Setting

The study was conducted at the Regional Hospital of Nkongsamba, Moungo Division of the Littoral Region of Cameroon. It is a second level reference public health facility with a catchment area of over 321,295 inhabitants [8]. The onsite HIV clinic was established in 2005 and offers HIV counselling and testing, ART and limited community outreach services to patients on ART. Screening for cervical disease in WLWHIV is not common practice.

Ethical consideration

The study was approved by the medical council of the Nkongsamba Regional Hospital. Permission to use data was granted by the directorate of the Nkongsamba Regional Hospital, the custodian of the database. Individual participant consent could not be obtained because we were using patients’ records for this study. All patient information was anonymised and de-identified prior to analysis.

Population and enrolment

Between February 26 and May 7, 2014 Nkongsamba Regional Hospital in Cameroon, held a voluntary cervical cancer screening campaign in honour of the International Women’s Day, an extension of its regular cervical cancer screening program. Free cervical disease screening was provided by a Peace Corps Cameroon Volunteer Activities Support and Training (VAST) grant. This voluntary screening was open to all women in the municipality. HIV positive women receiving treatment at the HIV clinic of Nkongsamba Regional Hospital as well as women from the community with unknown HIV status were eligible to participate in the screening if they were over the age of 35, not pregnant, and not severely ill. Women who were menstruating at enrolment were asked to return in one week to participate. Health workers screened potential eligible women and completed a brief intake questionnaire after verbally obtaining their consent. They collected information on socio-demographic characteristics, HIV and HAART statuses, lifestyle factors, reproductive characteristics, and family history of cancer (S1 File).

Procedures

Each woman was screened using three different methods in series: visual inspection with acetic acid (VIA), visual inspection with Lugol’s iodine (VILI) and conventional Pap smear cytology. During a pelvic examination, visual appearance of the cervix was recorded before application of acetic acid or Lugol’s iodine at The Nkongsamba Regional Hospital. VIA was performed by applying 3–5% acetic acid solution to the cervix and waiting for three minutes. VIA was considered positive with the presence of well-defined acetowhite areas with sharp borders. After VIA screening VILI was performed by applying Lugol’s iodine solution to the cervix. Similarly, results were interpreted by a laboratory technician after waiting for another three minutes. VILI was considered positive with the presence of areas of non-uptake, which appear pale yellowish-white. VIA and VILI results were interpreted by the same laboratory technician. Following VIA and VILI screening a traditional Pap smear was performed with cervical cells collected and smeared onto slides, which were read and analysed by a pathologist at The University Teaching Hospital in Yaounde, Cameroon. The pathologist was blinded to results from the direct visual inspection (DVI). Women were referred to immediate biopsy if they had high-grade squamous intra-epithelial lesions (HSIL), atypical squamous cells cannot exclude high-grade lesion (ASC-H), or atypical glandular cells of undetermined significance (AGU-S). Women with low-grade squamous intra-epithelial lesions (LSIL) and atypical squamous cells of undetermined significance (ASCUS) were referred to repeat Pap smear in six months.

Statistical analysis

Data analyses were performed using Stata^® 13.1 (StataCorp LP, Texas 77845, USA). A comparison of characteristics between WLWHIV on HAART and women from the general community was done. The main outcome of interest was cervical disease defined as the presence or absence of squamous intra-epithelial lesions (SIL) of any grade according to Pap smear cytology. The impact of HAART on human papilloma virus-related disease remains more clearly observable when evaluating SIL of any grade as the primary outcome[9]. Cervical cytology has been chosen by convention to define cervical lesion and was performed at a University Teaching Hospital, a first level of reference health facility. The characteristics compared were age, place of residence, occupation, religion, level of education, marital status, smoking habit, past pregnancy and childbearing history, and family history of cancer. Summary statistics were presented as proportions for categorical variables and as means (with standard deviations) for normal continuous variables or medians (with IQR—interquartile range) for skewed continuous variables. Pearson chi-squared test or Fisher exact tests for small samples were used where appropriate to assess for the association between categorical variables. The student t-test was used to test for the mean difference between two groups for continuous variables. A univariable logistic regression model was set up for the association between each exposure variable with cervical disease. Crude odd ratios, their 95% confidence intervals and p-values were reported. Factors associated with cervical disease at 5% significance level, were included in a multiple logistic regression model. Backwards elimination based on p-value was used to pick up factors independently associated with cervical lesion. The corresponding adjusted odds ratios, their 95% confidence intervals and p-values in the final model were reported.

Integrating cancer screening program with HIV care

Although access to HAART has been rolled out in the recent decade, the integration of cervical cancer screening with HIV care is rare or inexistent in most low and middle income countries as a result of persistent lack of resources and competing priorities in ensuring a sustainable and comprehensive HIV care. By adopting and successfully implementing for three successive years, the “see-and-treat-or-refer” approach for cervical cancer control, a level two referral health facility in a predominantly rural Cameroonian setting, has demonstrated that, it is feasible to nest a cancer control program into healthcare in general and HIV care in particular. The health facility provided the resources needed to implement the program with the low cost VIA/VILI technology but its extension to the more expensive cervical cytology was made possible by a Peace Corps Cameroon Volunteer Activities Support and Training (VAST) grant. This is an illustration of a pragmatic public-private partnership initiated at the operational level of a health system in which health programs are more vertical than horizontal in practice. Health facilities and HIV programs could be encouraged to take similar initiatives by making use of their own resources but their efforts need to be supplemented by support from other sources. HIV programs in LMICs are currently gaining substantial support from partners like the Global Fund and PEPFAR (President’s Emergency Plan for AIDS Relief) to help scale up the screening program [10].

