Cost and cost-effectiveness of cervical cancer HPV DNA self-sampling: a systematic review

Abstract

Objective

To gather evidence on the costs and cost-effectiveness of human papillomavirus DNA (HPV DNA) self-sampling worldwide and to determine whether these are influenced by the various HPV DNA self-sampling strategies employed.

Design

A systematic review conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Data sources

Articles were searched from Embase, Medline (Ovid), PubMed and EconLit, published from 2015 to 2025.

Eligibility criteria

Women eligible for cervical cancer screening. Studies on the economic evaluation of HPV DNA self-sampling worldwide. Peer-reviewed original articles published in English from 2015 to October 2025.

Data extraction

Data were extracted using the authors’ developed template. Analysis was done descriptively, applying synthesis without meta-analysis (SWiM).

Results

33 studies were included in 11 low- and middle-income countries (LMICs) and 9 high-income countries (HICs). Most studies (N=18) targeted women from the general population and 10 under/never-screeners. The HPV DNA self-sampling strategies include: opt-in, opt-out, point of care and community health campaigns. Community health campaigns were predominant in LMICs, while opt-in and opt-out in HICs. The cost per woman screened reported in micro-costing/cost-consequence analysis studies ranged from US$31.77 to US$72.20 in LMICs and from US$56.21 to US$78.34 in HICs. 21/23 studies found HPV DNA self-sampling strategies to be cost-effective/cost-saving: 11/12 from LMICs and 10/11 from HICs.

Conclusions

Community health campaigns and point-of-care HPV DNA self-sampling strategies in LMICs and opt-in and opt-out in HICs are cost-effective. Implementation studies are recommended, especially for populations at high risk.

PROSPERO registration number

CRD42024518867.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Human papillomavirus DNA (HPV DNA) self-sampling may be cost-effective in high-income countries (HICs), especially in Western Europe. However, there was a lack of evidence for low- and middle-income countries (LMICs) and some of the HPV DNA strategies, such as HPV self-sampling collected at the point of care.

WHAT THIS STUDY ADDS

• HPV DNA self-sampling strategies are cost-effective in both LMICs—11/12 studies (community health campaigns 4/4, point of care 2/3, combination of point of care and mailed 1/1; unspecified 4/4). 10/11 among studies that declared cost effectiveness in HICs with (3/3 opt in, 4/4 opt out (not mutually exclusive) and 5/6 unspecified). The budget for HPV DNA self-sampling varies across countries.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Future implementation studies are recommended for HPV DNA self-sampling strategies, such as community health campaigns in HICs and point of care, especially opportunistic HPV DNA self-sampling. Economic evaluation studies targeting a subgroup of high-risk population are encouraged.

• Implementation studies on these approaches, including subgroup preferences, implementation barriers are recommended when a country wants to introduce this screening strategy, especially in areas with low screening uptake and high burden of cervical cancer, such as sub-Saharan Africa.

Background

Cervical cancer is ranked the fourth most frequent cancer affecting women in the world. Globally, new cases were estimated to be 604 127 annually, causing 341 831 deaths in 2020.¹ This incidence is estimated to increase to 700 000 and deaths to 400 000 in 2030 annually.² Incidence and deaths from cervical cancer are higher in low- and middle-income countries (LMICs) due to a lack of supporting infrastructure in terms of human resources and affordability of technologies to screen for cervical cancer such as Papanicolaou smear (pap smear).

In November 2020, the WHO launched a global strategy to speed up the reduction and elimination of cervical cancer.² The strategy has three overarching targets for 2030: 90% of girls are vaccinated against the human papillomavirus (HPV) by the age of 15; 70% of women are screened two times in their lifetime with high-performance tests by the age of 35 and again at the age of 45 years; and 90% of those with a positive screening result or those with cervical lesions are treated including palliative care when necessary.² In the subsequent year, WHO published a new guideline that proposed the use of HPV DNA as the primary screening test. Countries using visual inspection with acetic acid (VIA) or pap smear as primary screening tests are encouraged to transition to HPV DNA tests.³ VIA and pap smears have been used as primary tests in LMICs and HICs, respectively.

