Mobile Screening Units for the Early Detection of Cancer: A Systematic Review

This article is featured in Highlights of This Issue, p. 1677

The implementation of screening and early detection programs is a cornerstone of cancer prevention (1, 2). Despite evidence that early detection saves lives, global disparities in access to services persist (1). Barriers to cancer screening include lack of provider availability, community access to screening and community demand for screening (3). Challenges related to health care providers can include shortages in screening delivery or policies that require a referral from a family physician to access screening. Community access to screening can be limited by prohibitive costs or inaccessibility of screening clinics due to distance, hours of operation, or a lack of knowledge about where to go for screening. Finally, client demand for screening can pose as a barrier in situations where individuals are unaware of the benefits of screening, do not perceive themselves at risk, or fear screening results.

Mobile screening units (MSUs) are an innovative alternative to screening exams in clinics or hospitals and may include vans, recreational vehicles, or other traveling clinics that are staffed by health workers and outfitted with equipment for early detection. MSUs allow care providers to increase their capacity for service delivery outside of fixed clinics, which is particularly important in areas without an infrastructure for cancer screening services (4). MSUs increase community access by offering screening in convenient locations thus decreasing the distance and travel time needed to access screening services (5). MSUs also present unique challenges for service delivery, such as more complicated referral systems as individuals with abnormal exams commonly need to report to a specialty clinic for follow-up.

Previous narrative literature reviews of community-based screening initiatives have concluded that eliminating structural barriers such as location, distance, and inconvenient hours of operation is highly effective in increasing screening in high-income countries (3, 5). In 1995, a National Cancer Institute survey of 1,057 mammogram facilities in the United States showed that 2.4% of screening services were offered in MSUs, which demonstrated similar quality indicators as fixed clinics with respect to certification of personnel, quality control practices, and frequency of equipment verification by medical physicists. In addition, the NCI survey found MSUs benefitted clients by offering lower fees, accessible evening operating hours, and serving a higher proportion of self-referred women (4). A survey on barriers to breast and cervical cancer screening among underserved populations in the United States found MSUs to be an effective strategy for reaching women over the age of 60 years, some of whom may be unable to travel to fixed clinics (6). Overall, current evidence to support the operations of MSU programs is limited, with summaries available only through dated reviews (6–8) and informal medical forums (9). In addition, none of the prior reviews included data on performance measures, such as recall rates and case detection rates.

To our knowledge, no systematic review has synthesized results on the implementation of MSU programs. This systematic review summarizes the evidence from qualitative, quantitative, and mixed-method studies on the implementation of MSUs for early cancer detection worldwide.

We adhered to the PRISMA guidelines and followed a review protocol (10). Empirical studies that included a description of cancer screening for breast, cervical, and colorectal cancer taking place on MSUs were eligible for inclusion. No restrictions were placed on study design, study location, and participants. Studies reporting on screening tests to detect precancerous lesions or early invasive cancer were eligible. Both organized and opportunistic screening programs were included. Only studies published in English, French, or Portuguese were considered. Studies were excluded if data on the MSU program were not reported separately from those of a fixed clinic.

The following databases were searched through to July 2015: EMBASE (via Ovid), MEDLINE and PubMed (via Ovid), Cochrane Library, WHO Global Health Library, Science Citation Index Expanded (via Web of Science), Social Sciences Citation Index (via Web of Science), and PsycINFO (via Ovid). Additional studies were identified through bibliographies of included studies and consulting subject matter experts. The search strategy was developed by a medical librarian and used the following search terms to query databases: mobile, movable, traveling, unit, clinic, facility, hospital, early detection, neoplasms, cancer screening, cancer prevention, and cancer diagnosis. A full search strategy is included in Supplementary Table S1. Only published articles in peer-reviewed journals or thesis dissertations were included. Conference abstracts/posters, review articles, and opinion pieces were excluded.

We first reviewed titles and abstracts of potentially relevant citations. The full text of eligible articles was then screened by one author and verified by a second. Disagreements between authors were resolved by discussion or consultation with a third author. Reasons for the exclusion of full-text articles were recorded.

