Cervical cancer is a major contributor to cancer-related morbidity and mortality worldwide. The World Health Organization estimates approximately 530,000 women develop cervical cancer, and 270,000 women die from the disease each year (1). Several epidemiological studies and heath technology assessments (HTAs) around the world have demonstrated the importance of regular cervical screening programs, which lead to early, effective, and cost-effective treatment of precancerous lesions (1). Organized screening programs are critical in identifying the disease before it enters advanced stages, when treatments are often less effective compared with early interventions (Reference Chesson, Blandford, Gift, Tao and Irwin2–Reference Subramanian4). Such programs are an important public health and policy goal, as they have contributed to steadily declining mortality rates in many countries (1).

Population screening for cervical cancer was first introduced in the 1950s using a method called the Papanicolaou (Pap) smear (5), which is also known as conventional cytology. Since its widespread introduction, conventional cytology has been acknowledged for its role in decreasing the global incidence of cervical cancer and associated mortality around the world (Reference Kulasingam, Havrilesky, Ghebre and Myers6). Before the 1950s, cervical cancer was a leading cause of death in the United States among women of childbearing age. Between 1955 and 1992, the incidence of cervical cancer in the United States declined by nearly 60 percent (5). Despite the profound impact of conventional cytology on cervical cancer incidence, uncertainty around test sensitivity and the occurrence of false-positive diagnoses prompted ongoing development of new technologies. Now, screening women for cervical cancer can be accomplished using various technologies and methodologies including, but not limited to, visual inspection using acetic acid (VIA), conventional and liquid-based cytology (LBC), molecular tests involving HPV genotype detection, colposcopy, biomarker detection, electrical impedance probes, and computer-assisted systems. Table 1 includes a description of each technology included in this review.

Table 1. Definitions of Cervical Cancer Screening Technologies Included in This Review

¹Mayo Clinic. Tests and Procedures: Pap Smear. http://www.mayoclinic.org/tests-procedures/pap-smear/basics/what-you-can-expect/prc-20013038.

²National Cancer Institute (NCI) Dictionary of Cancer Terms. http://www.cancer.gov/dictionary.

³ Wade R, Spackman E, Corbett M, Walker S, Light K, et al. Adjunctive colposcopy technologies for examination of the uterine cervix–DySIS, LuViva Advanced Cervical Scan and Niris Imaging System: a systematic review and economic evaluation. NIHR Health Technol Assess. 2013.

⁴ Kisser A, Zechmeister-Koss I. P16/Ki-67 dual stain in the triage of PAPIII/IIID cytology in cervical cancer screening. Vienna: Ludwig Boltzmann Institut fuer Health Technology Assessment (LBIHTA)., 2013; HTA-Projektbericht 72.

⁵ Adelaide Health Technology Assessment (AHTA). Zilico hand-held device for rapid cervical cancer detection. Australia and New Zealand Horizon Scanning Network (ANZHSN). 2010.

⁶ Willis BH, Barton P, Pearmain P, Bryan S, Hyde C. Cervical screening programmes: can automation help? Evidence from systematic reviews, an economic analysis and a simulation modelling exercise applied to the UK. Health Technology Assessment. 2005; 9(13).

⁷ University of Zimbabwe/JHPIEGO Cervical Cancer Project. Visual inspection with acetic acid for cervical-cancer screening: test qualities in a primary-care setting. Lancet. 1999; 353: 9156.

Unlike other cancers, cervical cancer is the first cancer known to be virally induced, as the majority of diagnoses are associated with a select few types of human papillomavirus (HPV) genotypes (Reference Walboomers, Jacobs and Manos7). In addition to organized screening programs, many countries have implemented primary prevention initiatives including HPV vaccination (Reference Hulstaert, Arbyn and Huybrechts8).

As new tests, systems, and algorithms are developed for managing cervical cancer screening programs, policy makers have a responsibility to monitor the body of robust evidence and apply changes to these programs when warranted. It is essential that government endorsed algorithms are both supported by evidence and accepted by the community of women that the programs are intended to serve. In many countries, HTAs are designed to bring evidence to bear on public health policy funding and implementation, informing the decision-making process that surrounds the adoption of new technologies and new clinical algorithms around screening or vaccination. Several national and regional HTA bodies have completed systematic evidence reviews and developed decision models to inform policy makers in implementing cervical screening guidelines.

