Distribution of cervical abnormalities detected by visual inspection with acetic acid in Swaziland, 2011–2014: A retrospective study

Background Cervical cancer is the fourth most common cancer worldwide among women, with the number of new cases increasing from 493 243 in 2002 to 527 000 in 2012. These numbers are likely to be underestimated because given the lack of registration resources, cervical cancer deaths are usually under-reported in low-income countries. Aim To describe the distribution of and trends in visual inspection with acetic acid (VIA) to detected cervical abnormalities in Swaziland by reviewing records of VIA examinations performed at two main hospitals in Swaziland between 2011 and 2014. Setting Mbabane Government Hospital and Realign Fitkin Memorial (RFM). Methods Records of cervical screening using VIA at the Mbabane government hospital and RFM hospital between 2011 and 2014 were retrieved. Positivity rates (PRs) of VIA with 95% confidence intervals (95% CI) were calculated and used as proxies of cervical abnormalities. Odds ratios of the association between VIA-detected cervical abnormalities and human immunodeficiency virus (HIV) status were estimated using logistic regressions. Results VIA was positive in 1828 of 12 151 VIA records used for analysis (15%, 95% CI: 14.4–15.7). VIA was positive in 9% (36 of 403) women under the age of 20, in 15.5% (1714 of 11 046) of women aged 20–49 years and in 11.1% (78 of 624) of women aged 50–64 years. A decreasing trend of VIA positivity was observed over time at both screening centres (p for trend < 0.001). Of 2697 records with Papanicolaou results, 20% (67 of 331) VIA-positives and only 5% (114 of 2366) VIA negatives had high-grade squamous intraepithelial lesion. Among 4578 women with reported HIV status, 1702 were HIV-positive (37.2%, 95% CI: 35.8–38.6). The prevalence of HIV in VIA-positive women was 62.5% (95% CI: 58.7–66.2), almost double that among VIA-negative women (33.0%, 95% CI: 31.6–34.5) and that among all women screened (p < 0.001). HIV-positive women were 3.4 times more likely to have cervical abnormalities on VIA than HIV-negative women (OR: 3.4, 95% CI: 2.8–4.0, p < 0.01). Conclusion The high VIA PRs observed over four years in this study may reflect the prevalence of cervical abnormalities, in particular, in HIV-positive women. VIA is not a robust screening test, but it can play a major role in strengthening and expanding cervical cancer screening prevention programmes in resource-limited countries.

Introduction persistent among HIV-positive women. 6 A significant number of HIV-positive women are now accessing life-prolonging antiretroviral therapy (ART); therefore, there are more highrisk women living longer at an increased risk of developing ICC in developing countries. 6 About 80% of cervical cancers in developing countries like Swaziland are diagnosed at an advanced stage, resulting in poor prognosis. 7 The challenge in these countries is to implement and maintain adequate cervical screening with very limited resources (e.g. infrastructure and health professionals).
The use of Papanicolaou (Pap) smears for early detection of cervical cancer over the past 50 years has led to more than a 70% decrease in the mortality rates in high-income countries, compared to a 19% decrease in low-income countries. 8 In 1983, Pap was implemented at the Swaziland Central Public Health Laboratory, which remains the only cytology laboratory in the country. This laboratory receives cervical smears of mainly symptomatic women attending health institutions from all over the country. Screening was offered only to a small proportion of women in the two main cities of the country: Mbabane and Manzini. In a report of the status of cytology in Swaziland, 5113 of 11 701 (44%) Pap smears, adequate for diagnosis, received at the Central Public Health Laboratory between 2004 and 2006 were abnormal. 9 The prevalence of cytological high-grade squamous intraepithelial lesion (HSIL) was 12% (602 of 5113), more than five times the usually reported 1% -2% in asymptomatic screening populations. 9,10 Since then, many international studies and programmes have provided evidence of the feasibility and cost-effectiveness of screen-and-treat approaches for cervical cancer prevention in countries with limited resources and difficult scenarios. 11,12 Visual inspection of the cervix after the application of acetic acid (VIA) is a more attractive and feasible screening strategy for cervical cancer precursors detection than cytology in low-income countries where high-quality cytology-based programmes are difficult to establish or maintain. 13 Visual inspection with acetic acid is not expensive, can be performed by non-medical health workers and its results are available immediately, allowing treatment of screened positives the same day. In 2009, to fast track the early detection of cervical lesions and facilitate the extension of cervical cancer prevention services in both urban and rural areas, the Swaziland Ministry of Health incorporated the use of VIA followed by treatment with cryotherapy in the national cervical cancer prevention programme.