Two authors then independently reviewed full-text articles. Data were extracted onto a piloted comprehensive and standardized data abstraction sheet by one author and validated by a second author. The following information was extracted: study design, country, cancer site(s), screening test(s), size of target population, number of screening exams and/or number of individuals screened, participant sociodemographic characteristics, type of MSU, operational characteristics of MSU (with respect to the number of screening tests offered, daily kilometer distance traveled, locations of follow-up facilities, etc.) and performance measures of screening including the rate of abnormal calls, biopsy rate, losses to follow-up, timeliness of follow-up, number of detected cases, and characteristics of detected cases including staging and extent of disease. Test validity measures including sensitivity, specificity, positive predictive value, and negative predictive value were extracted when reported or otherwise manually calculated (formulas provided in Supplementary Table S2). We considered screening programs to be organized if they were part of a systematic program where every individual in a target population was invited to attend screening.

We collected economic data when reported; however, no analyses were conducted due to the broad time range of included studies and nontransferability of costs of medical equipment and infrastructure from past to present. If key data were missing from studies, we contacted the investigators of original studies published within the last 10 years to obtain the information. Where multiple reports of the same study were published, we extracted data from the most recent or comprehensive publication.

The risk of bias for a subset of individual studies reporting detected cancer cases was independently appraised by two authors using the Mixed Methods Appraisal Tool (MMAT; ref. 11) provided in Supplementary Table S3. This tool was selected for its capacity to appraise quantitative, qualitative, and mixed-method studies. The content validity and reliability of the MMAT for judging methodologic quality has been previously validated (12). We rated each study according to four criteria via a score ranging from 100% (all criteria met) to 0% (no criteria met). MMAT methodologic criteria indicators vary by study design to assess systematic error such as sampling, selection, confounding, and information biases. Conflicts were resolved by consensus or consultation with a third author. We modified the tool to include a criterion to assess the methodologic quality of screening programs for studies using a quantitative descriptive design without comparison group. The modified criterion assessed whether the rate of follow-up among abnormal screening exams was above 80%. We used an 80% threshold as this has been previously established by WHO early detection programme guidelines as a minimum acceptable rate of follow-up (2). In addition, we hypothesized that loss to follow-up among abnormal screening exams would introduce information bias into studies, leading to an underestimation of case detection rates.

Because of the heterogeneity in study endpoints and the presence of noncomparative study designs, the studies were not amenable to statistical analyses and no meta-analysis was possible. Studies were thus summarized qualitatively in terms of key study characteristics, such as the number of screening exams performed on MSUs, and duration of program operations, and performance outcomes including the rate of abnormal calls, biopsy rate, number of detected cases, and characteristics of detected cases, without an overall summary effect measure. Outcomes were graphically displayed for continuous data, such as the performance of programs in terms of case detection rates per 1,000 screening exams.

The implementation of the MSU programs are described with respect to three evidence-based factors associated with successful delivery of population-based screening services: provider delivery, community access, and community demand (3). We also summarized available information on the referral systems of MSU programs and reported rates of loss to follow-up following abnormal exams.

The initial search yielded 441 articles. An additional 27 records were identified through other sources. After checking for duplicates, there were 268 unique hits. Careful analysis of titles and abstracts resulted in 140 eligible articles. Of these, 78 studies met the inclusion criteria and were included in the review (Fig. 1; Table 1). Twelve studies were excluded due to screening for sites of cancer for which current evidence does not support efficacy of screening among the general population, such as for skin cancer (n = 5), lung cancer (n = 3), gastric cancer (n = 2), oral cancer (n = 1) and prostate cancer (n = 1). Key operational aspects of these MSU studies are summarized in Supplementary Table S4.

The majority of MSU interventions (n = 55) targeted breast cancer while other studies involved cervical cancer (n = 12), colon cancer (n = 1), or multiphasic screening for multiple cancer sites (n = 10). A summary of operational measures by site of cancer screening is presented in Table 2. Studies were published between 1970 and 2016 and were conducted in 20 countries, with the majority from Europe (36%), North America (36%), Latin America (13%), and Asia (11%); 52% of MSUs offered services in both rural and urban areas, while 35% of programs exclusively served rural regions and 13% exclusively urban areas. Overall, 28% of studies were part of organized screening programs.

