Screening for Intimate Partner Violence, Elder Abuse, and Abuse of Vulnerable Adults: Evidence Report and Systematic Review for the US Preventive Services Task Force

Importance  Intimate partner violence (IPV), elder abuse, and abuse of vulnerable adults are common and result in adverse health outcomes.

Objective  To review the evidence on screening and interventions for IPV, elder abuse, and abuse of vulnerable adults to inform the US Preventive Services Task Force.

Data Sources  MEDLINE, Cochrane Library, EMBASE, and trial registries through October 4, 2017; references; experts; literature surveillance through August 1, 2018.

Study Selection  English-language randomized clinical trials (RCTs), studies evaluating test accuracy, and cohort studies with a concurrent control group assessing harms.

Data Extraction and Synthesis  Dual review of titles and abstracts, full-text articles, and study quality; qualitative synthesis of findings. Data were not pooled, primarily because of heterogeneity of populations, interventions, and outcomes.

Main Outcomes and Measures  Abuse or neglect, morbidity caused by abuse, test accuracy, and harms.

Results  Thirty studies were included (N = 14 959). Three RCTs (n = 3759) compared IPV screening with no screening; none found significant improvements in outcomes (eg, IPV or quality of life) over 3 to 18 months and 2 (n = 935) reported no harms of screening. Nine studies assessed tools to detect any past-year or current IPV in women; for past-year IPV (5 studies [n = 6331]), sensitivity of 5 tools ranged from 65% to 87% and specificity ranged from 80% to 95%. The accuracy of 5 tools (4 studies [n = 1795]) for detecting current abuse varied widely; sensitivity ranged from 46% to 94% and specificity ranged from 38% to 95%. Eleven RCTs (n = 6740) evaluated interventions for women with screen-detected IPV. Two enrolling pregnant women (n = 575) found significantly less IPV among women in the intervention group: 1 home visiting intervention (standardized mean difference [SMD], −0.34 [95% CI, −0.59 to −0.08]) and 1 behavioral counseling intervention for multiple risks (IPV, smoking, depression, tobacco exposure) (SMD, −0.40 [95% CI, −0.68 to −0.12]). No studies evaluated screening or interventions for elder abuse or abuse of vulnerable adults. One study assessing a screening tool for elder abuse had poor accuracy (sensitivity, 46% and specificity, 73% for detecting physical or verbal abuse).

Conclusions and Relevance  Although available screening tools may reasonably identify women experiencing IPV, trials of IPV screening in adult women did not show a reduction in IPV or improvement in quality of life over 3 to 18 months. Limited evidence suggested that home visiting and behavioral counseling interventions that address multiple risk factors may lead to reduced IPV among pregnant or postpartum women. No studies assessed screening or treatment for elder abuse and abuse of vulnerable adults.

Intimate partner violence (IPV), elder abuse, and abuse of vulnerable adults can cause acute and long-term adverse physical and mental health as well as adverse social consequences (eg, homelessness).¹ IPV refers to physical or sexual violence, psychological aggression, or stalking by a person with whom one has a close personal relationship,² such as a spouse. The 2015 National Intimate Partner and Sexual Violence Survey (N = 10 081) estimated that 5.4% of women and 5.1% of men experienced any past-year contact sexual violence, physical violence, and/or stalking.³ Elder abuse refers to an intentional act or failure to act by a caregiver or another person in a relationship involving an expectation of trust that causes or creates a serious risk of harm to an older adult⁴; similar criteria apply to abuse of vulnerable adults (those with impaired ability to perform normal activities of daily living or to provide for their own care because of physical or mental disability).⁵ In a 2008 nationwide telephone survey of older adults (N = 5777), 10% of respondents reported past-year emotional, physical, or sexual mistreatment or potential neglect.⁶

Routine screening of people without signs or symptoms of abuse could identify abuse not otherwise disclosed and provide opportunities for intervention that may reduce future abuse as well as short- and long-term adverse health consequences. In 2013, the US Preventive Services Task Force (USPSTF) recommended screening women of childbearing age for IPV but concluded that the evidence was insufficient for older or vulnerable adults.⁷ To inform an updated recommendation, the evidence on the benefits and harms of screening asymptomatic adults for IPV, elder abuse, and abuse of vulnerable adults in populations and settings relevant to US primary care was reviewed.

Detailed methods are available in the full evidence report at http://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/intimate-partner-violence-and-abuse-of-elderly-and-vulnerable-adults-screening1. Figure 1 shows the analytic framework and key questions (KQs) for IPV that guided the review; a similar figure for elder abuse and abuse of vulnerable adults is provided as eFigure 1 in the Supplement.

