Health-Related Quality of Life of Patients with HPV-Related Cancers in Indonesia

Didik Setiawan, PhD1,2,*, Arrum Dusafitri, BPharm2, Githa Fungie Galistiani, MSc2, Antoinette D.I. van Asselt, PhD1,3, Maarten J. Postma, PhD1,4

1 Unit of PharmacoEpidemiology & PharmacoEconomics (PE2), Department of Pharmacy, University of Groningen, Groningen, The Netherlands; 2 Faculty of Pharmacy, Universitas Muhammadiyah Purwokerto, Purwokerto, Indonesia; 3Health Technology Assessment Unit, Department of Epidemiology, University Medical Centre Groningen, Groningen, The Netherlands; 4Institute of Science in Healthy Aging & healthcaRE (SHARE), University Medical Center Groningen, Groningen, The Netherlands

ABSTRACT

Background: Human papillomavirus (HPV)-related cancers are a serious concern in developing countries. Valid estimates of a country-specific health-related quality of life (HRQOL) for patients with HPV-related cancers provide a substantial tool in determining the burden of the disease. Objectives: To investigate the HRQOL of patients with HPV-related cancers in Indonesia. Methods: The HRQOL of patients with HPV-related cancers (cervical, uterine, nasopharyngeal, head and neck, and anogenital cancer) was assessed using the EuroQol five-dimensional questionnaire (EQ-5D). Validity and reliability were assessed by means of construct validity and test-retest reliability methods, respectively. Subsequently, the EQ-5D utility index was calculated using the Thailand value set. Results: The EQ- 5D came out as a valid and reliable questionnaire for measuring the HRQOL of patients with HPV-related cancers in Indonesia. From a total of 520 patients diagnosed with HPV-related cancers, 404 patients were excluded because of not fulfilling the inclusion criteria, and so 116 patients finally participated in the study. The mean age of the patients was 47.5 ± 12.03 years. Most of the patients were women (56.0%) and married (97.4%), and less than half of them had finished high school (32.7%). Moreover, the proportions of nasopharyngeal, cervical, head and neck, anogenital, and uterine cancers in the study population were 29.3%, 24.6%, 22.4%, 14.2%, and 9.5%, respectively. The average HRQOL of the patients with HPV-related cancers was 0.69 ± 0.10, with the highest and lowest estimates applying to uterine cancer (0.84 ± 0.29) and head and neck cancer (0.58 ± 0.33), respectively. Conclusions: The HRQOL of patients with HPV-related cancers was found to be reduced to a certain extent in our study for Indonesia.Keywords: cervical cancer, health-related quality of life, HPV-related cancer, human papillomavirus, Indonesia.

Introduction

Human papillomavirus (HPV) infections are a well-established cause of not only cervical cancer [1–4] but also other cancers, including head and neck, anogenital (anus, vulva, vagina, and penis), uterine, and nasopharyngeal cancers [5,6]. In addition, the global burden of HPV-related cancers is increasing, and develop- ing countries, including Indonesia, bear a high proportion of this burden [7]. For example, data reported by the Ministry of Health in Indonesia showed that HPV-related cancers, especially cervical and nasopharyngeal cancers, were among the 10 cancers with the highest incidence and mortality rate in Indonesia [8].

Various health technologies such as chemotherapy [9,10], prevention strategies [11–14], new hormonal therapies [15,16], and clinical practice guidelines [17] are constantly developed and changed to combat cancer. The implementation of these health technologies has resulted in a wide range of improvements in outcome measures, with success being influenced by cancer type, study site, outcome measure, or even the cancer treatment and prevention itself [18,19]. Nevertheless, various, and sometimes life-threatening, side effects from the treatments could possibly occur, and the health technologies as well as their side effects occasionally influence not only the health status of the patients but also their social and emotional well-being.

Within this broad variety of issues, the need for comparability of outcomes has been recognized and the term “health-related quality of life” (HRQOL) for this purpose is widely accepted [20,21]. HRQOL generally captures the complete health state of an individual because it includes several important health parameters from, for example, the physical, psychological, and social health dimensions [22]. Any reduction in the value of HRQOL reflects a reduction in the health of the individual. Therefore, HRQOL is able to convey important information for assessing the overall burden of a disease and the effectiveness of interventions as well.

With regard to cancer disease, HRQOL measurement tools may be classified as generic, general cancer, cancer-site–specific, and cancer-problem–specific. General cancer, cancer-site–specific, and cancer-problem–specific HRQOL measurement tools generally provide more detailed information in a single cancer-type or cancer-related problem, whereas an ultimate advantage of generic HRQOL measurement tools is that because they can be implemented over a wide range of conditions and interventions, they enable comparison of outcomes across diseases [23,24]. One example of a generic questionnaire is the EuroQol five-dimensional questionnaire (EQ-5D), which was developed by the Euro Qol group and is widely used across the world [25–27]. Further advantages of the EQ-5D are that it has only five questions and it is easy to administer and complete. The EQ-5D is of two types: the three-level EQ-5D (EQ-5D-3L) and the more recently developed version with five levels (EQ-5D-5L). Several studies support both the validity and the sensitivity of the EQ-5D in patients with cancer [24,28]. Nevertheless, some studies suggest that the EQ-5D-5L has less ceiling effects and better discriminative abilities with potentially more power to detect differences between groups as compared with the EQ-5D-3L [29,30].

Indonesia-specific HRQOL of HPV-related cancers provides initial information in decision-making processes because it will assist subsequent processes such as cost-utility analyses. Subsequently, this will allow for comparing outcomes of interventions for HPV-related cancers with the outcomes of other interventions, within as well as outside the area of HPV-related cancers. The purpose of this study was to investigate the HRQOL of patients with HPV-related cancers in Indonesia using the EQ-5D-5L.

Methods

A descriptive cross-sectional study was conducted, directed at the HRQOL of patients with HPV-related cancers in Indonesia. The study was divided into two main activities, consisting of pilot testing and the main study. Notably, this study was approved by the ethics committee of the Faculty of Medicine, Gadjah Mada University, Yogyakarta, Indonesia.

Pilot Study

Initially, a pilot study was conducted to evaluate the validity and reliability of the EQ-5D for patients with HPV-related cancers in the specific Indonesian setting. A convenience sample of 30 patients with HPV-related cancers was recruited from Dadi Keluarga Public Hospital, Purwokerto, Indonesia. On the recruitment day (day 0), patients filled out both the EQ-5D-5L and the European Organization for Research and Treatment of Cancer Quality-of-Life Questionnaire C30 (EORTC QLQ-C30), a cancer- specific questionnaire. On day 14, patients filled out the EQ-5D-5L a second time.