Cervical cancer is preventable and treatable if it is detected early.⁴ It takes more than 10‒20 years for HPV to develop into cancer for women with strong immunity and 5‒10 years for women with weak immunity like those living with HIV.³ There are proven vaccinations that have been included in population-wide vaccination programmes for girls between 9 and 15 years of age.²

Current cervical cancer screening methods include nucleic acid amplification tests comprising high-risk HPV DNA tests and mRNA under the molecular category; pap smear, liquid-based cytology (LBC) and dual staining under cytology and VIA or with visual inspection with Lugol’s iodine (VILI) under visual inspection.³

These methods differ in sensitivity and specificity. HPV DNA testing has higher sensitivity in detecting cervical intraepithelial neoplasia grade two or higher (CIN2+) than cytological and visual inspection tests.^{5 6} Due to this and the new WHO guideline on screening and treatment of cervical cancer, several cytology-based primary screenings are being transformed into HPV DNA testing.⁷

HPV DNA tests can be done by a clinician or by self-sampling. HPV DNA self-sampling reduces discomfort, provides privacy, is convenient and causes less embarrassment.8,10 It can also be done through various strategies including outreach and point of care. Studies on the clinical performance of HPV DNA self-sampling compared with clinician-collected samples have shown effective performance in terms of sensitivity/specificity, increased uptake and patients’ acceptability in both LMICs and HICs.11,13 Because of this, several HICs, including the UK, France, Australia and Sweden, have adopted/piloted HPV DNA self-sampling for the general population or non-screeners or under-screeners.^{14 15} It has also been piloted in several LMICs and included in national guidelines and programmes like Tanzania, Kenya, Ethiopia, India, Brazil, Malaysia, Peru and Myanmar.^{15 16}

A previous systematic review on the cost-effectiveness of HPV DNA self-sampling, published in 2020, included studies on HPV DNA self-sampling outreach strategies up to 2019.¹⁷ Since then however, changes have occurred in global screening strategies and guidelines.^{2 3} The current systematic review, therefore, aimed to collect evidence on the cost and cost-effectiveness of HPV DNA self-sampling worldwide and whether they are affected by the types of HPV DNA self-sampling strategies.

Methods

This systematic review was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (onlinesupplemental tables I II). The protocol for this review was registered with PROSPERO, registration ID CRD42024518867.

The primary outcomes of this review are two, depending on the type of economic evaluation:

1. For partial economic evaluation (micro-costing or cost consequence analysis (CCA)): HPV DNA self-sampling cost per woman screened for LMICs and HICs countries and whether this varies by HPV DNA self-sampling strategies.

2. For full economic evaluation studies (cost effectiveness (CEA), cost utility (CUA) and cost benefit analysis (CBA)): cost effectiveness of HPV DNA self-sampling for LMICs and HICs and whether the effectiveness varies by HPV DNA self-sampling strategies.

The secondary outcome of the reviews is the budget impact analysis (BIA) reported by the included studies.

Inclusion and exclusion criteria

Inclusion and exclusion criteria were developed based on Population, Intervention, Comparator, Outcome and Setting (PICOS). The population included all women eligible for cervical cancer screening. The intervention included synonyms for cervical cancer HPV DNA self-sampling. In the current review, we included studies comparing cervical cancer HPV DNA self-sampling with any other cervical cancer screening test, including HPV DNA clinician-collected samples, VIA, pap smear or no screening. We included economic evaluation studies such as CEA, CUA, CCA and micro-costing. Studies were not excluded based on the costing perspectives, outcomes or economic modelling.

We included peer-reviewed original articles published between 2015 and October 2025 in the English language. Articles from the previous systematic review during this timeframe were included (online supplemental table III). This time period was selected since it was a time in which much evidence of HPV DNA self-sampling technologies including accuracy, uptake and acceptability was produced and it was a period in which HPV DNA self-sampling was included in the WHO cervical cancer screening guideline and some countries started integrating the method into their screening programmes.^{3 18 19}

Search strategy, screening and selection of articles

Based on previously published reviews^{17 20 21} and Scottish Intercollegiate Guidelines Network (SIGN) recommendations for researching studies on economic evaluations, a systematic search strategy was formulated. We used the PICOS of our research question to identify keywords and MeSH terminologies which were then combined using Boolean operators. The search strategy was developed by NAT and reviewed by other authors as well as a subject librarian (online supplemental table IV). The strategies were then adapted for various databases. The initial search was carried out between 21 and 23 December 2023, updated on 17 September 2024 and 27 October 2025. We searched for studies on the cost-effectiveness of cervical cancer HPV DNA self-sampling across the globe, using PubMed, Medline (Ovid), Embase (Ovid) and EconLit databases. Reference lists from selected articles for full-article screening were also checked through citation tracking. Studies were exported to Rayyan—a web and mobile app systematic review tool for article screening.²² Duplicate studies were removed, and articles were screened for eligibility through title, abstract and full-text screening. The article screening was done by NAT; EG and NJ screened 20% of the excluded studies and 20% of the fully screened articles.²³ Conflicts were resolved through discussions. We planned to do 100% duplicate screening if the conflicts were significantly high or the agreement was less than 80% for the proportion done in duplicate. All authors had access to the Rayyan software as reviewers.