Study designs, based on the classifications for mixed study designs (11), are reported in Table 1. The majority were of a quantitative descriptive nature (64% including 50 single-group studies of cancer incidence/prevalence or operational measures and three studies of cost-effectiveness); a further 22% were nonrandomized quantitative studies (including three prospective and four retrospective cohort studies, and 10 cross-sectional studies comparing outcomes between two groups); 6% were randomized control trials (RCT; n = 5) and 8% (n = 6) were qualitative descriptive or case studies.

As shown in Table 3, 32 studies reported performance measures. The quantitative descriptive studies (n = 22) measured various implementation, clinical and service-related outcomes, and generally explicitly or implicitly stated the objective of evaluating the performance of the MSU program. The nonrandomized observational studies (n = 7) sought to answer more specific questions, often with respect to comparisons in screening modalities such as the effectiveness of MSU as compared with fixed clinic screening (13–16); breast cancer studies comparing digital to film-screen mammography (17, 18); cervical cancer studies comparing conventional Pap cytology with liquid-based Pap cytology (Pap-LBC) tests (19, 20) or human papillomavirus (HPV) testing with Pap-LBC (21). Two RCTs for mammography screening sought to establish evidence of mortality reductions associated with screening (22, 23).

The mean overall score was 74.2% for the 29 studies assessed for methodologic quality (Table 3), indicating a fair level of methodologic quality across studies. Trends in the risk of bias by study design are presented in Fig. 2. The overall quality score was 72% for the 22 descriptive studies (Fig. 2A) and 82% for the seven nonrandomized observational studies (Fig. 2B). A common weakness was incompleteness in reporting outcomes, with 50% of the descriptive studies and 29% of nonrandomized observational studies providing no information on follow-up rates. Two RCTs for breast cancer received scores of 100% (22) and 50% (23).

Fifty-five studies described breast cancer screening using a range of tests including digital or analog film-screen mammography and clinical breast exams. In addition, all 10 multiphasic programs offered breast screening. The number of mammography screening exams reported per study ranged from 141 to 214,327, and study duration ranged up to 20 years; this indicates the flexibility of MSUs to be used at various capacities in different contexts. As displayed in Fig. 3, among 22 breast cancer screening studies reporting performance measures, case detection rates ranged from 2.15 to 14.65 per 1,000 mammography exams. In these studies, patients with abnormal screening results were recalled to fixed clinics for further clinical work-up and diagnostic exams. The broad range in detection rates may be due to differences in contextual factors between programs such as the ages of screened women, proportion of never-screened women, or differences in incidence rates of breast cancer in various settings.

Twelve studies described cervical cancer MSU programs involving screening exams performed by nurse technicians including: conventional or liquid-based Pap cytology, HPV testing, and visual inspection with acetic acid (VIA); one MSU also offered diagnostic colposcopy exam performed by a clinician (15). All 10 multiphasic programs reported screening for cervical cancer. Notably, many of the cervical cancer screening studies were in low-resource settings in Brazil, Peru, South Africa, India, Thailand, and Taiwan (15, 20, 21, 24–27). The studies report diverse outcomes of detection of precancerous cervical lesions and invasive cancer, as described in Table 3. Because of the inconsistency of reporting measures used, no summary figure of case detection is presented for the cervical cancer screening studies.

Examples of operations of MSUs in regions distant from centralized services include cervical cancer screening in tents in the Andes Mountains of Peru (26); a mammography unit transported by boat and plane to reach Arctic regions in Canada (14); an MSU mammography program serving residents of 26 inhabited islands in Scotland (28); a semi-mobile colonoscopy program for high-risk individuals in South Africa residing over 500 km away from the nearest hospital (29); and MSU interventions for breast and cervical cancer in the rural interior and Amazonian states of Brazil (21, 30). In addition, MSUs fill a unique niche in urban areas by targeting underscreened groups who may be within a reasonable physical proximity to a fixed clinic, but less likely to access services due costs, awareness, and perceived barriers (26, 31–37).

Key factors to the accessibility of services are the time and distance traveled to a screening center. Some studies reported patients' travel to the MSU location to be fairly minimal, with average distances traveled by clients to reach MSU sites of 4.5–34 km (38–40). One study from the United Kingdom found MSUs were located substantially closer to patients than fixed clinics; the mean distance per client to an MSU was 6.4 km as compared with fixed clinics that were on average 18.2 km away (41).