As mentioned, the EQ-5D is a generic quality-of-life instrument developed by the Euro Qol group and contains five questions and a visual analogue scale (VAS). The questions comprise five different dimensions, that is, mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. We adopted the updated version of the EQ-5D containing five levels of severity on each dimension because it was considered to be more sensitive and reliable compared with the previous three-level version (EQ-5D-3L). In addition, the VAS records a patient’s self-rated health perception, graded between 0 (worst) and 100 (best).

The EORTC QLQ-C30 is a disease-specific questionnaire developed mainly to assess the quality of life of patients with cancer. It consists of three different scales: functional (15 questions), symptoms (13 questions), and global quality of life (2 questions). The EORTC QLQ-C30 was chosen as the comparator in the validation process because the Indonesian version has been validated [31] and some questions, particularly on global health status, physical function, role function, emotional function, and pain, have been identified as significant predictors of the EQ-5D outcomes according to several mapping studies [32–35]. There- fore, construct validity testing, that is, assessing whether the EQ- 5D can be interpreted as a meaningful measure of quality of life, was performed by measuring the correlation between each mapped question of the EQ-5D and the EORTC QLQ-C30 because they have the same construct (convergent validity method) [36]. Each of the EORTC QLQ-C30 subscales was considered to be correlated with an EQ-5D dimension when the statistically significantly Pearson correlation coefficient indicated so [32,34,35]. Furthermore, the value of the Pearson correlation coefficient represents the magnitude and the direction of the correlation [37].

To ensure the reliability of the EQ-5D, the two measurements (day 0 and day 14) were compared within patients to assess test- retest reliability [38]. This method was chosen because in patients with cancer, significant changes in quality of life rarely occur within a 14-day interval and although a clear recommendation to avoid recall bias is scarcely available and often debatable, this 14-day interval is considered sufficient to avoid recall bias [39]. Test-retest reliability represents measurement stability, using intraclass coefficients (ICCs) [40,41]. ICC values were interpreted as follows: weak agreement if the ICC is lower than 0.40, good agreement if the ICC lies between 0.40 and 0.75, and excellent agreement if the ICC is 0.75 or higher [41]. In addition, the Cronbach α was calculated to assess internal consistency, that is, to determine whether the EQ-5D questions together measure the same construct of the HRQOL of patients with HPV-related cancers. Internal consistency was considered to be good if the Cronbach α was higher than or equal to 0.80 [38].

Main Study

All registered patients from 2010 to 2015 in Margono Public Hospital, Purwokerto, Indonesia, were screened for their eligibility. Inclusion criteria were as follows: patients were older than 18 years, were diagnosed with any HPV-related cancer (cervical, uterine, nasopharyngeal, head and neck, or anogenital cancer), and had received any care in the hospital. Patients with chronic disease comorbidity were excluded. A clinical convenient sample was implemented for all patients who were eligible and received and signed informed consent before both the EQ-5D and the EORTC QLQ-C30 were filled out while the patient’s sociodemographic and clinical information, such as age, sex, education level, and diagnosis of cancer, were collected from medical records. These two separate data sets were matched according to the medical record number for each patient.

Each dimension of the EQ-5D-5L is presented as a dichotomous outcome: no problem (answer level 1) or problems (answer levels 2–5). Furthermore, a major outcome provided by the EQ-5D is an index-based value (utility index) allowing the calculation of quality-adjusted life-years [42]. This utility index, ranging from lower than 0 (worse than death) to 1 (perfect health), is generated by transforming patients’ responses to the five questions, using a country-specific value set. Because an Indonesia-specific value set is not available yet, the utility index was calculated using the Thailand value set and subsequently presented by cancer type [43].

Statistical analysis was performed to identify the differences of both the EQ-5D utility index and the EQ-5D VAS score on the basis of the study characteristics and cancer groups. Because the results of the Shapiro-Wilk normality test showed that both parameters were not normally distributed (P value of 0.000 for both parameters), a nonparametric test was implemented. The Mann-Whitney U test was used for the variables sex and marital status, whereas the Kruskal-Wallis H test was used for education level and cancer groups.

Table 1 – Reliability test of the EQ-5D for patients with HPV-related cancer.
Dimension	ICC	Cronbach α
Mobility	0,97	0,84
Self-care	0,95
Usual activities	0,79
Pain/discomfort	0,84
Anxiety/depression	0,82
EQ-5D VAS	0,73
EQ-5D, EuroQol five-dimensional questionnaire; HPV, human papillomavirus; ICC, intraclass correlation coefficient; VAS, visual analogue scale.

Results

Pilot Study

The age of the patients in the pilot study ranged from 22 to 68 years with a mean of 51.5 ± 11.5 years (data not shown). Most of the patients were women (56.7%) and diagnosed with head and neck cancer (70.0%), followed by cervical cancer at 13.4% and nasopharyngeal cancer at 10.0% (data not shown). According to the ICCs generated from test-retest reliability, each sub scale in the Indonesian version of the EQ-5D had an excellent agreement (≥0.75). The EQ-5D VAS scores had a good agreement with the ICC of 0.73. In addition, the EQ-5D had a good internal consistency because the value of Cronbach α was higher than 0.80 (Table 1).

There were significant relationships between almost all the dimensions of the Indonesian version of the EQ-5D with mapped subscales of the EORTC QLQ-C30 including physical function, role function, fatigue, and pain. Only the mobility dimension of the EQ-5D seemed uncorrelated with the social function subscale of the EORTC QLQ-C30. Meanwhile, only the global health status sub scale of the EORTC QLQ-C30 correlated with the EQ-5D anxiety/depression dimension, whereas only the cognitive function sub scale correlated with the pain/discomfort dimension. The EQ-5D VAS was apparently correlated with all the EORTC QLQ- C30 sub scales except cognitive function (Table 2). Strong correlations were generated by some of the EQ-5D dimensions and the EORTC QLQ-C30 subscales, such as between self-care and physical function, usual activities and physical function, and pain/ discomfort and pain, with correlation coefficients of −0.870, −0.855, and 0.842, respectively.

Main Study

From the hospital database, a total of 520 patients were identified for this study of whom 374 were eligible for inclusion and subsequently, if possible, were visited at their home. Finally, 116 patients were included in the study (Fig. 1).