Data extraction, synthesis and analysis

We developed a data extraction form that included the target population, country, intervention, comparator, outcome measure, time horizon, costing perspective, modelling used, type of costing study, willingness to pay (WTP) threshold, discounting and sensitivity analysis for characteristics of included studies. For the results, we extracted cost reference year, cost per woman screened, cost per CIN2 detected, incremental cost-effectiveness ratio (ICER) and the remark which shows if the intervention was cost-effective. We also extracted the assumption of the clinical characteristics including screening interval, uptake/coverage, sensitivity, specificity, follow-up for HPV-positive cases and adherence to follow-up. The extraction form was pre-piloted and adapted to capture the necessary data. Data extraction was done by NAT and was reviewed by other authors.

Data synthesis was based on ‘synthesis without meta-analysis’ (SWiM) using a descriptive narrative approach.²⁴ This method was chosen since cost and cost-effectiveness studies included do not have common objectives and have heterogeneity in costing designs and methodology. For comparability, all costs were expressed in USD using the World Bank (WB) exchange rate 2023²⁵ and inflated to 2023 cost values using the gross domestic product (GDP) deflator from WB.²⁶ The synthesis of the result is presented separately for LMICs and HICs. The presentation follows types of economic evaluation, starting with partial economic evaluation micro-costing and CCA and the full economic evaluation: CEA/CUA and CBA for each HPV DNA self-sampling strategy and BIA.

Critical appraisal and methodological assessment

We used a quality checklist derived from the HPV-FRAME quality framework for HPV screening models and the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement. The tool is adopted from Malone et al.¹⁷ A single reviewer (NAT) assessed the quality of the included studies. The tool included 13 items on the target population, economic evaluation, strategies compared, time horizon, source of data, currency, sensitivity analysis and discussion of limitations. It was rated on a scale of excellent (including all 13 items), good (including 9‒12 items), fair (including 6‒8 items) and poor (including 0‒5 items).

Patient and public involvement

None.

Results

980 articles were identified from the databases and citation searches (n=440 Embase Ovid, n=289 Medline Ovid, n=246 PubMed, n=1 EcoLit and 4 from citation search). After the removal of duplicates, titles and abstract screening 55 studies remained for full-text screening, out of which 33 articles qualified for inclusion27,59 (figure 1).

Figure 1. Flow chart of the cost and cost-effectiveness studies of human papillomavirus DNA self-sampling.

Study characteristics

The included studies were conducted across 11 LMICs33,3840 and 9 HICs27,3239 48 (figure 2). 26/33 studies carried out CEA/CUA27,3032 33 37 (figure 3). 18/33 studies targeted women from the general population of which 13 were from LMICs.3335,38 40 10/33 studies targeted women who were under/never screeners of which 9/10 were from HICs countries.2830,32 39 50 55 56 59 13/33 studies used a form of mailed HPV DNA self-sampling kits which includes (opt-in, opt-out or both).27,2931 32 39 48 54 55 57 Studies which used mailed strategies were from HICs except one that had a combination of point of care and mailed kits from LMICs.⁴⁰ Nine studies used community health campaigns,34,3638 42 43 45 46 50 eight of these studies were done in LMICs and one in HICs⁵⁰ (figure 4).

Figure 2. Distribution of studies per country. *Include Uganda, India and Nicaragua.

Figure 3. Distribution of studies by economic evaluation design.

Figure 4. Human papillomavirus DNA self-sampling strategies used. HICs, high-income countries; LMICs, low- and middle-income countries.

13/33 studies reported using individual-based microsimulation models. Eight of these used a common individual-based microsimulation model referred to as the ‘Harvard-based model’.3839 44,47 50 58 Two studies used a dynamic individual-based microsimulation model referred to as ‘Policy1-Cervix’^{37 41} and two studies used a microsimulation Markov model used for cervical cancer screening (SiMCerC).^{49 53} One study used a Microsimulation Screening Analyses-Cervix (MISCAN) cervix model⁵⁹ (online supplemental table V).