The aim of this study was to identify all published HTAs in cervical cancer screening from 2000 to 2014 and to report the most common metrics used by HTA agencies when evaluating a new technology in this space. We abstracted information on the population of women included in the reviews, the interventions and comparators that were examined, and how various clinical and economic effectiveness outcomes were measured. Moreover, we indicated if a recommended change in practice resulted. Based on the results of our review, we highlight some of the emerging trends regarding this public health initiative, as well as some of the challenges HTA bodies face during this resource intensive process.

Our study is the first of its kind to review HTAs in cervical cancer screening and to centralize key components of all such reports in one article. Our review can inform users of HTAs, such as policy makers and manufacturers, of the most common metrics used across geographies to assess new cervical cancer screening technologies. Furthermore, our study can inform HTA agencies responsible for carrying out these assessments of some of the current trends in cervical cancer HTA.

RESULTS

The search identified 110 unique HTA reports (see Figure 1). Forty reports were excluded because the primary focus did not pertain to cervical cancer screening and nine reports were excluded because they were a form of peer-reviewed literature rather than a formal HTA report. Thirteen reports were eliminated because they were not accessible or did not offer an adequate summary online. Finally, four reports were excluded because the report and/or a sufficient summary were not available in English.

Figure 1. HTA Report Selection. HTA reports were identified through the Center for Research and Dissemination (CRD), MedLine, and country-specific HTA databases. All searches were limited to years 2000–2014.

Forty-four discrete, full-text or summarized HTA reports with a principal focus on cervical cancer screening met full study criteria and were included for analysis. The included reports came from seventeen countries: United Kingdom (eight); Australia (eight); Germany (five); Spain, Italy, and Canada (three each); Sweden, New Zealand, United States (two each); and Austria, South Korea, Thailand, France, Belgium, Netherlands, Scotland, Denmark (one each).

(P)opulation

All reports carried out their assessment on the female population, only. The majority of reports did not specify an age range for screening, but rather described the intended population as women of screening age. For those that did specify an age range, the range was from 14 years old to 70 years old across all reports. The vast majority of HTAs included in this assessment evaluated new technology for a primary screening population, while six reports evaluated new technology for use as a triage tool for more targeted screening after an initial diagnosis had been made.

(I)ntervention and (C)omparators

Ten cervical screening interventions were evaluated across all forty-four reports: two cytology tests (conventional cytology and LBC), two HPV molecular tests (DNA and mRNA), colposcope, adjunctive colposcopic digital image analysis systems, oncogenic biomarker detection and staining, electrical impedance device, computer-assisted image analysis (CAIA), and visual inspection using acetic acid (VIA).

Of the ten cervical screening technologies evaluated across these forty-four reports, the most common intervention of interest from 2000 to 2014 was the HPV DNA test, with eleven reports evaluating this technology as a primary or triage test. Nine reports evaluated LBC as a primary screening intervention, while eight reports evaluated the introduction of CAIA for use as a primary screening intervention. Rather than evaluating the implementation of one intervention compared with the standard of care, ten reports evaluated a combination of cytology and HPV testing to assess the optimal placement of each technology in a population screening algorithm.

In addition to cytology and HPV DNA testing, there were a few reports that evaluated less common screening interventions. One report from Spain in 2010 compared the use of mRNA testing with that of HPV DNA testing (Reference Varela Lema and Queiro Verdes9). One report from Germany in 2007 evaluated the use of colposcopes in a primary screening population (Reference Nocon, Mittendorf and Roll10), and another report from the United Kingdom in 2013 compared the use of various methods and technologies that work in conjunction with a colposcope (Reference Wade, Spackman and Corbett11). Another report from Austria in 2013 evaluated a new biomarker technology for use as a triage for abnormal cytology (Reference Kisser and Zechmeister-Koss12). Moreover, a 2010 HTA from Australia evaluated the use of an electric impedance probe as an adjunct to primary colposcopy screening (13). Finally, one report from Thailand in 2008 evaluated primary screening with the use of VIA and conventional cytology at various ages and screening intervals (14). All interventions and comparators can be found in Figure 2.

Figure 2. Interventions and their respective comparators from 44 included HTAs. The interventions and their sub-categories are in rectangles, while the comparators are in ovals.