The mean age of the patients was 47.5 ± 12.03 years (range 18–75 years). Most of the patients were women (56.0%) and married (97.4%), and less than half of them had finished high school (32.7%). Moreover, the proportions of nasopharyngeal, cervical, head and neck, anogenital, and uterine cancers in the study population were 29.3%, 24.6%, 22.4%, 14.2%, and 9.5%, respectively (Table 3).

The statistical analysis showed that utility index was not statistically significantly different with respect to most of the patients’ characteristics at baseline, except for education level (P ¼ 0.039). Nevertheless, because the analysis was performed using nonparametric methods, the specific differences in each group could not be identified. Furthermore, the EQ-5D VAS score was not statistically significantly different with respect to patients’ characteristics.

On the basis of information collected using the Indonesian version of the EQ-5D-5L (Table 4), the highest reduction on the mobility dimension was seen in cervical cancer (46.42%). With regard to the self-care dimension, only a few patients with HPV-related cancers (o20%) encountered problems. Meanwhile, almost half (41.38%) of the patients reported difficulties on the usual activity dimension. Problems on the pain/discomfort dimension were heterogeneous. Patients with nasopharyngeal cancer reported a high proportion of pain/discomfort issues (67.65%), whereas none of the patients with uterine cancer reported pain/ discomfort. In addition, a high proportion of patients with naso- pharyngeal cancer (55.88%), cervical cancer (53.85), and anogenital cancer (53.57%) experienced anxiety/depression problems.

The overall value of the EQ-5D VAS for all included patients with HPV-related cancers was relatively high (77.38 ± 6.42), ranging from 72.05 ± 25.55 to 88.54 ± 14.60 for anogenital cancer and uterine cancer, respectively. The utility index was also moderately high (0.69 ± 0.10) for the overall patient population included. The highest value for the EQ-5D utility index was found in uterine cancer (0.84 ± 0.29), followed by nasopharyngeal cancer (0.75 ± 0.30), anogenital cancer (0.68 ± 0.35), cervical cancer (0.61 ± 0.39), and head and neck cancer (0.58 ± 0.33). Finally, the Kruskal Wallis H test showed that both utility index and VAS were not statistically significantly different because the P values were 0.059 and 0.144, respectively.

Table 2 – Convergent validity reflected with Pearson correlation coefficients.
EORTC QLQ-C30	EQ-5D
	Mobility	y Self-care	Usual activities	Pain/discomfort	Anxiety/depression	EQ-5D VAS
Global health status	−0.198	−0.08	−0.081	−0.288	−0.482*	0.489*
Physical function	−0.777*	−0.870*	−0.855*	−0.502*	−0.616*	0.691*
Role function	−0.675*	−0.534*	−0.719*	−0.613*	−0.481*	0.618*
Emotional function	−0.535	* −0.456†	−0.327	−0.337	−0.592*	0.426†
Cognitive function	0.114	0.059	−0.025	−0.490*	0.06	−0.023
Social function	−0.304	−0.406†	−0.364†	−0.498*	−0.433†	0.363†
Fatigue	0.498*	0.430†	0.400†	0.520*	0.519*	−0.527*
Pain	0.536*	0.503*	0.625*	0.842*	0.538*	−0.663*
EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality-of-Life Questionnaire C30; EQ-5D, EuroQol five- dimensional questionnaire; VAS, visual analogue scale.
⁎ Significantly correlated at the P o 0.01 level.
† Significantly correlated at the P o 0.05 level.

Fig. 1 – Flow chart of the HPV-related cancer patients studied.

Discussion

Our study shows that there are various reductions in the HRQOL of patients with HPV-related cancers in Margono Public Hospital, Purwokerto, Indonesia. Although the reference utility index for the healthy population in Indonesia is not available yet, two studies showed that the utility index in a healthy population in Sri Lanka and South Australia was 0.85 (95% confidence interval 0.84–0.87) and 0.91 (95% confidence interval 0.90–0.91), respectively. These findings are apparently congruent with a study from Australia that suggested that the HRQOL of patients with HPV- related cancers ranged from 0.57 to 0.79 [44]. In particular, our finding on head and neck cancer is similar to findings in studies from Italy (0.69 ± 0.30) and Australia (0.58 ± 0.50), but these findings were considerably lower than the finding in a study from Canada (0.83 ± 0.14) [44–46]. Several studies in patients with cervical cancer reported similar utility indexes, such as an Italian study (0.58 ± 0.31), a US study (0.68), and a study in another region of Indonesia using the EQ-5D-3L and the Malaysian value set (0.76 ± 0.20) [45,47,48]. Furthermore, the utility indexes for anogenital cancer and nasopharyngeal cancer from this study were slightly higher compared with what were found in Italy (0.50 ± 0.26) and Australia (0.58 ± 0.05), respectively [44,45]. The differences in utility indexes across studies might be caused by several issues including different perceptions of health across populations and countries as partly represented by the different value sets [49], several characteristics including age and education [50],and also the use of different types of EQ-5D (notably, 5L and 3L) [30,51].

According to our study, the most commonly reported problems by all types of patients with HPV-related cancers were pain/discomfort followed by anxiety/depression. These findings were considerably strengthened by previously published studies in patients with cervical cancer [48] and gynecological cancer [31] in Indonesia using the EQ-5D and the EORTC QLQ-C30, respectively. The finding of zero cases of pain/discomfort reported by patients with uterine cancer may have been caused by the small number of patients, or the fact that patients were fully adapted to their condition. Nevertheless, most studies on the HRQOL of patients with cancer explained that the pain/discomfort dimension had the highest proportion of problems reported in comparison with other dimensions in the EQ-5D [25,27,52,53]. Although several recommendations on cancer-related pain management exist [17,54], cancer-related pain, commonly caused by both the cancer itself and the cancer treatment, is still an important issue [55,56].

The EQ-5D was found to be a reliable and valid instrument for measuring the HRQOL of patients with HPV-related cancers in Indonesia, as evidenced by excellent agreement on test-retest reliability (ICC ≥ 0.75 for all dimensions) and strong correlations with most of the dimensions of the EORTC QLQ-C30. The EORTC QLQ-C30 was chosen as the comparator in the validation process because it has been mapped for each dimension in the EQ-5D for patients with cancer [35]. Furthermore, our methods and results were consistent with other EQ-5D validation reports in several diseases and countries including cervical cancer in Taiwan [27] and chronic diseases (cardiovascular disease, respiratory disease, depression, diabetes, liver disease, personality disorders, arthritis, and stroke) in a multicountry setting [30].