Most studies (13/33) adopted a societal perspective.2833 38,40 43 17/33 studies used a lifetime horizon.2930 33 37,40 44 Countries had different WTP thresholds with eight studies using established WTP^{29 30 39 40 49 50 58 59} and 13/33 studies using a WTP threshold based on the GDP per capita.3337 38 41 43,47 50 52 53 56 The discount rate ranged from 3% to 4% for both costs and benefits. Most studies (16/33) used the 3% discount rate.3033 34 37 38 40 41 44,47 50 52 53 56 59 The most frequently used outcome measures were the lifetime risk of cervical cancer.3839 44,47 50 58 Only a few studies documented whether the participants were from rural/urban areas33,3638 41 (online supplemental table V).

Clinical characteristics

Uptake of the HPV DNA self-sampling strategies in micro-costing studies in LMICs ranged from 50.1%³⁶ to 100%.³⁴ Most of these studies assessed one-off screening. The presumably micro-costing study from an HIC had assumed coverage of 32%, triage compliance ranging from 79% to 82% and colposcopy adherence of 90%.³¹ The CCA study from the UK had triage of pap smear followed by colposcopy for abnormal results. Adherence was assumed to be 66% for triage and 75% for colposcopy.⁵⁴

In case of CEA/CUA in LMICs the assumed coverage of HPV DNA self-sampling ranged from 8.7%⁵³ to 100%.⁴⁶ The assumed sensitivity ranged from 67%⁴⁵ to 97.6%⁴¹ while specificity ranged from 66%³⁸ to 95%.⁴⁷ The follow-up strategy after a positive test in the CEA undertaken in HICs was pap smear and this was followed up by colposcopy.2729 30 39 49 55,59 Two studies included triage which was not followed up by colposcopy^{28 32} while two also reported on using only colposcopy triage.^{30 50} Triage compliance for pap smear ranged from 70%³⁰ to 100%.^{28 58 59} Sensitivity of the screening test ranged from 80%⁵⁹ to 100%³⁹ while the assumed specificity ranged from 72%⁵⁰ to 100%⁵⁹ (online supplemental table VI).

Critical appraisal and methodological assessment

11 studies were graded as excellent.2937,39 41 45 46 50 54 55 59 21 studies were graded as good.2728 30,36 40 43 45 47 9/21 studies that were graded good had missed reporting only 1 item out of 13 items.2730 31 33 40 43 45 47 48 56,58 The most unreported item (10/21) was the description of the intervention,3032 33 40 45 49 51,53 57 followed by the description of diagnostic follow-up (8/21).3234,36 43 47 48 56 One study was graded fair, which did not report on the outcomes, comparator, diagnostic follow-ups, source of data for costs and parameters and sensitivity analysis⁴² (online supplemental table VII).

Synthesis of the results

This section presents study results on cost per woman screened with HPV DNA self-sampling and cost-effectiveness of HPV DNA self-sampling, differentiating between LMICs, HICs and self-sampling strategies.

Cost per woman screened in LMICs

The cost per woman screened ranged from US$31.77⁴² to US$72.22 under community health campaigns³⁴ and was US$42.62 under point of care³⁵ (online supplemental table VIII).

Cost per woman in HICs

Cost per woman screened for the general population was US$56.21 for HPV DNA self-sampling compared with US$79.10 for HPV DNA clinician-collected samples.⁵⁴ Non-attendees were studied in scenario analysis in which the cost per woman screened was US$78.42 for HPV DNA self-sampling and US$79.10 for HPV clinician-collected samples.⁵⁴ Cost per woman screened was US$45.60 for the reminder letter to attend pap smear, US$59.41 for the first reminder with HPV DNA self-sampling kit triage with pap smear, US$73.23 for the second reminder letter and HPV DNA self-sampling kit triage with pap smear and US$69.09 for the reminder letter and HPV DNA self-sampling in the opt-in strategy³¹ (online supplemental table VIII).

Cost-effectiveness of HPV DNA self-sampling strategies for LMICs

Community health campaigns

Four studies evaluated this strategy and found the strategy to be cost-effective.^{38 43 45 46} This includes a study comparing no screening, HPV DNA self-sampling and HPV clinician-collected samples at different levels of coverage in Uganda which found HPV DNA self-sampling to be highly cost-effective at 75% and 100% coverage.⁴⁶ Another study in Uganda found that HPV DNA self-sampling in a ‘screen and treatment’ approach dominated HPV DNA clinician-collected sample triage with VIA and VIA without triage.³⁸ HPV DNA self-sampling was also cost-effective in El Salvador⁴⁵ and India.⁴³