Several MSU programs aimed to increase community access to screening (24, 34, 42–48). An increased screening rate from 44.3% to 63.4% was reported in regions served exclusively by MSUs in Quebec, Canada over a course of eight years (14). A Thai study found an increase in the proportion of women reporting ever receiving a Pap smear from 20% in 1991, to 70% in 1999 due to MSU operations (20).

In cases where before and after measures of screening uptake were unavailable, using the proportion of never-screened individuals who received their first screening exam through the program can serve as a surrogate measure for increased access. Among breast screening programs, studies reported the proportion of first-screen clients to range from 11% to 72% (18, 30, 36). Cervical cancer screening programs targeting low-income groups reached particularly underserved populations composed of 75% never-screened clients in South Africa (15), and 70% never-screened clients in the United Kingdom (49, 50).

MSU programs have addressed economic barriers to screening by reaching underserved populations. Notably, 31% (n = 24) of studies offered screening services free of charge. The majority of studies that discussed fee-for-service economic barriers to screening were based in the United States where no organized breast cancer screening program existed and attending screening was likely to be affected by insurance status (17, 34, 35, 42–44, 47, 48, 51–53). Furthermore, a number of MSU programs have decreased structural barriers by offering screening at the workplace (49, 54–58), shopping centers (50), and retirement homes (59).

Client demand for screening is an essential part of attaining population-level screening coverage. MSU qualitative studies used focus groups and surveys to identify women's attitudes towards screening, their perception of the utility of screening and intent to screen (28, 38, 39, 46, 60–64). Among women surveyed following MSU screening, nonadherence to screening guidelines was reported as related to a lack of awareness of early detection, lack of accessible information, negative beliefs or attitudes toward screening, as well as fear of discomfort/embarrassment, cost, and bad news (34, 61, 63).

Studies reported high levels of overall client satisfaction with the quality of service (24, 62, 64) and adequate levels of comfort and privacy with screening (24, 62, 65). Two studies measured patient preferences and documented preference for MSU relative to fixed-site clinic (24, 62). In contrast, one study reported a lower client satisfaction for MSUs relative to hospital screening in terms of receipt of test results, staff interpersonal skills, privacy, physical surroundings, and general satisfaction (64).

Studies described various methods to promote screening attendance, including direct invitation by mail or telephone, publicity (flyers, radio, car loudspeaker, newspaper), word of mouth, or physician referral (32, 38, 49, 62, 66–69). A study conducted during the initial implementation of a breast and cervical cancer screening program in Brazil found that among various promotion strategies, community health care agents making home visits was the most effective means of recruiting women to begin screening (70).

Compliance to the recommended follow-up of an abnormal screening exam serves as an indicator of the quality of a screening program. Rates of loss to follow-up generally tended to be higher in low and middle-income countries. A 21% loss to follow-up rate of the MSU mammography screening program in Mexico City (71) and a 66% loss to follow-up for colposcopy referrals in the MSU cervical cancer screening program in Brazil (21) were found compared with a 10%–15% loss to follow-up in the MSU breast screening programs in France (72), Greece (73), and Belgium (74). MSU operations should strive to minimize potential delays in the referral pathway and loss to follow-up that may occur due to a delay between the screening test and dissemination of results/availability of treatment. A cervical cancer screening study in South Africa assessed the effect of a “see, screen, treat” intervention to minimize loss to follow-up and found improved rates of completed follow-up when clients with suspected lesions were referred to the MSU for diagnostic colposcopy and lesion excision (15). Ninety-eight percent of patients with high-grade squamous intraepithelial lesions (HSIL) attended MSU colposcopy as compared with 33% complete follow-up when referred for colposcopy at the nearest hospital, which was 30 km away from the MSU site.

This review summarizes the evidence regarding the feasibility of MSU interventions for the early detection of cancer. We described existing programs, their performance, and the mechanisms through which they can increase the delivery, access, and demand for screening services. It was evidenced that MSUs can increase access for underscreened groups. Offering early detection services through MSUs is a practical way for service providers to increase physical and economic access to screening while reducing barriers for clients (structural barriers and out-of-pocket costs).