The EQ-5D was chosen because this questionnaire provides a single utility index and has several advantages, including its brevity and ease of administration, and the availability of previous mapping exercise results. Also, it is recommended by several guidelines [32,35,57–59]. The utility index produced by the EQ-5D provides a simpler comparison between interventions for different health problems than do other patient-specificor disease-specific questionnaires such as the Patient Generated Index or the Cancer Patient Experiences Questionnaire [60,61]. Furthermore, the implementation of utility information for economic evaluation is generally acceptable, quite often even required, for health technology assessment processes in almost all regions in the world.

Although the prevalence of chronic diseases in patients with cancer is substantial [62], we excluded patients with chronic diseases because several studies showed that chronic diseases, such as stroke, chronic obstructive pulmonary disease, or chronic kidney disease, further significantly decrease the HRQOL of patients with cancer [50,63–65]. A study from Serbia showed that patients with cancer with chronic disease complications showed a higher level of depressive symptoms [65]. There was also a substantial reduction in the physical and social functioning of patients with breast cancer with chronic disease complications [50]. Although an Indonesian-specific HRQOL estimate with respect to the influence of chronic disease complications in patients with cancer would be generally interesting, this would require a substantially larger sample size.

One of the limitations of this study was that the cancer substages could not be presented because the medical records or hospital database did not provide sufficient information. Cancer stage information is valuable in cancer treatment because guidelines generally provide treatment recommendations on the basis of cancer stage. The lack of information about the substage of each cancer might be a reason why no cases of pain/discomfort got reported by patients with uterine cancer. It may indeed be that most patients were in the early stage of cancer. A possible explanation for the unavailability of data on cancer stage is that the documentation process in the hospitals concerned might not be standardized; there may even be a lack of standard operating procedures altogether in this area. An Indonesia- specific EQ-5D value set for the five-level version is not available yet. Therefore, we converted the descriptive information produced by the questionnaire into a health utility index by using the value set from Thailand [43]. The Thailand-specific value set was chosen because Indonesia and Thailand have several similarities including social, cultural, and economic factors. More- over, the adoption of another country’s value set is practically acceptable [66].

Because the patient data were collected from a single public hospital in Central Java, the generalizability of the result could be considered questionable. Nevertheless, Margono Public Hospital can be considered as a representative of a public hospital that provides services for patients with cancer at the district and province level in the Indonesian region. According to the data from the Ministry of Health, every province has at least one hospital that has comparable characteristics with Margono Public Hospital. In addition, an analysis in terms of demographic information for respondent versus nonrespondent patients could increase the generalizability of the results. Therefore, further study should also collect the data from nonrespondent patients.

The descriptive cross-sectional approach used in this study limits the amount of information provided. Further research adopting a comparative cohort study could be an interesting option for providing the complete picture, including the ratio of HPV- related cancers’ utility indexes compared with a healthy population in Indonesia. Nevertheless, this study provides a significant addition to the literature on HPV-related cancers’ utility values in Indonesia by implementing a valid and reliable questionnaire and coming up with findings mostly consistent with various other studies in the same field in other regions of the world.

Further research into the HRQOL of patients with precancer, substages of cancer, or even HPV-related genital warts is worth - while in completing the information on the full scale of consequences of HPV infections in Indonesia. In terms of the decision-making process, a cost-effectiveness analysis on HPV- related cancer prevention strategies using our findings will provide useful information for decision makers in Indonesia.

Acknowledgments

We acknowledge the assistance and support received from Fretty Setiawati, Mayang Setianing Hadi, Siti Robi’atul Adawiyah, and Sulastri for data collection in this research.

Source of financial support: These findings are the result of work supported by the Directorate General of Higher Education Scholarship, Ministry of National Education, Indonesia. The authors’ work was independent of the funders, who had no role in the study design, analysis of data, writing of the manuscript, or decision to submit for publication. The views expressed in this article are those of the authors, and no official endorsement by the Ministry of National Education is intended or should be inferred.

Cost-Utility Analysis of Human Papillomavirus Vaccination and Cervical Screening on Cervical Cancer Patient in Indonesia

Didik Setiawan, MSc, Apt1,2,*, Franklin Christiaan Dolk, MSc1, Auliya A. Suwantika, PhD1,3, Tjalke Arend Westra, PhD1, Jan C. WIlschut, PhD4, Maarten Jacobus Postma, PhD1

1Unit of PharmacoEpidemiology & PharmacoEconomics (PE2), Department of Pharmacy, University of Groningen, Groningen, TheNetherlands; 2Faculty of Pharmacy, University of Muhammadiyah Purwokerto, Purwokerto, Indonesia; 3Faculty of Pharmacy, University of Padjadjaran, Bandung, Indonesia; 4Department of Medical Microbiology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands; 5Institute of Science in Healthy Aging & healthcaRE (SHARE), University Medical Center Groningen (UMCG), Groningen, The Netherlands

ABSTRACT

Background: Although cervical cancer is a preventable disease, the clinical and economic burdens of cervical cancer are still substantial issues in Indonesia. 

Objectives: The main purpose of this study was to model the costs, clinical benefits, and cost-utility of both visual inspection with acetic acid (VIA) screening alone and human papillomavirus (HPV) vaccination in addition to VIA screening in Indonesia. 

Methods: We developed a population-based Markov model, consisting of three health states (susceptible, cervical cancer, and death), to assess future costs, health effects, and the cost-utility of cervical cancer prevention strategies in Indonesia. We followed a cohort of 100,000 females 12 to 100 years old and compared VIA screening alone with the addition of HPV vaccination on top of the screening to “no intervention.” 

Results: The implementation of VIA screening alone and in combination with HPV vaccination would reduce the cervical cancer incidence by 7.9% and 58.5%, corresponding to 25 and 98 deaths avoided within the cohort of 100,000, respectively. We also estimated that HPV vaccination combined with VIA screening apparently yielded a lower incremental cost-effectiveness ratio at international dollar 1863/quality-adjusted life-year (QALY), compared with VIA screening alone (I$3126/QALY). Both strategies could how- ever be definitely labeled as very cost-effective interventions, based on a threshold suggested by the World Health Organization. The incremental cost-effectiveness ratio was sensitive to the discount rate, cervical cancer treatment costs, and quality of life as part of the QALY. Conclusions: The addition of HPV vaccination on top of VIA screening could be a cost-effective strategy in Indonesia even if relatively conservative assumptions are applied. This population- based model can be considered as an essential tool to inform decision makers on designing optimal strategies for cervical cancer prevention in Indonesia.