Point of care

Four studies evaluated HPV DNA self-sampling at the point of care,^{37 41 47} with one study evaluating a combination of mailed kits and point of care.⁴⁰ Three of the studies found HPV DNA self-sampling strategy to be cost-effective. One of the studies compared a scenario where only HPV vaccination is provided with: vaccination combined with HPV DNA self-sampling; and vaccination combined with HPV DNA self-sampling and triage.⁴¹ The scenarios with HPV DNA self-sampling were cost-effective compared with vaccination only.⁴¹ HPV DNA self-sampling screen and treatment approach was cost-effective compared with VIA.³⁷ Kositamongkol et al used a combination of mailed and point-of-care screening strategy in Thailand and found HPV DNA self-sampling to be cost-saving.⁴⁰ One study, which was done across three countries: Uganda, India and Nicaragua found this strategy not cost-effective compared with HPV DNA clinician-collected samples.⁴⁷

HPV DNA self-sampling strategy unspecified

Four studies did not specify the types of HPV DNA self-sampling strategy used. They all found HPV DNA self-sampling to be cost-effective.^{33 44 52 53} A Chinese study on the screen and treatment approach that compared HPV DNA self-sampling strategies with or without triage and HPV DNA clinician-collected samples with or without triage found that in comparison to current practice all screen and treatment strategies under both HPV DNA self-sampling and clinician-collected samples were cost-effective. The screening and treatment strategies remained cost-effective whether the analysis was done for urban or rural areas.³³ One study from El Salvador used HPV DNA self-sampling only in the scenario analysis. In this case, HPV DNA self-sampling was highly cost-effective.⁴⁴ The other two studies compared HPV self-sampling and VIA^{52 53} (online supplemental table VIII).

Cost-effectiveness of HPV DNA self-sampling strategies in HICs

Community health campaigns

Only one community health campaign study was conducted in HICs, namely the USA. It found that community health campaigns were cost-effective under the American College of Cardiology and the American Heart Association threshold but not cost-effective under the USA GDP per capita threshold.⁵⁰

Opt in (three studies appear in both opt-in and opt-out strategies^{29 48 55})

Five studies evaluated an opt-in strategy.^{28 29 48 55 58} However, two of these^{28 55} did not specify a WTP threshold and did not declare cost-effectiveness. Boyard et al compared a combination of HPV DNA self-sampling (opt-in) and pap smear as a control with a combination of HPV DNA self-sampling (opt-out) and pap smear and the ICER per woman screened was US$63.98.²⁸ Knauss et al compared reminder letters for pap smears as a control to an opt-in and reported an ICER of US$161.84 per woman screened.⁵⁵ The other three studies found HPV DNA self-sampling cost-effective^{29 48 58} (online supplemental table VIII).

Opt out (three studies appear in both opt-in and opt-out strategies^{29 48 55})

Six studies reported on an opt-out strategy.^{27 29 39 48 55 59} Two had no threshold and did not declare cost-effectiveness.^{27 55} In Aarnio et al study, the ICER was US$−423.49 per screened woman compared with the pap smear while Knauss et al’ study reported an ICER of US$213.49 per woman screened compared with the pap smear reminder letter. The remaining studies found an opt-out strategy to be either cost-effective or cost-saving^{29 39 48 59} (online supplemental table VIII).

HPV DNA self-sampling unspecified

Six studies did not specify the type of HPV DNA self-sampling.^{30 32 49 51 56 57} Five of these studies found HPV self-sampling cost-effective.^{30 32 56 57} Among them, two studies had a mailed HPV DNA self-sampling strategy,^{32 57} one HPV DNA self-sampling triage with pap smear/colposcopy³⁰ and a study comparing LBC with reflex HPV plus 25% HPV DNA self-sampling to pap smear.⁵⁶ One study did headroom analysis for the combination of HPV DNA clinician-collected samples and HPV DNA self-sampling triaging with molecular for high-risk HPV and others with cytology and found the molecular test to be cost-effective with the corresponding maximum reimbursable price ranging from US$200.25 to US$771.32.⁴⁹ One study found that HPV DNA self-sampling followed by different molecular triage was not cost-effective⁵¹ (online supplemental table VIII).

Estimated budget impact for HPV self-sampling

Four studies included BIA in LMICs^{37 40 41 47} and one from HICs.⁴⁸ The annual budget for HPV DNA self-sampling varied across countries, from about US$20 million in Thailand⁴⁰ to about US$50 million (including a combination of vaccination and HPV DNA self-sampling costs) in Malaysia.⁴¹ For the screen and treatment approach, the budget ranged between US$1.55 and US$5.59 million, depending on the screening frequency, age at screening and whether it was a 5-year or 10-year estimate.³⁷ For HICs, it ranged from US$0.10 to US$0.43 million depending on the target population and the type of visit⁴⁸ (table 1).

