Study design and study population

This study was conducted in Tanzania, which is a low-income country located in Eastern Africa with a population of 56 million people.10 Women were enrolled from three existing cervical cancer screening clinics located in urban and semirural areas: (1) ORCI in Dar-es-Salaam as well as (2) KCMC and (3) Mawenzi regional referral hospital in the Kilimanjaro region. ORCI is a national cancer hospital that provides clinical care and treatment for all patients with cancer in the country. Additionally, they conduct cervical cancer screening three times a week for the general population. KCMC is a northern zonal tertiary facility that provides cervical screening three times a week for the general population, and Mawenzi Hospital is a regional hospital that provides cervical cancer screening two times a week. In Dar-es-Salaam, women from Ilala, Temeke and Mwananyamala district were included, while in the Kilimanjaro region, women originating from the urban and rural district of Moshi—including Hai and Rombo—were included. Originally, the study was designed as a double-site study (KCMC/ORCI), however, due to a slower-than-anticipated recruitment rate, a third study site (Mawenzi) was added 6 months into the enrolment period. Women were eligible for inclusion if they were 25–60 years and attended a patient-initiated routine cervical cancer screening at one of the study sites. Women were excluded if they were pregnant, on their menstrual period, had a history of premalignant lesions of the cervix within the last 12 months, had previously been diagnosed with cervical cancer or had undergone abdominal hysterectomy. Women on their menstrual period were encouraged to return once their menstrual period was over. Following a detailed explanation of the study, all participants provided written informed consent. Fingerprints were used for illiterate participants. HIV positive women were oversampled from care and treatment clinics (CTC) at the study sites from where they were referred to the screening clinics. The total number of women and HIV positives required for the study was found through a power calculation based on McNemar’s test comparing two diagnostic tests (S1: standard test (VIA) vs S2: new test (careHPV) with cervical cytology as the reference test and the threshold being HSIL+ (HSIL, carcinoma in situ and carcinoma). The power calculation was based on the research group’s previous study in Tanzania.22 It was estimated that 180–200 women would have precancerous lesions at baseline (~true positives), and assuming a significance level of 5%, 80% power and a sensitivity of VIA of 30%, it would be possible to detect a significant difference if the sensitivity of the new test would be at least 44%. It was anticipated that careHPV testing would have a higher sensitivity than VIA.

Data were collected during the enrolment visit, at 14 months (first follow-up) and at 28 months (second follow-up, ongoing). As there is no predefined optimal duration of time to investigate the natural history of HPV, the length of follow-up was based on a number of factors, including the recommended duration of time between cervical cancer screenings for HIV positive women (12 months),34 available resources, risk for developing cervical lesions and limiting the workload at the screening clinics by minimising overlaps between enrolment and follow-up visits. Healthcare providers working at the screening clinics enrolled participants and collected data following protocols developed specifically for the project. At inclusion, all participants were given a 14-month follow-up appointment written on an appointment card. If the women did not attend their follow-up visit within 1 month of their appointment, an active follow-up procedure was initiated. First, non-attendees were contacted by phone and encouraged to attend (tracing method I). If the woman did not show up within 2 weeks of the phone call, the women had a home-visit by an outreach nurse who encouraged her to attend (tracing method II). If the woman still did not attend and consented to it, an outreach nurse visited her again and conducted the follow-up visit at home (tracing method III). Transportation costs were compensated for those women who were reminded to come. Women who participated in the first follow-up were appointed a second follow-up visit 28 months after initial recruitment. If the woman did not attend the second follow-up, tracing method I and III were reinitiated.

Assessment of exposure

At inclusion, sociodemographic, lifestyle, reproductive and sexual characteristics were collected during a personal interview using a modified version of a standardised questionnaire adopted from a previous study conducted in Tanzania35 (online supplementary appendix 3). The questionnaire was hard copy, developed in English while the interview was conducted in Kiswahili with direct translation. A Kiswahili version of the questionnaire was available to guide the interviewers. A detailed contact information form was filled at enrolment and updated at follow-up. Participants were screened by trained healthcare providers according to the standard national cervical cancer screening prevention programme in Tanzania.36 This entails a cost-free gynaecological examination with VIA, HIV testing and HIV counselling. Venous blood from the index finger was tested by using a quick HIV-1/2 test (www.alere.com), and a supplementary quick HIV-1/2 test (Abbott Laboratories) was performed as confirmatory for positive results. Discordant results were further confirmed using Unigold (Trinity Biotech). Women who screened VIA positive or were suspected of manifest cancer were managed on-site based on the National Cervical Cancer Service Delivery Guidelines. This included treatment with cryotherapy or loop electrosurgical excision procedure for VIA positives, and punch biopsy and referral for treatment at the oncology clinic at ORCI for cancer suspicions.36 Further, weight and height were measured and registered on a hard-copy registration sheet together with the HIV and VIA result (table 1).