Keywords: Cervical cancer, cost-utility analysis, human papillomavirus,  Indonesia, vaccination.

Introduction

Cervical cancer is the second most common cancer among women in Indonesia. The age-standardized cervical cancer incidence rate and mortality rate per 100,000 women in 2012 was 17.3 and 8.1, respectively [1]. The long onset of cervical cancer development [2,3] enables the application of cervical screening to prevent and control cervical cancer. Despite the poor sensitivity of visual inspection with acetic acid (VIA) screening [4],it is the most commonly recommended screening strategy for countries with limited resources [5]. Well-organized VIA screening programs definitely decrease the burden of cervical cancer at relatively low costs [6,7]. The Ministry of Health, Republic of Indonesia launched a cervical cancer control program in 2007 and started a campaign recommending VIA screening for all susceptible women [8–10]. Yet, several studies reported various barriers to the implementation of this program, such as limited screening coverage, poor quality of services, and subsequent poor cryotherapy performance [11].\

In addition to the screening program, the introduction of prophylactic human papillomavirus (HPV) vaccination of girls against two high-risk HPV types (16 and 18) [12], which together are responsible for the large majority of cervical cancer development cases, offers primary prevention of cervical cancer [5,7]. There are two available HPV vaccines in the market, and their efficacies against HPV infections and cervical intraepithelial neoplasm have been demonstrated in numerous clinical trials [13–17]. Next to the efficacy and safety of vaccines, the available national budget for vaccination and affordability presents the other main consideration for a country to implement a vaccination program. Although the cost-effectiveness of HPV vaccination has been proven in many studies [18–22], those findings not necessarily apply to Indonesia because many differences in clinical profiles, patient and population characteristics, and health care systems among countries exist.

Although a new health insurance system has been implemented from 2014 onward in Indonesia, pharmacoeconomic studies have not yet been incorporated as a criterion into the decision-making process. However, cost-utility studies on cervical cancer prevention can provide valuable information for the decision maker to design the most cost-effective strategy to reduce the clinical and economic burdens of HPV-related disease among Indonesian women, within the limited budget. The main purpose of this study was to model the costs, clinical benefits, and cost-utility of both VIA screening alone and HPV vaccination in addition to VIA screening in Indonesia. To interpret the findings from this study, we applied the World Health Organization’s (WHO’s) threshold on cost-effectiveness of immunization programs [23,24].

Methods

Model Overview

We developed a population-based Markov model for Indonesia by using Microsofts Excel (Microsoft, Redmond, WA). The model (as shown in Fig. 1) consists of three health states (susceptible, cervical cancer, and death), which represent the major stages throughout the natural history of infection and cervical cancer. In our model, “susceptible for cancer” refers not only to healthy women but also to infected women with cervical intraepithelial neoplasm but (yet) without cancer. This simplification was made to accommodate with the limited data availability in Indonesia; notably, more complicated models would lack the data to populate them. In addition, in annual cycles, women may move through “cervical cancer” and “death” states. We hypothesized a cohort of 100,000 12-year-old girls before sexual debut as the initial situation in the model [25], followed until 100 years old. To estimate the natural history of cervical cancer, we applied the 2012 WHO’s life table on age-dependent incidence and mortality rate specific for cervical cancer in Indonesia [26,27]. The transition from cervical cancer to death resulted from death caused by cancer as well as by other diseases. We performed our analysis from the payer’s perspective, based on national tariffs that were recently launched by the Ministry of Health for all treatments in primary care and hospitals [28].

We compared three strategies in the base case: 1) without any intervention (reference), 2) with VIA screening, and 3) with VIA screening and HPV vaccination. Both unvaccinated and vaccinated groups were followed in the model with differing risks until the potential screening process. Cryotherapy treatment was assumed to be given among a part of positive individuals when precancer stages would be detected. Specific proportions of deaths, cancer cases, and recovered patients followed from VIA screening efficacies related to the prevention of cervical cancer [29]. In addition, we assumed that 15.8% of new patients with cervical cancer will have a recurrence and undergo an additional/ recurrence treatment [30]. The model parameters and baseline values specifically adopted for Indonesia are presented in Table 1.

Fig. 1 – Markov model for the development of cervical cancer.

Screening and Vaccination

Despite the extensive communication and the introduction of the national VIA screening program in 2007, the performance of this program remains sub optimal [11]. In our study, we assumed implementation of the screening for 30- to 60-year-old women within an annual interval of 3 years if the previous test result was negative, according to the recommendation [8,9]. We assumed that 63.6% of eligible (“susceptible” in the model) women would undergo VIA screening every 3 years [29]. We applied a detection rate of VIA screening of 69.4% [31] and an adherence rate to cryotherapy of 83.1% [11], based on previous studies in Indonesia. Furthermore, a study by Sankaranarayanan et al. found that the incidence ratio and the mortality hazard ratio for screened women was 0.75 and 0.65, respectively, compared with unscreened women [29].

The vaccine"s efficacy against HPV type 16 and 18 infection was estimated from available clinical trials [13,32,33], without taking cross-protection against HPV types other than 16 and 18 into account in the base case. The proportion of high-risk HPV was estimated from three studies in Indonesia [34–36]. Although the duration of immunity induced by vaccination is formally unknown, we assumed lifelong vaccine-induced protection in the base case as in other studies [37–39]. We also assumed that vaccination would be performed only at the start of the followed cohort (i.e., at age 12 years), with vaccination decreasing the transition probabilities from susceptible to cervical cancer. Vac- cine coverage was assumed to be 76.6% on the basis of school enrollment rates [40] and the coverage of other vaccinations (measles, diphtheria, and tetanus for 7–12-year-old girls) in Indonesia [41].

Costs and Utilities

In this study, all costs were converted to 2013 international dollars (I$), using purchasing power parities conversion factors [42]. With respect to the economic perspective, we considered only direct medical costs for cervical cancer treatment and all VIA screening–related activities according to the national tariffs for primary and secondary health care services [28]. Cervical cancer treatment costs (both initial and recurrent) were weighted by the cervical cancer treatment patterns for each stage of cervical cancer in Indonesia [34–36,43–46], and applied for every newly detected cervical cancer patient. The total cost for initial and recurrent treatments was I$4140 and I$3169, respectively. In the absence of a national vaccine price and availability of related relevant Indonesian information, we estimated all vaccine- related costs on the basis of Pan American Health Organization (PAHO) revolving fund, which consists of the price of a three-dose vaccination (I$39.71), revolving fund (I$1.39), freight (I$1.19), and insurance and wastage cost (I$1.99) [47]. Thus, our assumption for the total vaccination cost would be I$44.27.