Table 1. Budget impact of HPV DNA self-sampling.

Author, country	Screening strategy	Inflation and currency exchange	Average budget per year (in million Thai Baht)	GDP deflator 2023	Inflated average annual budget to 2023 (in million Thai Baht)	Exchange to dollar (US$1=34.80 Thai Baht, 2023) (in million $)
Kositamongkol et al, Thailand40	HPV DNA self-sampling	Consumer Price Index, cost reported in 2022 Thai Baht US$1=35.07 Thai Baht in 2022
	5 years		฿661.00	1.012	฿669.12	$19.23
	10 years		฿681.00	1.012	฿689.37	$19.81
	HPV DNA self-sampling and clinician-collected sample
	5 years		฿1338.00	1.012	฿1354.44	$38.92
	10 years		฿1377.00	1.012	฿1393.92	$40.06
Keane et al, Malaysia41			Average budget per year—USD millions		Inflated average annual budget to 2023	N/A
	Vaccination only	Inflated from 2014 to 2019 (~12.7%). Malaysian Ringgits (RM) 4.14=US$1)	$30.8‒$30.9	1.10	$33.9‒$34.02
	Vaccination+HPV self-sampling		$45.2‒$45.4	1.10	$49.8‒$50.00
	Vaccination+HPV self-sampling+triage		$44.3‒$44.4	1.10	$48.8‒$48.9
Nguyen et al, Papua New Guinea37			Average budget per year in US$ million		Inflated average annual budget to 2023 in US$ million	N/A
	No screening	Did not account for inflation PNG Currency (PGK) 1=US$0.3, 17 October 2019	$0.38	1.13	$0.43
	HPV DNA self-sampling screen and treat
	5-year budget
	Once in life years (at age 35)		$1.38	1.13	$2.46
	Twice in life years at the age of 35 and 40		$2.18	1.13	$2.68
	Twice in life years at the age of 30 and 40		$2.38	1.13	$2.30
	Twice in life years at the age of 35 and 45		$2.04	1.13	$3.72
	Thrice in life years at age 30, 35 and 40		$3.30	1.13	$5.59
	Six times in life years (30–55)		$4.96
	10-year budget
	Once in life years (at age 35)		$1.39	1.13	$1.57
	Twice in life years at the age of 35 and 40		$2.13	1.13	$2.40
	Twice in life years at the age of 30 and 40		$2.34	1.13	$2.64
	Twice in life years at the age of 35 and 45		$2.03	1.13	$2.29
	Thrice in life years at age 30, 35 and 40		$3.17	1.13	$3.57
	Six times in life years (30–55)		$4.69	1.13	$5.28
	5-year and 10-year budgets were calculated as the financial costs associated with cervical cancer screening, diagnosis and treatment over the first 5 years (2023–2027) and 10 years (2023–2032) of implementation. Averaged per year by taking the sum of the year total budget divided by 5-year and 10-year budget by 10
Meenan et al48			Average budget per year in US$ million		Inflated average annual budget to 2023	N/A
	Direct mail	Reference year 2022
	Women adherent		$0.11–0.41	1.04	$0.11–0.43
	Overdue		$0.10–0.34	1.04	$0.10–0.36
	Opt in
	Women with unknown screening status		$0.11–0.39	1.04	$ 0.11–0.40
	Average budget was calculated by dividing the total 4 years BIA by 448

BIA, budget impact analysis; GDP, gross domestic product; HPV, human papillomavirus; USD, US dollar.

Discussion

This study has documented the current economic evidence for HPV DNA self-sampling and its strategies. It includes 33 studies on the economic evaluation HPV DNA self-sampling. Costs per woman screened for partial economic evaluation studies (micro-costing and CCA) and cost-effectiveness for CEA/CUA for both LMICs and HICs are presented. It is the first to study the health economics of all HPV DNA self-sampling strategies. A previous systematic review included only outreach strategies.¹⁷

The included studies used different types of economic evaluation and costing designs, including micro-costing, CCA and CEA/CUA. There is no CBA reported.