(O)utcomes

Clinical Effectiveness

Forty-three of the forty-four HTAs evaluated diagnostic accuracy using test sensitivity and specificity as the primary measure of clinical effectiveness. One HTA did not specifically assess diagnostic accuracy because the focus of the report was on productivity related to CAIA (Reference Della Palma, Moresco and Giorgi Rossi15). Of the reports that did assess the diagnostic accuracy of a technology, an evaluation was made between the main intervention(s) and comparator(s).

Twenty-nine reports compared sensitivity and specificity to histological outcomes (Reference Hulstaert, Arbyn and Huybrechts8;Reference Wade, Spackman and Corbett11–14;16–Reference Queiro Verdes and Puñal Riobóo36), 76 percent (twenty-two of twenty-nine) of which assessed cervical intraepithelial neoplasia CIN2+ or CIN2/3+ as the endpoint of interest (Reference Hulstaert, Arbyn and Huybrechts8;Reference Varela Lema and Queiro Verdes9;Reference Wade, Spackman and Corbett11–13;16–18;23;25–27;Reference Kitchener, Blanks and Cubie29–34;Reference Queiro Verdes and Puñal Riobóo36;37). Six reports analyzed sensitivity and specificity to cytological outcomes such as low grade squamous intraepithelial lesions (LSIL) or high grade squamous intraepithelial lesions (HSIL) (Reference Krahn, McLachlin and Pham22;23;Reference Queiro Verdes and Puñal Riobóo36;Reference Broadstock38–Reference Noorani, Brown, Skidmore and Stuart40). There were three reports that compared both histological and cytological outcomes (Reference Krahn, McLachlin and Pham22;23;Reference Queiro Verdes and Puñal Riobóo36).

In addition to diagnostic accuracy, eight reports also examined cervical cancer incidence (Reference Hulstaert, Arbyn and Huybrechts8;14;Reference Karnon, Peters and Platt20;26;Reference Sroczynski, Schnell-Inderst and Mühlberger28;37;39;41), and five reports examined mortality due to cervical cancer as clinical endpoints (Reference Hulstaert, Arbyn and Huybrechts8;26;Reference Sroczynski, Schnell-Inderst and Mühlberger28;37;41). Finally, specimen quality was examined in 45 percent (twenty of forty-four) of all reports (Reference Hulstaert, Arbyn and Huybrechts8;14;19;Reference Karnon, Peters and Platt20;Reference Krahn, McLachlin and Pham22–26;Reference Vesco, Whitlock and Eder30–Reference Ronco, Biggeri and Confortini32;39;Reference Noorani, Brown, Skidmore and Stuart40;Reference Payne, Chilcott and McGoogan42–Reference Ruiz-Aragón, Márquez-Peláez, Carlos-Gil, Romero-Tabares and Beltrán-Calvo47).

Economic Effectiveness

In total, thrity-six of forty-four HTAs carried out a health economic analysis. A primary cost-effectiveness analysis was the most commonly used health economic model, with twenty HTAs carrying out this type of analysis (Reference Wade, Spackman and Corbett11;14–16;Reference Varela Lema and Queiro Verdes9;Reference Karnon, Peters and Platt20;Reference Krahn, McLachlin and Pham22;Reference Kitchener, Almonte and Gilham24–26;Reference Sroczynski, Schnell-Inderst and Mühlberger28;Reference Kitchener, Blanks and Cubie29;31;Reference Kitchener, Canfell and Gilham35;39;Reference Payne, Chilcott and McGoogan42;Reference Lapointe, Erickson and Brophy46–Reference Ko, Kim and Hong49). Nine HTAs conducted a secondary cost-effectiveness analysis (Reference Nocon, Mittendorf and Roll10;34;Reference Queiro Verdes and Puñal Riobóo36;Reference Broadstock38;Reference Noorani, Brown, Skidmore and Stuart40;41;43;Reference Latapy, Barre and Midy50;Reference Mittendorf, Nocon and Roll51), seven HTAs developed a budget impact model (Reference Hulstaert, Arbyn and Huybrechts8;13;23;27;Reference Ronco, Biggeri and Confortini32;Reference Ronco, Biggeri and Confortini33;45), and three HTAs carried out both a cost-effectiveness and budget impact analysis (Reference Krahn, McLachlin and Pham22;Reference Lapointe, Erickson and Brophy46;Reference Willis, Barton, Pearmain, Bryan and Hyde48). A major factor that contributed to the economic effectiveness of a new technology was related to the increased productivity it would bring to the laboratory, a metric which was evaluated in nine reports (Reference Della Palma, Moresco and Giorgi Rossi15;19;Reference Karnon, Peters and Platt20;25;Reference Kitchener, Blanks and Cubie29;Reference Ronco, Biggeri and Confortini32;Reference Queiro Verdes and Puñal Riobóo36;Reference Payne, Chilcott and McGoogan42;43). Highlighted economic metrics are summarized in Figure 3.