We adopted utilities associated with patients with cervical cancer on the basis of Health and Activity Limitation Index [48], which allows the calculation of quality-adjusted life-years (QALYs) by taking utilities and durations of health states into account. Finally, we systematically applied an annual discount rate of 3% for both future costs and utilities.

Table 1 – Parameters used in the economic model: Base-case values and distributions applied in the probabilistic sensitivity analysis.

Parameters	Value	Distribution	References
Estimated proportion of HPV 16/18 in cervical cancer	75.40%	Triangular (71.0%; 75.4%; 100.0%)	[34–36]
Cervical cancer recurrence	15.80%	Triangular (3.4%; 15.8%; 23.2%)	[30]
HPV vaccination
Vaccine coverage	76.60%	Triangular (76.1%; 76.6%; 77.1%)
Vaccine price (I$)	14.76	–
Vaccine efficacy-perent reduction in HPV 16/18 persistent infection	95.00%	Triangular (90.4%; 95.0%; 98.1%)
Screening
Age range (y)	30-60	–
Coverage (3 yearly)	63.60%	Triangular (50.1%; 63.6%; 70.5%)
Efficacy to cervical cancer incidence	75.00%	Triangular (59.0%; 75.0%; 95.0%)
Efficacy to cervical cancer mortality	65.00%	Triangular (47.0%; 65.0%; 89.0%)
Detection rate (confirmed)	9.90%	Triangular (2.5%; 9.9%; 12.8%)
Cryotherapy coverage	83.10%	Triangular (13.0%; 83.1%; 100.0%)
Costs
Initial treatment of cervical cancer (I$)	4140	Triangular (2335; 4140; 5825)
Recurrence treatment of cervical cancer (I$)	3169	Triangular (1842; 3169; 4415)
Cryotherapy (I$)	26.29	–
Screening (I$)	4.38	–
Discount rate	3%	–
Utility
Susceptible	1	–
Cervical cancer	0.68	Triangular (0.48; 0.63; 0.84)
Death	0	–
Discount rate	3%	–
HPV, human papillomavirus.
* Estimated from school enrollment rate and vaccination coverage for measles, diphtheria, and tetanus for 7–12-y-old girls in Indonesia.
† Includes three doses of vaccine, revolving fund, freight, insurance, and wastage cost.
‡ Estimated from the national tariffs and weighted by the pattern of cervical cancer treatment in Indonesia.

Model Outcome

We critically addressed the estimated epidemiologic and economic outcomes from each strategy. Predicted epidemiologic outcomes were the number of both prevented cervical cancer cases and deaths. Furthermore, as an economic outcome, we estimated the incremental cost-effectiveness ratio (ICER) from the incremental costs divided by the incremental QALYs from preventive strategies, compared with no intervention. All out- comes were expressed for a cohort of 100,000 women through their lifetime in Indonesia.

Scenario and Sensitivity Analysis

We investigated the robustness of the ICERs by developing several scenarios with regard to booster dosing at age 30 years (scenario I) if the booster dose would be required to obtain lifelong effectiveness of the vaccine. We also investigated the effect of cross-protection against HPV types 31/33/45/52/58 at 25% efficacy (scenario II: low cross-protection) [33] and at 53% efficacy (scenario III: high cross-protection) [13].Also, we considered limited duration of vaccine-induced protection, specifically at 10 years (scenario IV: short protection) and at 20 years (scenario V: medium protection) and waning of vaccine-induced immunity at 95% efficacy for 10 years, followed by exponential decrease at 50% efficacy during each following period of 20 years (scenario VI: slow waning) or 5 years (scenario VII: fast waning).

We based the vaccine price on PAHO revolving fund for the base-case scenario [47]. In sensitivity analyses, we also explored potential reductions in the market price (75%, 50%, and 25% discounts on I$125.17) [50,51], both with and without booster dosing. Price reductions indicate potential advantages of economic up scaling and tendering effects if widespread vaccination would be considered to be implemented.

Univariate sensitivity analyses were performed by estimating the ICERs on the basis of changes in maximum and minimum values for each parameter and assumption, so as to investigate the most influential parameters or assumptions in the model. Parameters included in the univariate sensitivity analyses were vaccine efficacy, vaccine coverage, efficacy of VIA screening for cervical cancer incidence and mortality, screening coverage, utilities for patients with cervical cancer, and cryotheraphy coverage and its costs.

Probabilistic sensitivity analysis was taken into account by drawing one value for each parameter from its respective distributions simultaneously and estimating the ICER for each strategy correspondingly. We repeated this process up to 1000 times to provide a range for the ICER. We developed a cost- effectiveness acceptability curve to describe the relationship between potential Indonesian cost-effectiveness thresholds and the ICER, using the net monetary benefit approach. Based on the WHO’s criterion [24], a new intervention in Indonesia would be deemed very cost-effective and cost-effective if the ICER would be less than 1 time and 1 to 3 times the gross domestic product (GDP) per capita [52], respectively (2013 GDP per capita was I $3475).

Fig. 2 – Estimated annual cases of cervical cancer prevented (A) and life-years saved (B) by VIA screening, or VIA screening in combination with HPV vaccination. HPV, human papillomavirus; VIA, visual inspection with acetic acid.

Results

Clinical Outcomes

The projected annual reduction in cervical cancer cases and deaths as a consequence of VIA screening or in combination with HPV vaccination is presented in Fig. 2. Because the cervical cancer progression increases strongly after the age of 40 years, the effect of cervical screening is most evident in those particular ages. All susceptible women in the VIA screening group have an equal risk again on cervical cancer to that of unscreened women when the screening program stops after the age of 60 years. In contrast, women in the vaccination group remain protected by the effect of HPV vaccination until the end of the model analysis. Assuming a 3-year screening coverage of 63.6% [29], screening would reduce the total incidence of cervical cancer from 1842 cases to 1697 cases (7.9% reduction) compared with no intervention. In addition to the screening, the effectiveness of HPV vaccination in reducing the incidence of cervical cancer is high, as shown in Figure 2A. Specifically, it reduces the incidence of cervical cancer by up to 58.5% and 55.0% compared with no intervention and screening alone, respectively.