The included HPV DNA self-sampling strategies included community health campaigns where the collection of samples is done outside the facility in outreach programmes, such as the use of tents and door-to-door. The cost per woman screened for the strategy ranged from US$31.77⁴² to US$72.22.³⁴ All studies under this strategy (one from HICs) found the study to be cost-effective.^{38 43 45 46 50}

The second HPV DNA self-sampling strategy reported was point-of-care. In this case, HPV DNA self-sampling is done at the facility. Most of the studies did not specify whether this was opportunistic (targeting women attending other services at the facility). Four studies reported on this method which were all done in LMICs.^{35 37 41 47} Cost per woman screened was US$42.62.³⁵ 2/3 studies which used CEA/CUA found the strategy to be cost-effective.^{37 41}

Mailed HPV DNA self-sampling kits included opt-in where a kit is sent to a client prior to a request and opt-out in which a kit is sent to eligible women directly without requesting it.27,2939 48 54 55 58 59 Cost per woman screened was US$56.21.⁵⁴ All studies under opt-in/opt-out declaring cost-effectives found this strategy to be cost-effective.^{29 39 48 58 59}

The previous systematic review included a limited number of studies from LMICs.¹⁷ However, the burden of cervical cancer in LMICs is higher than in HICs, with age-standardised rates of 14 per 1 00 000 women per year.⁶⁰ Of the 604 127 cases of cervical cancer reported in 2020, 532 239 (88.1%) were from LMICs.⁶⁰ Although we have found more studies from LMICs (n=10) compared with the previous review (n=4), more cost-effectiveness studies are needed especially when a country wants to introduce a particular strategy, given the heterogeneity of the included studies. In sub-Saharan Africa where the burden of cervical cancer is high and where more than ten countries with high incidence of cervical cancer are found,⁶⁰ studies were found done in only two countries with CEA in Uganda^{38 46 47} and micro-costing in Kenya.^{35 36 42} For HICs, only Japan and USA were outside Europe.

Our findings show that studies differ in terms of the models, screening frequencies, target population, self-sampling strategies, follow-up after a positive result, referral systems and treatment guidelines, and hence limit formal synthesis and transferability of the findings. There are further differences in infrastructure, HIV incidence and HPV vaccination. This was also observed Sun et al⁶¹ The majority of studies in HICs targeted women who are non/under-screeners using opt-in/opt-out, while studies in LMICs targeted women in the general population using community health campaigns.

Models to assess cost-effectiveness included individual microsimulations, decision trees, hybrid models and descriptives. The Harvard model was used in eight studies and has been applied by the same authors using data from different countries.3839 44,47 50 58 Some of these models are extensively validated cervical microsimulation models.^{41 62} The Harvard and Policy1-Cervix models have been found to produce similar results.⁴¹ The Policy1-Cervix model was used to evaluate the timeline for the elimination of cervical cancer globally, and the impact of WHO interventions as a comparative analysis of 78 LMICs.^{63 64} Adopting these models in LMICs faces the challenge of data availability for individual parameters to inform the modelling process and treatment options.^{37 44 45}

Community health campaign studies were predominantly undertaken in LMICs. It is an important strategy for reaching the underserved population since it has a higher uptake and has face-to-face contact.^{65 66} Point-of-care self-sampling has been less studied especially in HICs, even though it has been found to increase the uptake of cervical cancer screening due to the confidence built on the presence of a clinician and instructions offered face-to-face.⁶⁵ Although HPV DNA self-sampling is generally acceptable, however, there are differences in preferences for the different strategies of HPV DNA self-sampling, though with limited evidence.⁶⁷ There is no one approach that fits all hence calling for further analysis considering different HPV DNA self-sampling strategies to different groups including socioeconomic, demographic, disabilities as well as localities.⁶⁷

Studies in LMICs targeted women eligible for cervical cancer screening to increase the uptake which is still low while most studies in HICs targeted the under-screeners/never-screeners. For LMICs there was lack of studies for the rural population while for HICs, the under-screened population was defined as those who have never attended screening or those who do not attend according to the required intervals for regular screening. Humphreys et al criticised the inclusion of such a population which does not describe the specific population at risk. They recommend that, for equity purposes specific populations at risk should be studied, since studies have shown high risk for cervical cancer for example for prisoners, women with HIV and sex workers. They called for further cost-effectiveness exploring the population at risk.^{68 69} Only two studies from the USA targeted a specific subgroup of the under-screened population, in this case low-income and African American populations.^{32 50} Lack of subgroup analysis was also observed by other authors, who found that the strategies to increase uptake among under-screened populations were general and did not consider subgroups such as race, income or migrants.⁶¹

Some studies used the WTP threshold based on the GDP per capita. However, WHO is currently cautioning against using this threshold since it overestimates WTP. GDP values were never intended to be used as thresholds and recent estimates for WTP thresholds show this threshold to be lower. It was found that on average the WTP per disability adjusted life years averted was 0.34 times the GDP per capita for countries with low human development index, reflecting what the country can afford rather than the GDP per capita-based thresholds.⁷⁰