Figure 3. Common metrics assessed by each cervical cancer HTA and whether a recommended change in practice was made.

Other Key Outcomes Metrics

As new technologies have emerged over the years, the complexity of evaluating new interventions grew in scope. Aside from clinical and economic outcomes, common metrics we extracted from the forty-four reports include the evaluation of certain technologies as an adjunct to other tests, screening interval impact due to new technology, age specific interventions, and the impact on patient reported outcomes.

In total, there were twenty-eight reports which examined adjunct testing for use with the main intervention, either as a triage or co-test (Reference Hulstaert, Arbyn and Huybrechts8;Reference Varela Lema and Queiro Verdes9;Reference Wade, Spackman and Corbett11–14;16;18;Reference Krahn, McLachlin and Pham22;Reference Kitchener, Almonte and Gilham24;26–Reference Sroczynski, Schnell-Inderst and Mühlberger28;Reference Vesco, Whitlock and Eder30–Reference Kitchener, Canfell and Gilham35;37;Reference Noorani, Brown, Skidmore and Stuart40;43;45;Reference Lapointe, Erickson and Brophy46;Reference Ko, Kim and Hong49–Reference Mittendorf, Nocon and Roll51). Of the forty-four HTA reports included in this review, 50 percent (twenty-two of forty-four) examined modifying testing intervals for various screening technologies to find optimal screening algorithms (Reference Hulstaert, Arbyn and Huybrechts8;14;16;19;Reference Karnon, Peters and Platt20;Reference Krahn, McLachlin and Pham22;Reference Kitchener, Almonte and Gilham24;26;Reference Sroczynski, Schnell-Inderst and Mühlberger28;Reference Vesco, Whitlock and Eder30;31;Reference Ronco, Biggeri and Confortini33–Reference Kitchener, Canfell and Gilham35;37–Reference Payne, Chilcott and McGoogan42;Reference Ko, Kim and Hong49–Reference Mittendorf, Nocon and Roll51). Furthermore, rather than assessing one screening program for all women of screening age, seventeen HTAs considered a patient's age group in making specific screening recommendations (Reference Hulstaert, Arbyn and Huybrechts8;13;14;Reference Krahn, McLachlin and Pham22;Reference Kitchener, Almonte and Gilham24;26–Reference Sroczynski, Schnell-Inderst and Mühlberger28;Reference Vesco, Whitlock and Eder30;31;Reference Ronco, Biggeri and Confortini33;34;41;45;Reference Ko, Kim and Hong49–Reference Mittendorf, Nocon and Roll51). Twelve reports also considered the impact of introducing a new test on a patient's stress and anxiety through Patient Reported Outcomes (Reference Hulstaert, Arbyn and Huybrechts8;13;18;Reference Kitchener, Almonte and Gilham24;26;Reference Sroczynski, Schnell-Inderst and Mühlberger28;Reference Ronco, Biggeri and Confortini33;Reference Kitchener, Canfell and Gilham35;37;Reference Broadstock38;Reference Latapy, Barre and Midy50;Reference Mittendorf, Nocon and Roll51), a common metric used in HTAs evaluating HPV tests. A summary of these important outcome metrics can be found in Table 2.

Was a Change in Screening Practice Recommended?

Based on our analysis of the forty-four included reports, fifteen HTAs from five countries recommended a change in practice for their cervical cancer screening program. It is important to note that not all HTAs seek to formally make recommendations, and some exist to simply summarize all existing evidence in data. Four reports had both a recommendation for a broader population and a subset of the population, and thus fell into the categories of “Recommend” and “Optimized.” Six reports were “Recommend” only, and five reports were “Optimized” only.

“In research only” was a finding of one report from the year 2000, which concluded that HPV testing should only be used within the framework of clinical trials. Finally, twenty-nine HTA reports fell into the “Not Recommend” category. Reasons for this conclusion could include insufficient evidence, poor clinical/economic effectiveness, or that the objective of the HTA was to simply summarize data rather than formally recommend a technology and/or change in practice. Figure 3 details the findings of each included HTA.