PubMed/MEDLINE, the Cochrane Library, and EMBASE were searched for English-language articles from 2011 through October 4, 2017 (eMethods in the Supplement). Studies published before 2011 were identified from the prior systematic reviews for the USPSTF.^9,10 For IPV screening and treatment in men and adolescents (not included in the prior reviews), searches were conducted from database inception through October 4, 2017. ClinicalTrials.gov, the National Institutes of Health Research Portfolio Online Report Tools, and the World Health Organization International Clinical Trials Registry Platform were also searched. To supplement searches, investigators reviewed reference lists of pertinent articles and studies suggested by reviewers. Since October 2017, ongoing surveillance was conducted through article alerts and targeted searches of journals to identify major studies published in the interim that may affect the conclusions or understanding of the evidence and the related USPSTF recommendation. The last surveillance was conducted on August 1, 2018.

Two investigators independently reviewed titles, abstracts, and full-text articles using prespecified eligibility criteria (eTable 1 in the Supplement). English-language studies conducted in countries categorized as “very high” on the United Nations Human Development Index were included. Only studies rated as good or fair quality were included.

Randomized clinical trials (RCTs) comparing screened groups with unscreened groups were eligible for KQ1 (direct evidence that screening improves health outcomes). Eligible outcomes for KQ1 included abuse or neglect victimization, health outcomes, health care utilization attributed to abuse, quality of life, and mortality.

Studies assessing the accuracy of tools designed to detect current, past, or risk of abuse were eligible for KQ2 (screening test accuracy). Only studies that compared a screening tool with an acceptable reference standard, such as the Conflicts Tactics Scale (CTS), were eligible.

RCTs and cohort studies with a concurrent control group comparing screened groups with unscreened groups were eligible for KQ3 (harms of screening). Eligible harm outcomes included labeling, stigma, false-positive results, increased abuse and retaliation, and others.

RCTs assessing interventions that could be offered in primary care or referred to by primary care were eligible for KQ4 (benefits of interventions) and KQ5 (harms of intervention). RCTs comparing intervention groups with no treatment, usual care, attention control, or waitlist control were eligible. Cohort studies with a concurrent control group were also eligible for KQ5 (harms of interventions).

For each included study, 1 investigator extracted information about populations, tests or interventions, comparators, outcomes, settings, and designs, and a second investigator reviewed for completeness and accuracy. Two independent investigators assessed the quality of each study as good, fair, or poor using predefined criteria developed by the USPSTF and adapted for this topic (eMethods in the Supplement).¹¹ Individual study quality ratings are provided in eTables 2 through 11 in the Supplement.

Findings for each question were summarized in tables, figures, and narrative format. For KQ4 (benefits of IPV interventions), studies reported on similar outcomes (eg, incident IPV) using both continuous and dichotomous measures. To create figures displaying commonly reported outcomes, the results were reexpressed as a standardized mean difference when sufficient data were available. Statistical significance was assumed when 95% CIs did not cross the null. All testing was 2-sided. Meta-analysis was not performed because we identified few trials focused on heterogeneous populations, intervention types, and outcomes.

The overall strength of the body of evidence was assessed for each KQ as high, moderate, low, or insufficient using methods developed for the USPSTF (and the Evidence-based Practice program), based on the overall quality of studies, consistency of results between studies, precision of findings, and risk of reporting bias.⁸ The applicability of the findings to US primary care populations and settings was also assessed.

A total of 30 studies (34 articles) with 14 959 participants were included (Figure 2). Because of the paucity of eligible studies on elder abuse and abuse of vulnerable adults, this literature is not organized by KQ but is briefly described at the end of this section.

Key Question 1. Does screening for current, past, or increased risk for IPV in adults and adolescents reduce exposure to IPV, physical or mental morbidity, or mortality?

Three RCTs (n = 3759) compared universal screening for IPV in a health care setting with no screening (Table 1); of these, 1 (n = 2708) enrolled women from 10 US primary care clinics,¹² 1 (n = 344) enrolled participants from a single New Zealand emergency department,¹⁴ and 1 (n = 707) enrolled participants from a variety of Canadian clinical settings.¹⁵ All studies enrolled adult women only (mean ages, 34-40 years); prevalence of past-year IPV ranged from 12% to 18% across studies. Responses to positive screening results in the intervention group included brief counseling and referral.

Overall, consistent evidence from 3 RCTs found no benefit of screening adult women for IPV followed by brief counseling or referral (eTable 12 in the Supplement). No study found a significant reduction in IPV among the screened group compared with a nonscreened control group (Figure 3).^12-15 Two RCTs (n = 3415) measured quality of life and found similar scores between groups (eFigure 2 in the Supplement)^12,15; one of these (n = 707) found no significant association between the intervention and improved posttraumatic stress disorder symptoms¹⁵ (eFigure 2 in the Supplement).

Key Question 2. What is the accuracy of screening questionnaires or tools for identifying adults and adolescents with current, past, or increased risk for IPV?

Fifteen studies (n = 4460) assessed the accuracy of 12 IPV screening tools (Table 2).^16-30 Studies used the following validated reference standards to establish screening test accuracy: Composite Abuse Scale, CTS or CTS-2, and Index of Spousal Abuse. One study²⁵ used a semistructured interview as the reference standard to determine the presence of IPV. All studies enrolled adults, and most enrolled only women or a majority of women; 1 study (n = 53) included only men.²⁴ None focused on pregnant women; only 2 (n = 5717) reported on the percentage of women who were pregnant (8%-9%).^16,28 Recruitment settings varied across studies (Table 2): emergency departments,^{18,19,24,25,29} primary care,^16,17,27,30 urgent care,²⁶ mixed settings,^23,28 and telephone or mail surveys.^20-22 All but 3 studies^23,27,28 were conducted in the United States. The prevalence of current or recent IPV ranged from 11% to 29%, with a median of 24%.

