Using comparative effectiveness analyses to optimize cervical cancer screening

描述(由申请人提供)：随着致癌人乳头瘤病毒(HPV)类型测试和特定类型HPV疫苗的引入，美国预防宫颈癌的潜在战略在过去十年中得到了极大的扩展。多种选择带来了复杂性，确定如何最好地最大化筛查好处和最大限度地减少危害(包括资源投入)已成为一个巨大的挑战。例如，将HPV DNA检测与细胞学进行比较的随机试验的初步结果表明，HPV检测比细胞学具有更高的敏感性(检测出更多的宫颈肿瘤病例)，而特异性更低(至少是阳性检测的两倍)。鉴于大多数宫颈癌发生在从未接受过或未充分筛查的妇女中，与每年约8000万处于危险之中的妇女相比，接受筛查的美国妇女潜在的可预防负担相对较小(n~5000)。虽然筛查的好处可以通过在一生中以越来越频繁的间隔进行更敏感的测试来最大化，但随着这种策略的实施，危害大大增加，特别是在健康女性中。衡量危害具有挑战性，因为它们可以采取多种形式，包括假阳性测试，这会导致不必要的干预，伴随着副作用、生活中断和与健康相关的生活质量的其他潜在下降。为了充分了解筛查危害的程度和影响，需要了解妇女的观点和偏好。最大化收益和最小化危害的一种方法是通过个体风险因素评估进行个性化筛查。例如，接种HPV疫苗似乎降低了宫颈肿瘤的风险；因此，继续以与未接种疫苗妇女相同的警惕对妇女进行筛查可能会加剧筛查危害。此外，十多年来一直建议对免疫功能低下的妇女(如艾滋病毒感染)进行年度筛查，尽管尚不清楚这种方法是否适当地平衡了益处和危害。如果我们仅仅依靠随机试验，确定如何以最佳方式使用新的检测方法(单独或联合使用)，以及是否应该在妇女亚组中考虑“个性化”方法(例如，接种疫苗、免疫功能低下)，即使不是不可能，也仍然是困难的。另一种方法是使用决策分析模型和比较有效性分析来确定提供类似益处和危害(或改进的益处/危害平衡)的新策略。为宫颈癌筛查定义一系列合理的选项，以及这些选项可能因个别风险因素而变化，对于医学研究所定义的比较有效性研究的核心目标来说，立即是有用和综合的：确定哪种策略对谁最有效，在什么情况下。目前的研究将使用最先进的方法来解决宫颈癌预防的这个问题。

DESCRIPTION (provided by applicant): With the introduction of tests for oncogenic human papillomavirus (HPV) types and type-specific HPV vaccines, potential strategies for cervical cancer prevention in the US have expanded tremendously over the last decade. With multiple options has come complexity, and determining how best to maximize screening benefits and minimize harms (including resource inputs) has become a great challenge. As an example, preliminary results from randomized trials comparing HPV DNA tests to cytology indicate that HPV testing has a higher sensitivity (detects more cases of cervical neoplasia) than cytology and a lower specificity (at least doubling the number of positive tests). Given that most cervical cancer occurs among never- and inadequately-screened women, the potential preventable burden among screened US women is relatively small (n~5000) compared to the approximately 80 million women at risk per year. While screening benefits can be maximized by apply more sensitive tests at increasingly frequent intervals over a lifetime, harms increase substantially with such strategies, especially among healthy women. Measuring harms is challenging since they can take many forms including false-positive tests that lead to unnecessary interventions with concomitant side effects, life disruptions and other potential decrements in health-related quality of life. To fully capture the magnitude and effect of screening harms, the perspectives and preferences of women are needed. One approach to maximizing benefits and minimizing harms is to personalize screening by individual risk factor assessment. For example, HPV vaccination appears to lower risk of cervical neoplasia; continuing to screen vaccinated women with the same vigilance as unvaccinated women, therefore, may exacerbate screening harms. Further, annual screening of immunocompromised women (e.g., HIV infection) has been recommended for over a decade, though it is unclear if such an approach appropriately balances benefits and harms. Determining how to optimally use new tests (singly or in combination) and whether "personalized" approaches should be considered in subgroups of women (e.g., vaccinated, immunocompromised) will remain difficult, if not impossible, if we are to rely solely on randomized trials. An alternative approach is using decision analytic models and comparative effectiveness analyses to identify novel strategies that provide similar benefits and harms (or an improved benefit/harm balance). Defining a "range of reasonable options" for cervical cancer screening and how these might vary by individual risk factors, would be immediately useful and synthetic to the core goal of comparative effectiveness research as defined by the Institute of Medicine: determining which strategy works best, for whom, and under what circumstances. The current study will address this question for cervical cancer prevention using state-of-the-art methodology.