Cervical cancer screening and management programme in Swaziland
In 1983, the government of the Kingdom of Swaziland (Ministry of Health) introduced the Pap smear for cervical cancer screening at the National Referral Laboratory, which remains the only functional cytology laboratory in the country. This laboratory receives cervical smears of mainly symptomatic women attending health institutions from all over the country.
However, many international researches and programmes have presented evidence of the feasibility and costeffectiveness of 'screen-and-treat' approaches for cervical cancer prevention in countries of scarce resources. Visual inspection of the cervix after the application of acetic acid is a more attractive and feasible screening approach for cervical cancer precursors detection than cytology in developing countries where high-quality cytology-based programmes are hard to institute or sustain. 13 Visual inspection with acetic acid is so cheap and user-friendly, can be achieved by any trained nurse and its results are available immediately, allowing treatment of screened positives the same day. In 2009, to quicken the early detection of cervical lesions and facilitate the extension of cervical cancer prevention services across four political regions, the Swaziland Ministry of Health incorporated the use of VIA, followed by treatment with cryotherapy to the national cervical cancer prevention programme. Other treatment methods used include loop electrosurgical excision procedure (LEEP), which is the treatment of choice when a lesion is too large for the cryoprobe or involves the endo-cervical canal or when a histological specimen is needed. 14 For cervical cancer treatment, according to the national cervical cancer guidelines, patients may be diagnosed by clinicians at clinic level or health centre level and then referred to a regional hospital, which subsequently refers to a National Referral Hospital for further investigations. 14 However, most of the necessary tests, staging equipment and treatments are not available in Swaziland. Therefore, almost all cervical cancer patients are transferred and treated in South Africa, so a heavey economic burden is experienced by the country. Therefore, we conducted a retrospective study to describe the distribution and trends of VIA-detected cervical abnormalities over four years in two hospitals that pioneered the implementation of the VIA programme in Swaziland.

Materials and methods
Records of cervical screening using VIA at the Mbabane Government Hospital and the Realign Fitkin Memorial (RFM) hospital between 2011 and 2014 were retrieved. Visual inspection with acetic acid visits are recorded on designed record books. Nurses performing VIA registered their results themselves and transfer the data to an Excel database with some clerical support.
Following the Ministry of Health guidelines, 14 VIA was reported as: (1) negative if no acetowhite change is observed, (2) positive if an acetowhite area is noticed in the transformation zone and (3) suspicious for invasive cancer if a growth or ulcerative lesion is observed.
Visual inspection with acetic acid records include the date and result of screening, as well as the age and area of residence of women screened at both centres. No histology data were available from either centre. Age was categorised into five groups: 15-19, 20-29, 30-39, 40-49 and 50-64, and records with ages less than 15 and over 64 years were excluded from the analyses. As the number of screened women younger than 20 years was small, this group was merged with those 20-29 years, or age was split into two groups, less than 30 years and 30-64 years, for certain analyses.
In addition, information on Pap smears and the HIV status from women screened at the Mbabane hospital was retrieved from hospital clinical records whenever available. Papanicolaou results were classified following the Bethesda system 15 and further grouped into four (or three) categories for analyses: negative, atypical typical squamous cells of undetermined significance (ASCUS) and/or atypical glandular cells of undetermined significance (AGUS), lowgrade squamous intraepithelial lesion (LSIL) (or ASCUS and/or LSIL) and HSIL or worse (HSIL+) for statistical analysis. Papanicolaou HSIL was considered a proxy of highgrade cervical disease and was used to calculate positive and negative predictive values of VIA.
HIV status collected from clinical data could be based on recorded ART prescriptions or could be self-reported at the time of screening. None of these data were available for the RFM hospital.
Visual inspection with acetic acid positivity included positive and suspicious for invasive cancer results. Positivity rates (PRs) of VIA with 95% confidence intervals (95% CI) were calculated and used as proxies of cervical abnormalities, overall, by age, by Pap results, by screening centre and by year and month of screening. Visual inspection with acetic acid PRs by these characteristics was compared using the t-test for proportions, or the score test for trends as appropriate.
The prevalence of HIV (with 95% CI) was estimated overall by VIA result and age. Odds ratios of HIV status for VIA result (positive vs. negative) were estimated overall and by age using logistic regressions. Potential interactions were evaluated using the likelihood ratio test (LRT). Statistical analysis was carried out using Stata 14.0 SE.