Prior to the routine VIA examination, cervical swabs were taken using (1) an Aryes spatula for careHPV test (www.qiagen.com), and another specimen was taken using (2) a ThinPrep Pap test plastic spatula for liquid-based cytology, HPV DNA testing and genotyping by use of HC2 and LiPaExtra (Innogenetics, Gent, Belgium). The cervical samples for careHPV analysis were kept in a careHPV collection medium and stored at room temperature (max 25°C) in laboratories at ORCI or KCMC. Once 90 samples had been collected, they were analysed for HR HPV using a careHPV machine. A test was considered positive if one or more of the following 14 HR HPV types were detected: HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68. The results were registered on a careHPV results sheet (table 1).

The samples for HC2 testing, genotyping and cytology were kept in a PreServCyt solution (Hologic, 250 Campus Drive, Marlborough, MA 01752, USA) and stored at room temperature in laboratories at ORCI and KCMC. Once enrolment had finished, the PreServCyt vials were sent to the Department of Pathology at Lillebaelt Hospital, Vejle, Denmark and processed at the ThinPrep5000 Autoloader Instrument, Hologic, according to manufacturer’s instruction and stained with ThinPrep Stain. The slides were scanned by the Thin Prep Imaging System with selection of 22 fields of view which was reviewed by cytotechnologist in review scopes. The specimens were evaluated for adequacy and for abnormal cells. If abnormal cells were detected, the slides were consulted with a gynae-pathologist who made the final diagnosis. The specimens were diagnosed according to the Bethesda Nomenclature for System for Cervical Cytology 201437 into following categories: negative for intraepithelial lesion (NILM), atypical squamous cell of undetermined significance (ASCUS), atypical squamous cell in which high-grade squamous intraepithelial lesion cannot be excluded (ASCH), low-grade squamous intraepithelial lesion (LSIL), high-grade intraepithelial lesion (HSIL), atypical glandular cell (AGC), adenocarcinoma in situ (AIS) and adenocarcinoma. The remaining material of the PreServCyt vials were sent to the Section for Experimental Virology, Tubingen University, Germany, for HPV DNA testing and genotyping. HPV DNA testing was done using HC2 DNA test (www.qiagen.com) with a HR cocktail probe. A test was considered positive if one or more of the following 14 HR HPV types were found: HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68. A threshold of 1.0 pg HPV DNA/ml, which corresponds to 1.0 relative light unit coefficient, was used, as recommended by the US Food and Drug Authority. HPV positive samples were genotyped using LiPaExtra, which can detect 28 HPV types, 18 HR risk types (HPV 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 73, 82) and 10 low risk types (HPV 6, 11, 40, 43, 44, 54, 69, 70, 71, 74).38

Outcome measures

Women who attended their follow-up appointment at the screening clinics were HIV tested (if negative at baseline), had a cervical swab collected with a ThinPrep Pap test plastic spatula—for HPV DNA testing using HC2 and genotyping using LiPaExtra—and underwent VIA (table 1). Further, sexual and reproductive characteristics were updated using a structured questionnaire (online supplementary appendix 4). Women who did not attend their follow-up appointment at the clinic but consented to having a home-visit appointment (cf. tracing method III) responded to the questionnaire and had cervical specimens collected by use of an Evalyn self-sampling/self-swab device (www.roversmedicaldevices.com). The samples were transferred to laboratories at ORCI and KCMC where they were kept in a PreServCyt solution and stored at room temperature.