The effectiveness of VIA screening and HPV vaccination in reducing mortality would increase gradually after 30 years and attain a peak at 65 years after introduction. Fig. 2B shows that the addition of HPV vaccination on top of cervical cancer screening would prevent substantial mortality. Specifically, these strategies reduce cervical cancer–related death during lifetime by 24.58 and 97.49 cases per 100,000 women for screening alone and screening plus vaccination, respectively.

Costs, QALYs, and ICERs

Discounted costs and QALYs from each strategy are presented in Table 2. Discounted costs and QALYs from VIA screening combined with HPV vaccination (I$5,588,654 and I$2,724,504) are higher than discounted costs and QALYs from VIA screening alone (I$3,393,833 and I$2,723,129), both compared with no intervention. We also estimated that the ICER of VIA screening combined with HPV vaccination (I$1863) would be slightly lower than the ICER of VIA screening alone (I$3126). Apparently, based on PAHO revolving fund policy, both ICERs were still lower than the GDP per capita of Indonesia in 2013 (I$3475).

Table 2 – Discounted costs and QALYs and cost-effectiveness in the base case for VIA screening and vaccination in a cohort of 100,000 women followed from age 12 to 100 y.

Base case	Cost	QALYs	Incremental
			Cost	QALYs	ICER
No intervention	2,486,717	2,722,839	Reference	Reference	Reference
VIA screening	3,393,034	2,723,129	906,317	290	3126
VIA screening þ HPV vaccination	5,588,654	2,724,504	3,101,937	1665	1863
HPV, human papillomavirus; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; VIA, visual inspection with acetic acid.

Sensitivity Analysis

The impact of all scenarios on costs and QALYs is presented in Table 3. An addition of a booster dose to achieve lifelong protection has a limited effect on the ICER (I$3040). In addition, vaccine- induced cross-protection against type 31/33/45/52/58 would increase the ICER up to I$1716 and I$1570 for low (scenario II) and high cross-protection (scenario III), respectively. The duration of vaccine-induced protection and waning immunity has a significant effect on the ICER. Specifically, a short duration of vaccine- induced protection (scenario IV) affected the ICER strongly, raising it up to 5 times higher than the ICER in the base case (I$8795).

We also investigated the influence of the vaccine price (com- pared with the assumed market price) for both with and without booster dose scenarios (Fig. 3). The implementation of HPV vaccination on top of VIA screening in Indonesia would not be cost-effective under the normal market price of HPV vaccine (I $125.17) because the ICER would be far above the Indonesian cost-effectiveness threshold of I$10,425 per QALY gained (notably, I $17,106 per QALY without a booster dose and I$27,092 per QALY with a booster dose). If a booster dose is not required to obtain lifelong protection, a 50% reduction in the vaccine’s marketprice (I $62.59) would achieve the ICER (I$8466), being below the threshold. With the booster dose taken into account, a 75% reduction (I$31.29) keeps the ICER (I$6,642) below the threshold [24,52].

We performed a probabilistic sensitivity analysis by running a Monte-Carlo simulation to test the robustness of the model regarding the uncertainty surrounding the input parameters. A cost-effectiveness acceptability curve is presented in Fig. 4. Applying a threshold of 1 time the GDP (I$3475), the probability to be cost-effective would be 72.2% and 99.8% for VIA screening alone and VIA screening combined with HPV vaccination, respectively. In addition, the full range of simulations fell below I$7200/ QALY and I$3150/QALY for VIA screening alone and VIA screening combined with HPV vaccination, respectively.

We tested the influence of each parameter’s changes on the cost-effectiveness ratio in a univariate sensitivity analysis. A minor change in a very sensitive parameter that alters the ICER strongly would be found on the top in the tornado diagram. We see that the most sensitive parameters in the VIA screening strategy are the utilities, the discount rate, and cervical cancer treatment costs. In addition, the ICER was mildly sensitive to cryotherapy coverage, detection rate of screening, cost of recurrence, and VIA coverage (Fig. 5A). The most sensitive parameters in the HPV vaccination in addition to VIA screening strategy were discount rates, utilities, and cervical cancer treatment cost (Fig. 5B).

Discussion

We developed a population-based Markov model to determine the cost-utility of cervical cancer prevention programs in Indonesia, including VIA screening with or without HPV vaccination. Our study revealed that either screening alone or screening in combination with HPV vaccination can relevantly decrease the incidence of cervical cancer and improve quality of life and survival. Because most of the developing countries, including Indonesia, have no explicit cost-effectiveness criteria to justify the implementation of a new intervention, we applied the WHO’s recommendation on cost-effectiveness thresholds, stating that an intervention can be categorized as a cost-effective intervention if the ICER lies below 3 times the GDP per capita [24]. Because the GDP per capita of Indonesia in 2013 was approximately I$3475 [52], both VIA screening (I$3126) and VIA screening in combination with HPV vaccination (I$1863) compared with doing nothing can be considered as very cost-effective strategies. Specifically, the most cost-effective strategy is the combination of VIA screen- ing and HPV vaccination. To our knowledge, this is the first cost- utility analysis of cervical cancer prevention strategies in Indonesia. However, the result of this study, that HPV vaccination on top of cervical screening could be a cost-effective intervention, is in line with results of previous studies in other developing countries [53–58].

Table 3 – Discounted costs and QALYs and cost-effectiveness of various scenarios in a cohort of 100,000 women followed from age 12 to 100 y for vaccination in combination with VIA screening vs. natural progression of cervical cancer.

Scenario	Cost	QALYs	Incremental
			Cost	QALYs	ICER
Scenario I: Booster dose	7,548,979	2,724,504	5,062,262	1665	3040
Scenario II: Low cross-protection	5,506,716	2,724,599	3,019,999	1760	1716
Scenario III: High cross-protection	5,414,894	2,724,704	2,928,178	1865	1570
Scenario IV: Short vaccine-induced protection (10 y)	6,679,563	2,723,316	4,192,846	477	8795
Scenario V: Medium vaccine-induced protection (20 y)	6,603,330	2,723,449	4,116,613	610	6754
Scenario VI: Slow waning of immunity (after 10 y, efficacy decrease 50% every 20 y)	6,391,817	2,723,674	3,905,100	835	4678
Scenario VII: Fast waning of immunity (after 10 y, efficacy decrease 50% every 5 y)	6,630,473	2,723,395	4,143,756	556	7458
ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; VIA, visual inspection with acetic acid.