Ethical consideration
This study was approved by the Scientific Ethics Committee of Swaziland (SEC) (reference no.: MH599C/FW00015267/ IRB0009688) and the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (BREC) (reference no.: BE 242/14). As the study was a retrospective study, consent to participate was not applicable.

Results
A total of 13 561 records of women who attended cervical screening from January 2011 to October 2014 at the Mbabane Government Hospital and up to June 2014 at the RFM hospital were reviewed. Records without conclusive VIA results (n = 1254) and those from women outside the age range of 15-64 years (n = 156) were excluded ( Figure 1). Table 1 shows the main characteristics of women screened with VIA, overall, and separately by VIA abnormal results: positive or suspected cancer. Of the 12 151 VIA exams considered for analysis, 76% were performed in women aged 20-39 years, 60% in women from the Manzini region, 32% in those from Hhohho, 6% in women from Lubombo and 2% in those from Shiselweni.   Mbabane over 2011 was very small, while at RFM, more than 100 VIAs were performed monthly from March 2011 onwards; a larger proportion (37%) of VIA exams were conducted in 2013 than in other years. In a multivariate logistic model including age, cytology abnormalities (Pap ASCUS+) and HIV positivity, after adjusting for the other two factors, HIV-positive women were 2.9 times more likely to have a positive VIA (OR: 2.9, 95% CI: 2.3-3.8) than HIV-negative women, similarly, compared to women with normal Pap smears; those with abnormal cytology were 4.8 times more likely to be VIA-positive (Table 4).   Our study showed an overall VIA PR of 15% (95% CI: 14.4-15.7) among women aged 15-64 years with 1.1% of VIA exams classified as suspicious for cervical cancer. In addition, there was a much smaller proportion of VIA classified as suspected cancer in RFM than at Mbabane (0.1% v 2.3%). These differences could be explained by differences in the criteria used to classify exams and by observer variability. Visual inspection with acetic acid exams were conducted by one single nurse in each centre, both of them were trained at the same time by experts in Zambia; therefore, it is unlikely that differences are because of VIA positivity definition. In addition, there are potential issues of overcall versus under-call between the two nurses. It is known that VIA PRs vary largely compared to other screening test results and that VIA heavily depends on human judgement. 16 As only one nurse per centre performed VIA, it is likely that the differences represent observer variability. It has been proposed that this variability can be lessened with the establishment of VIA proficiency schemes that can provide certification to VIA providers. Unfortunately, in poor-resource settings, the availability of training and supervision may be limited, and other options to ensure quality of VIA should be explored. Nevertheless, overall observed VIA PRs are within limits of previous studies. 17,18,19     The prevalence of HIV was 37.2% among 4578 women aged 15-64 years with known HIV status (self-reported or based on clinical records) attending the Mbabane hospital. Even though HIV status was only available for one hospital, the results are likely to be representative of all women in Swaziland, as they are similar to those reported by the 2011 Swaziland HIV Incidence Measurement Survey (SHIMS), in which the HIV prevalence in women aged 18-49 years was 39%. 20 It is clear that cervical abnormalities are more prevalent among HIV-positive women than among HIV-negative women. A review of studies published before 2007 that evaluated the association between HIV infection, HPV infection and cervical neoplasia, and the impact of using highly active ART (HAART) on the association, reported ratios of the prevalence of HPV infection, LSIL and HSIL in HIV-positives compared to HIV-negatives. 21 Among women who participated in 11 studies in Africa, prevalence ratios ranged between 1.0 and 3.6 for HPV, between 1.6 and 8.8 for LSIL and 3.1 and 5.1 for HSIL; thus, prevalence ratios increased by grade of abnormality, confirming that HIV infection is strongly associated with higher HPV-persistent infection, which leads to higher prevalence of squamous intraepithelial lesions, and eventually to higher rates of ICC. 21 However, the evidence was not that clear for HAART use, and although it seems that HAART has little or no beneficial effect on the natural history of cervical lesions in HIV-positive women, its use increases their life expectancy. 21  Invasive cervical cancer cases are reported to a referral scheme called Phalala, whose committee decides whether a cancer case is referred for chemotherapy and radiotherapy to South Africa. This process can take up to four months, and the decision for referral, the treatment given and the clinical status after treatment are generally not reported back to the local doctor. Thus, we do not know the final diagnoses of these women and whether they were treated or not.