While it might be feasible to broaden an HIV program to include cancer screening, such a screening would be of less relevance and unsuccessful according to the principles of Wilson & Jungner [11], if post-screening follow-up, treatment or referral systems are not effective. Women in HIV care can be traced and followed up while receiving free HAART but they cannot be provided treatment for cervical lesions in HIV clinics. They need to seek care in tertiary health facilities by paying out of their pockets. To date, we have little or no idea about the outcome of women who were referred for confirmatory diagnostic investigations and subsequent treatment. Women with a ‘false positive’ results after screening may suffer from psychological disturbances if they cannot have a confirmatory test done. Similarly, those who require treatment but cannot afford would consider screening as a ‘necessary evil’. These barriers to a successful screening program and thus its linkage to HIV care are likely to annihilate achievements made so far and thus discourage further efforts.

The screening program employed three different methods including VIA, VILI and the conventional Pap smear. While VIA and VILI techniques showed good agreement in detecting cervical abnormalities, there was considerable discrepancy between the DVI techniques and Pap smear. The DVI procedures were performed in the same centre and by the team of nurses who also collected, smeared, stored and transported cell samples for cytology. The cytology results were read and interpreted by a trained pathologist at a tertiary hospital. The pathologist was blinded to DVI results. None of these procedures or interpretation of results were double-checked to ensure quality assurance (QA). This practice is a reflection of the routine clinical procedure in the health facility. Lack of QA may be responsible for the observed disagreement between DVI techniques and Pap smear. The impact of the discrepancy on individual patients and the screening program is negligible because follow up visits and confirmatory diagnostic tests can correct for false results. However, for the purpose of research, inadequate QA may introduce systematic errors in our findings. To reduce this bias, we used SIL of any grade as our primary outcome[9, 12] and because the cervical cytology was performed in a first level of referral hospital in which we have a higher degree of trust.

HAART and evolution of cervical disease prevalence in Cameroon

The 11.5% prevalence of SIL in women receiving HAART in this study was much lower than the 48.6% and 43.5% prevalence rates observed in 2009 and 2008 respectively in women initiating HAART in similar settings of urban Cameroon [7, 13]. This is an indication that cervical disease prevalence has decreased over time. Women in the earlier cohorts had low median CD4 counts of 253.7 cells/μl and 197cells/μl respectively while women in our cohort were HAART-experienced and supposedly had a median CD4 of about 436 cells/μl [14].This relatively low prevalence can thus be associated with an increased regression and decreased incidence of human papillomavirus (HPV)-related cervical lesions among HIV-infected women on HAART [12, 15] owing to improved levels of immune status[16]; and to the improved access to HAART and retention in HIV care in Cameroon during this period of time [17, 18]. Our findings may thus be tilting in favour of reduced prevalence of cervical disease in the era of HAART scale up despite increased survival. We could not however, account for differences in methods and quality assurance in the ascertainment of cervical lesions as well as other risk factors for cervical cancer that must have changed over time. As a result, it is difficult to attribute the observed decrease in prevalence of SIL to HAART alone.

HAART and risk of cervical disease

It is known that women infected with HIV have a significantly elevated risk of cervical cancer and its precursors when compared with the general population[19, 20]. We speculated that HAART by partially restoring the immune status will reduce the risk of cervical disease relative to HAART-naïve HIV-infected women, but by prolonging life HAART will also increase the risk as a result of ageing and prolonged exposure to oncogenic factors. The overall risk was expected to remain higher or at most similar to women in the general population. Current evidence about the effect of HAART on cervical cancer risk is mixed and had compared HAART and non-HAART groups[5]. Our findings suggest that the risk of cervical disease was much lower in women receiving HAART compared with women in the general population even after controlling for known differences between the two groups. Though the findings may be internally valid for HIV-infected women, there is no reason at this moment why we should think that HAART per se is protective against cervical disease in the general population even though HAART may provide some protection against HPV [21–23], the main risk factor for cervical cancer, it has never been tested for this purpose. Yet we want to imagine that by assuming that both groups had similar risk of cervical disease as “healthy volunteers” some unmeasured but predictable differences between the groups may have also been responsible for the low risk of SIL in the HAART group. A supposedly change in risky lifestyles such as cessation of smoking, moderation of alcohol intake, restriction of sexual partners and safer sex practices adopted more by women in HAART[24, 25] as they are taught during their regular visits to the HIV clinic for continuous care, may have reduced their risk of developing cervical disease. This study therefore indicates that HAART is directly or indirectly via comprehensive HIV care associated with a reduction in the odds of cervical cancer. However, we want to warn against the threat of relaxing its cancer screening in women on HAART because as they live longer and especially after the age of 55 years, the risk triples.

Our study had some limitations. The Pap smears lacked quality assurance but there is no indication that there was a differential interpretation of results between the HAART group and the community group because the pathologist was blinded to the participants’ information. The observational design cannot provide a causal relationship between HAART and cervical disease but has provided an additional argument in favour of a protective effect of HAART in HIV-infected women. The use of hormonal contraception, the number of sexual partners and HPV vaccination has not been accounted for.