Data management

Questionnaires, registrations forms, contact forms and careHPV result sheets were stored in different cabinets in locked offices at KCMC and ORCI, after which they were double entered into EpiData by data clerks. Together with laboratory results, these data were sent to the Research Unit for Gynaecology & Obstetrics, Odense University Hospital (OUH), Odense, Denmark, where they were merged and constructed into a baseline database. Data with invalid identification numbers (IDs) or no data registered were excluded on creation of the database. Similar to baseline, follow-up data were sent to OUH, after which it was merged with the baseline database. Follow-up IDs that could not match a baseline ID were excluded.

Patient and public involvement

Study participants were not involved in the design or recruitment of the study. In order to increase public awareness, government and religious leaders were informed about the project, the latter through mosques and churches. When the study finishes, the results and their potential implication to the public will be communicated through meetings with health authorities, policy briefings and announcements in the mainstream media.

Baseline findings

A total of 4080 women were enrolled into the CONCEPT study. After inclusion, 37 study participants were excluded, leaving the total cohort to consist of 4043 women (figure 1). The baseline sociodemographic distribution is presented in table 2. A total of 718 women (17.8%) were HIV positive and the majority of these were 35–39 years old (22.9%) while the majority of HIV negative women were 40–44 years old (19.2%). Among the HIV positives, 49.7% were married (n=356) while 73.6% of HIV negatives were married (n=2434). Most of both HIV positive (70.4%; n=504) and HIV negative women (64.1%;n=2127) had completed primary school. More HIV negative women (87.0%; n=2879) reported to have had sex within the last 12 months compared with HIV positive women (72.7%; n=520), while more HIV positives (26.2%) than HIV negatives (9.8%) reported to have had used condoms at every intercourse. In relation to number of lifetime partners, 71.9% of HIV positives reported having more than one lifetime partner while the corresponding number for HIV negative women was 61.0%. The CD4 count for HIV positive women were as follows: 8.6% (n=62) reported having a CD4 count <199; 30.5% (n=219) had a CD4 ranging from 200 to 499; and 48.9% (n=347) had a CD4 count >500. Further, 12.5% (n=90) of the HIV positives did not report the CD4 count.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flowchart of enrolment and follow-up of CONCEPT cohort.

Among the 4043 participants, the cervical sample was insufficient for HPV analysis for 396 women (9.8%) at baseline, leaving 3667 women eligible for analysis of HPV positivity. Further, 27 women (0.5%) did not have a sample for cervical cytology, leaving 4016 women available for cytological analysis of cervical lesions. All 4043 women were either tested for HIV or had previously tested HIV positive. At baseline, 696 out of 4043 women (17.2%) tested HR HPV positive by HC2. The cytology showed that overall 3.4% (n=139/4043) had HSIL+, while 8.1% (n=329/4043) of women had LSIL. A total 3416 women had both HPV- and cytology results. Among this subgroup of women, 18.9% were HPV-positive (n=644/3416), and the four most common HR HPV types were HPV 52 (3.8%), HPV 16 (3.6%), HPV 58 (2.5%) and HPV 18 (2.4%). Among HIV positive women, 33.7% were HR HPV positive, while the corresponding figure among HIV-negative women was 15.6%. Among women with high-grade lesions (HSIL+), 32.5% had HPV 16, 19.3% had HPV 58, 17.5% had HPV 31, 16.7% had HPV 18 and 16.7% had HPV 52. A full description of the HPV distributions according to HIV status and cytology results is published elsewhere.33

First follow-up findings

A total of 3805 women (94%) were eligible for first follow-up—238 women (6%) were ineligible due to becoming pregnant, having moved, having had a hysterectomy or having become sick or died (figure 1). Of the 3805 women, 3074 women (81%) attended the first follow-up visit approximately 14th months after enrolment. However, among the attendees, only 500 (16%) attended within 30 days of their scheduled appointment date and without being traced for follow-up. A total of 1088 (35%) attended the clinic after a phone call reminder (tracing method I), 62 women (2%) attended the clinic after a nurse home-visit (tracing method II), while 1253 women (41%) were followed up at home and had specimens collected using self-sampling device (tracing method III). A total of 731 women (19%) were lost to follow-up (figure 1).

The women who participated in the first follow-up were very similar to those who did not attend when looking at sociodemographic factors (table 2). However, overall, fewer women were HR HPV positive at follow-up compared with baseline (14.8% vs 17.2%), and fewer HIV positive women were HR HPV positive at follow-up compared with baseline (24.1% vs 31.6%) (table 3).