To our knowledge, this study represents the first evaluation of VIA results after its introduction in the Swaziland National Cervical Screening programme. The main strength of our study relies on the large number of VIA screens performed in women, residents from all political regions of the country, who attended two major hospitals over four years. However, although the records retrieved, we assumed, were representative of all women in Swaziland, we cannot rule out potential information and reporting bias, as we have used routine-collected data. Therefore, we accept that biases may have been greater because this is an analysis based on routine-collected data, that is, we did not have any control over the data collection. In fact, we were able to retrieve data on Pap smears and HIV-reported status only from a small subset of women. In addition, although we can assume that at least in the two hospitals included in this report adherence to the screen-and-treat protocol was high, and that the majority of VIA positives eligible for cryotherapy were treated, we were unable to retrieve data to confirm this. Another major limitation of our study is the lack of colposcopy and histological confirmation of disease, which was nonexistent, as only recently two pathologists have been appointed in the country. We still estimated the performance of VIA using Pap HSIL as gold standard; in this scenario, it is of concern the very low sensitivity of VIA, 37%, highlighting that there may be a number of women with cervical disease left untreated. Such may be because of challenges of implementing screening such as the low number of health service providers and limited training opportunities.
There is overwhelming evidence that human papillomavirus (HPV) DNA testing is more sensitive than cytology and VIA for the detection of high-grade cervical lesions. 23,24 It has also been demonstrated that a negative HPV test provides longterm protection against cervical neoplasia and stronger reassurance of not developing invasive cancer, and this protection can be translated into fewer screens over life, facilitating adequate screening of vulnerable populations in particular in low-and middle-income countries. Nevertheless, even within HPV-based screening, the major challenges associated with the process of screening mentioned above will remain; in particular, those associated with ensuring that women receive their screening results and are properly evaluated and treated. The difficulties associated with following-up women could be reduced by using HPV testing within screen-and-treat (HPV and cryotherapy) schemes, with or without VIA triage, that should be considered as an alternative to VIA-based screening once a validated HPV test becomes affordable enough to be introduced and scaled up in countries with limited resources.

Conclusion
In conclusion, based on our findings, cervical abnormalities represent a public health problem that needs to be addressed urgently by strengthening, and expanding the existing cervical cancer screening in the country, with a priority among HIV-positive women. Lobbying and advocating for the development of a cervical cancer screening policy, as well as the introduction of HPV vaccine at national level is needed to implement a comprehensive cervical cancer prevention programme. The results found in this retrospective study call for improved health education programmes on cervical cancer screening and potential-associated risk factors such as sexually transmitted infections, in particular HPV, among high-risk groups. It also calls for the evaluation of alternative screening schemes that include HPV testing that can offer further reassurance to screened negatives of not developing cervical cancer for several years. Nonetheless, populationbased screening using low cost VIA, if followed by treatment of VIA positives, is currently the best screening alternative in limited-resourced settings until a validated HPV test becomes affordable.