Fig. 3 – The effect of market vaccine price on ICER in terms of booster dose is needed (black square) and not needed (gray square) to achieve lifelong protection. ICER, incremental cost-effectiveness ratio; I$, international dollar.

 We selected the payer’s perspective for our study, which is inline with the new policy that has been implemented by the Indonesian government in early 2014 to cover almost all health care services in primary and secondary settings [28,59]. This method provides a clear picture of the average cost of cervical cancer treatment in Indonesia. Although the cost of cervical cancer treatment in this study is estimated as either lower [60] or higher [53,61] compared with that in other countries, the addition of HPV vaccination on top of VIA screening is considered as a cost-effective strategy as in those other countries.

Because the long-term efficacy of the current HPV vaccination has not been established, we investigated the possibility that a booster dose would be needed to achieve lifelong protection. As expected, the addition of a booster dose yielded a higher ICER, but the value itself remained below the GDP per capita of Indonesia. This finding is also in accordance with other studies in several settings [37,57,62]. Moreover, the vaccine’s effectiveness is influ- enced not only by the implementation of a booster dose but also by other variables such as vaccination coverage, the distribution of HPV types, and adherence [29,57].

In this study, the effect of cross-protection against HPV types 31/33/45/52/58 is limited, as illustrated by limited reduction from I$1630 in the base case to I$1488 and I$1346 for scenarios with low and high effect of cross-protection, respectively. This finding is similar to that from several studies from other countries that investigated the effect of cross-protection on cost-effectiveness [38,63]. Despite the fact that the distributions of HPV types in various countries are evidently different [34–36] and that this considerably influences the overall vaccine effectiveness from a clinical perspective, cross-protection against other high-risk HPV types can be highly interesting in other settings in Southeast Asia.

Fig. 4 – Cost-effectiveness acceptability curve, specifically 1 to 3 times the gross domestic product per capita is indicated with the corresponding probability to be cost-effective. HPV, human papillomavirus; I$, international dollar; QALY, quality- adjusted life-year; VIA, visual inspection with acetic acid.

Fig. 5 – Univariate sensitivity analyses for VIA screening alone (A) and VIA screening combined with vaccination (B) compared with no intervention. CC, cervical cancer; QALY, quality-adjusted life-year; VIA, visual inspection with acetic acid.

Based on a vaccine price derived from the PAHO revolving fund policy, the addition of vaccination yielded a cost-effective strategy in preventing cervical cancer. Yet, at the current market price of HPV vaccines, it appears that the addition of HPV vaccination to VIA screening is not a cost-effective intervention in Indonesia. Reduction in the range of 50% to 75% from the vaccine’s market price is required to maintain HPV vaccination in combination with VIA screening as a cost-effective strategy. This result suggests that a reduction in HPV vaccine price, compared with the market price, will be essential for the HPV vaccine to be included in the immunization schedule in Indonesia.

Notwithstanding the lack of data related to cervical intraepithelial neoplasm or precancer in Indonesia, our model can still be considered to validly and adequately estimate the natural history of patients with cervical cancer in Indonesia on the basis of actual epidemiological data from the WHO. For example, the natural history of patients with cervical cancer, cervical cancer incidence, and mortality rates for the population at risk could be described and implemented in the model. Notably, fewer assumptions were required in our model than in a more complex Markov structure or even dynamic model because we did not incorporate any transition to HPV infection or staging on pre- cancer and cancer stages. More complex modeling can be embarked upon if more data become available.

Despite the novelty of this study, it still has several limitations. First, we did not take the vaccine protection for low- risk HPV (type 6 and 11) into account. Although the data related to the effectiveness of both vaccines against other types of HPV are already available [64], the information related to the costs and QALYs caused by low-risk HPV in Indonesia is scarce. Second, incorporation of genital warts as a consequence of HPV types 6 and 11 also will introduce further differences in clinical benefits (i.e., QALYs) between both available vaccines in the market. However, to which extend this will be the case should be further investigated [65]. Therefore, further research should be directed at the clinical burden and costs of genital warts in Indonesia to make a more precise comparison between both vaccines. Another limitation in this study is the potential benefit of HPV vaccines against noncervical HPV-related cancers. Anal, vaginal, vulvar, and oropharingeal cancer, recurrent respiratory papilomatosis, and other precancerous lesions were not taken into account in the current model. The inclusion of these types of HPV-induced diseases will increase the savings and quality-of-life gains of HPV vaccination and consequently improve the cost-effectiveness of HPV vaccination [66–70].

Although we assumed 3-yearly screening in this model, the efficacy of the screening in preventing cervical cancer incidence and mortality is still considerably low. This can potentially be related to the fact that we did not incorporate the cumulative effect of repeated screening in the model [29]. Moreover, women who have negative results on their previous screening noticeably have a lower risk of cervical cancer incidence and mortality than do unscreened women. In this model we assumed that they have equal risk because the straightforward static Markov model has no ability to remember where the patient has come from nor the exact timing of that transition [71]. Yet, univariate sensitivity analysis showed that the influence of screening efficacy on the ICER is very low.

Notably, our results are consistent with results from previous studies from neighboring countries [20–22,38,54,57,58] by con- firming that HPV vaccination in addition to screening can be a cost-effective intervention if it can be obtained at a price similar to, for example, the PAHO price. This finding may encourage policymakers in Indonesia to further consider, decide, and implement optimal cervical cancer prevention strategies.

Conclusions

The addition of HPV vaccination on top of VIA screening in Indonesia, even in the context of various conservative assumptions (need a booster dose to obtain full protection, low cross- protection, short vaccine protection, and fast waning immunity), is a very cost-effective strategy. Substantial clinical and economic benefits can be obtained by implementing an HPV vaccination program. Nevertheless, improvement of the screening program itself also remains important and provides further potential to achieve optimal cervical cancer prevention strategies.

Acknowledgment

We thank Dr. Agusdini Banun Septaningsih from “Dharmais” National Cancer Hospital, Indonesia, for her advice on unit cost estimates in Indonesia.

Source of financial support: This work was supported by the Directorate General of Higher Education (DIKTI) Scholarship, Ministry of National Education, Indonesia. The authors’ work was independent of the funders, who had no role in the study design, analysis of data, writing of the manuscript, or the decision to submit for publication.