However, there are several challenges and risks involved with offering screening via MSUs as compared with fixed clinics. Referral systems and challenges with follow-up are a major issue. In an MSU setting, the exam is generally conducted by a nurse technician on the MSU and any abnormalities that are found upon interpreting the exam, will require the patient to travel to a fixed clinic to visit a physician on a separate date, whereas a fixed clinic may have the flexibility to conduct a diagnostic exam on the same day and minimize potential losses to follow-up. In addition, MSUs face unique risks when compared with fixed clinics—poor roads may necessitate the recalibration of sensitive diagnostic machinery if transported by MSU. Finally, client perceptions of MSUs may differ from fixed-clinics, as MSUs are generally perceived as less clinical than hospital-based clinics. This review did not focus on the clinical considerations of screening tests with respect to their effectiveness in reducing morbidity or mortality for cancer sites. However this serves as an area for future study to determine whether MSUs are as clinically effective as fixed facilities.

The body of evidence summarized by these studies is modest. The study designs were largely descriptive single-group studies, with few high-quality observational studies and RCTs documenting the performance of MSUs, the latter type representing the highest level of evidence, followed by observational studies and descriptive studies (75). Among studies that were appraised for methodologic quality, we found a low risk of bias for measures including sampling, selection bias, and valid measurements. However, several studies were methodologically weak and at high risk of bias for the incomplete reporting of outcome measures. Therefore, although the overall methodologic quality of the studies summarized by our review was at a moderately low risk of bias, the strength of evidence continues to be limited. A weakness of this quality review is that some studies scored decently on the MMAT scale, despite shortcomings in design and reporting. We found the MMAT scale was restrictive with respect to scoring the methodologic quality of screening studies. The design and methods employed by future studies of MSUs should pay particular attention to reporting of complete outcomes.

There are potential risks associated with screening in general, particularly if the test and the medical procedures that are required in consequence are inefficiently administered. This includes the potential for missing cancers (false negatives) and patients undergoing unnecessary clinical follow-up if they test positive but do not have the disease (false positives). Regrettably, risks of screening may be worsened in settings where follow-up is limited, because individuals with abnormal tests may not return and the period between screening rounds may extend beyond the recommended intervals, thus permitting development of interval cancers. Few studies reported the potential for false negative test results which can be measured as the number of interval cancers occurring within 24 months of a negative test result or by reporting test specificity (13, 21, 72, 76). Using a descriptive single-group study design measuring rates of cancer detection at baseline (cross-sectional), and without follow-up data, it is not possible to assess the rates of false-negative results. With respect to prevalence of false positives, due to the wide ranges of reported positive predictive values of tests and the lack of reporting on confounding characteristics in the study population (most notably age) it was not possible to summarize rates of false positives across studies.

The findings of individual studies have limited generalizability. The quantitative findings in terms of test performance cannot be directly compared between studies or generalized to other settings due to differences in baseline risk of cancer and the non-reporting of age-standardized detection rates. The qualitative outcomes such as the acceptability of a given cancer screening test and client satisfaction levels with MSU services are also highly contextual, and cannot be generalized across studies.

Insufficient evidence was available to describe the cost-effectiveness of MSU interventions but we hypothesize that MSUs have lower start-up costs compared with fixed sites but may face greater maintenance costs for technology and equipment due to stress caused by road travel. Conducting studies to investigate the cost-effectiveness of MSU programs relative to fixed clinics should be a priority.

The MSU studies included in this review were not restricted by study design, types of screening tests used, or types of outcomes reported. This broad inclusion criterion is both a strength and a limitation of this review. Studying multiple MSU interventions emphasized the versatility of MSU programs but limits the depth of analysis for performance outcomes of individual programs. The literature search was limited to programs that explicitly described the methods of screening as mobile or used a synonym listed within the search (Supplementary Table S1). The search was designed to be as robust and detailed as possible. However, it is possible that some studies were overlooked. Publication bias may have led to nonpublication of MSU studies of high and poor quality that do not report their findings in peer-reviewed journals. This review was only able to synthesize published results.

This review synthesizes the available knowledge regarding MSU interventions for cancer screening. MSUs can be an important health system component to expand screening coverage to rural areas or to urban residents who fail to be reached by health promotion messages or lack medical insurance. Overall, our findings highlight the value of MSU interventions, specifically with respect to increasing access to early detection services. More evidence is needed to provide comparative information on the cost-effectiveness and effectiveness of MSU interventions as compared with fixed-clinics to inform decision making for health care providers or policymakers.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.