Even though the CEA/CUA studies differed in terms of time horizon, costing perspectives and other dimensions, the result of HPV DNA self-sampling being cost-effective was almost unanimous. Our findings concur with the previous review which found that 14 of the 16 studies included in their review showed HPV DNA self-sampling to be cost-effective.¹⁷ This study also concurs with the cost-effectiveness study for cervical cancer screening for the under-screened population in HICs which found that HPV DNA self-sampling cost-effective.⁶¹ This finding also supports the WHO’s 2030 cervical cancer elimination strategy which recommends the use of HPV DNA as the primary screening strategy in place of pap smear and VIA.² The cost-effectiveness of HPV DNA self-sampling from this review complements clinical effectiveness studies including accuracy, agreement, uptake and acceptability of HPV DNA self-sampling compared with the clinician-collected samples.^{19 65 71} The uptake of HPV DNA self-sampling compared with clinician-collected samples was higher regardless of the HPV DNA self-sampling strategy used.⁶⁵ The agreement between HPV DNA self-sampling and the clinician-collected sample was 88.7% and a kappa of 0.72, while the pooled sensitivity ratio on detection of CIN2+ was 0.99 when using the PCR. It was lower when signal amplification tests were used.⁷¹ Campos et al provide a framework for decisions when the sensitivity of self-sampling is lower than that of clinician-collected samples in which increased uptake of 20% of HPV DNA self-sampling could offset the loss of sensitivity of HPV DNA self-sampling.⁷² The average acceptability of HPV DNA self-sampling was 97%.¹⁹

The affordability of HPV DNA self-sampling is important in informing its implementation as countries are transitioning to this new screening method. Our findings show that BIA varied across countries calling for a country-specific BIA when a country is evaluating the affordability and implementation feasibility of this cervical screening method.

Cervical cancer is highly preventable through primary and secondary prevention strategies.² The primary prevention includes education and vaccination.² Although different countries have introduced vaccination programmes for HPV, only three studies have modelled the cost-effectiveness of HPV DNA self-sampling in combination with vaccination.^{41 56 58}

Strengths and limitations of the study

This review is the first to include all HPV DNA self-sampling strategies and the first comprehensive systematic review since WHO published a guideline that encourages the use of HPV DNA screening as the primary test in place of pap smear and VIA. It includes studies across the globe with almost equal representation of LMICs and HICs. However, the review also has limitations. Our search included only studies published in English. The review did not include non-peer-reviewed evidence like grey literature and reports. Therefore, may miss evidence from non-English publications and evidence from grey literature which could provide more evidence to base the conclusion. One reviewer screened all articles; two reviewers screened 20% of articles, which may introduce errors and bias. However, we tried to minimise any bias or errors arising. We inflated costs based on the 2023 World Bank GDP deflator; however, countries have different base years and currencies. We did not use the international dollar because nearly all studies that had reported the costs in US dollars did not use the international dollar conversion. Hence, purchasing power parity was not considered, limiting the comparability and generalisability of the findings.

Conclusions

HPV DNA self-sampling strategies appear to be cost-effective for cervical cancer screening and can be used to reduce the burden of incidence and deaths from cervical cancer in both LMICs and HICs. Different strategies can be used to deliver HPV DNA self-sampling depending on a country’s healthcare settings, clients preferences and resources. However, due to the variability of costing studies in terms of healthcare costs and study design, transferability of the findings to other settings may not be advisable.

Recommendation for further studies

Further economic evaluation studies are recommended for community health campaigns in HICs, and point of care in both LMICs and HICs, especially opportunistic screening. Further studies evaluating cost-effectiveness in subgroups of the under-screened and general population are recommended. Implementation of the possible alternatives for HPV DNA self-sampling requires different resources, client preferences and the consideration of different health outcomes. Implementation studies on these strategies considering the subgroups, including implementation barriers, are recommended when a country plans to introduce this screening approach. There are future proposed screening methods such as DNA methylation, protein biomarkers that have HPV antibodies and oncoproteins and automated visual evaluation of digital images which may be more cost-effective than the current methods and research evaluating the effectiveness of these newer methods is required.

Recommendation for practice and policy

Countries planning to introduce HPV DNA self-sampling should carry out thorough economic evaluation and implementation studies given the variety of strategies through which HPV DNA self-sampling can be delivered. Being cost-effective does not mean a strategy is inexpensive, studies such as BIA and stakeholder buy-in are recommended before implementation. As data become more available, future modelling studies for LMICs are encouraged and will be able to better the challenges of including real-world data collection to inform the modelling process.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.