se. Strengths: histology as reference standard. Limitations: small sample size (19 CIN2+). |
| Gallay et al., 2017 [17] | Madagascar 30–65 y n= 56 * |
Four clinicians assessed D-VIA images and classified them in an app called “Exam”. Smartphone Galaxy S4/S5. |
Outcome: evaluation of image quality and inter-observer agreement. Results: adequate quality for visual assessment in 93.3% of cases. Moderate inter-observer agreement, with kappa value = 0.45 (0.23–0.56). |
Comment: small study, designed only for quality of images. Limitations: no histology for diagnosis confirmation. |
| Urner et al., 2017 [22] | Madagascar 30–69 y n = 187 * |
Fifteen clinician evaluated D-VIA images (off-site). Samsung Galaxy S4/S5. |
Outcome: performance in the detection of CIN2+. Results: Se 94.1% (95%CI 81.6–98.3); Sp 50.4% (95%CI 35.9–64.8). |
Comment: Se to detect CIN2+ lesion better than generally reported. Strengths: histology as reference. Limitations: small sample size and limited number of CIN2+ (n = 14). |
| Catarino et al., 2015 [23] | Madagascar 30–65 y n = 137 * |
Comparison of VIA (on-site) and D-VIA (off-site). Samsung Galaxy S4/S5. |
Outcome: performance to detect CIN2+ and inter-observer agreement. Results on-sites: Se 66.7% (95%CI 30–90.3); Sp 85.7% (95%CI 76.7–91.6). Results off-site: Se 66.7% (95%CI 30–90.3); Sp 82.3% (95%CI72.4–89.1). Moderate to poor inter-observer agreement: kappa 0.28. |
Comment: higher Sp than generally reported, demonstration that off-site assessment is feasible. Strengths: histology as reference Limitations: 30.7% drop-out rate, small sample size |
| Ricard-Gauthier et al., 2015 [16] | Madagascar 30–65 y n = 122 * |
Comparison of VIA and D-VIA (on-site) and D-VIA (off-site). Samsung Galaxy S4. |
Outcome: performance to detect CIN 2+. Results on-site: Se 28.6% (95%CI 3.7–71%), Sp 87.2% (95%CI 77.7–93.7%). Results off-site: Se ranging from 42.9 (95%CI 9.9–81.6) to 85.7% (95%CI 42.1–99.6); Sp from 48.1 (95% CI 38.5–59.7) to 79.2% (95%CI 68.5–87.6). |
Comment; Off-site assessment feasible, lower Se for on-site assessment than reported in literature. Strengths: histology as reference. Limitations: 27.9% drop-out rate, small sample size |
Abbreviations: CIN (cervical intraepithelial neoplasia), DC (digital colposcopy), D-VIA (smartphone-based visual inspection with acetic acid), D-VILI (smartphone-based visual inspection with Lugol iodine), ECC (Endocervical curettage), HPV–Hr (human papilloma virus–high risk), HPV-positive (human papilloma virus positive), NR (not reported), Se (sensitivity), Sp (specificity), y (years old). * All HPV-positive; ** 56/250 women were HPV-positive.