Included studies assessed 12 different screening tools (eTable 13 in the Supplement shows copies of the tools). Most assessed a tool designed to identify persons experiencing past-year IPV; however, 5 studies (n = 1908) reported on the accuracy for identifying current (ongoing) abuse,^{16,18,26,29,30} 1 (n = 75) assessed the accuracy of detecting lifetime abuse,³¹ and 1 (n = 409) assessed the accuracy for predicting future IPV (3-5 months).²²

Only studies that enrolled women and reported on accuracy of a tool designed to detect past-year or current overall IPV (not subtypes of abuse only) are described below; results for all tools are provided in eTable 14 in the Supplement.

Five studies (n = 6331) reported on accuracy of a tool for detecting past-year IPV: Humiliation, Afraid, Rape, Kick (HARK); Hurt, Insulted, Threaten, Scream (HITS); E-HITS (an extended version of the HITS, with an additional item assessing sexual abuse); Parent Screening Questionnaire; Partner Violence Screen (PVS); and Woman Abuse Screening Tool (WAST). Across all screeners, sensitivity ranged from 65% to 87% and specificity from 80% to 95% (Figure 4). Most were assessed by only 1 study; the HITS was assessed in 2 studies enrolling women veterans, 1 of which also evaluated the E-HITS. Accuracy estimates for the HITS and E-HITS were consistent but imprecise: sensitivity ranged from 75% to 78% and specificity from 80% to 83% (Figure 4). The largest study (n = 5605) evaluated the WAST and reported a sensitivity of 87% (95% CI, 85%-90%) and specificity of 89% (95% CI, 88%-90%) in a population of women from a variety of clinical settings.²⁸

Four studies (n = 1795) reported on the accuracy of 5 tools in identifying ongoing or current relationship violence: Abuse Assessment Screen (AAS); Ongoing Abuse Screen (OAS); Ongoing Violence Assessment Tool (OVAT); Slapped, Things Threatened (STaT); and an unnamed tool.^18,26,29,30 As shown in Figure 4, accuracy varied widely; sensitivity ranged from 46% to 94% and specificity from 38% to 95%. Only 1 tool, the OVAT, had a sensitivity and specificity greater than 80%.

Key Question 3. What are the harms of screening for IPV in adults and adolescents?

Two RCTs reported on harms of screening.^14,15 Both were included in KQ1; detailed results are reported in eTable 15 in the Supplement. The RCT enrolling women from various Canadian health care settings¹⁵ actively monitored harms of screening using the Consequences of Screening Tool (COST), which includes an 8-item Effects on Quality of Life subscale.³² The COST was administered to a subset of 591 women (of 3271 screened) within 14 days of screening. Mean scores were similar among women who screened positive and negative for abuse, with no significant differences, and reflected no harms associated with screening (mean scores ranged from 3.3 to 3.7 across groups [range, 16 to −16, with negative scores reflecting harm]). The second RCT (n = 344) stated that participants, clinicians, or research staff reported no adverse events; however, it is not clear whether adverse events were prespecified or how they were monitored.¹⁴

Key Question 4. How well do interventions reduce exposure to IPV, physical or mental morbidity, or mortality among screen-detected adults and adolescents with current, past, or increased risk for IPV?

Eleven RCTs (n = 6740) evaluated an intervention for women with screen-detected IPV or who were considered at risk for IPV, 5 (n = 1959) enrolled women during the perinatal period,^31,33-38 and 6 (n = 5712) enrolled nonpregnant women (Table 3).^39-45 All but 3 RCTs were conducted in the United States: 1 in Australia³⁹ and 2 in Hong Kong.^37,44,45 Studies assessed heterogeneous interventions (details of components are shown in eTable 16 in the Supplement). Outcomes measured included IPV, quality of life, depression, and anxiety (eTables 17 and 18 in the Supplement).

Two RCTs (n = 882) assessed the benefit of multiple home visits during the perinatal period, conducted either by paraprofessionals or trained nonprofessionals, focused on empowerment, support, and linkages to needed services.^33,36 One (n = 239) found a statistically significant reduction in IPV at follow-up (SMD, −0.34 [95% CI, −0.59 to −0.08]) (Figure 5).³⁶

Three RCTs evaluated brief clinic-based counseling (ranging from 1-8 sessions across studies) and found mixed results (Figure 5).^31,34,37,38 A 1-session brief counseling intervention (n = 110) was associated with lower rates of psychological abuse (SMD, −0.39 [95% CI, −0.78 to −0.01]) and minor physical violence (SMD, −0.47 [95% CI, −0.86 to −0.09]) and with mixed results for quality of life (significant improvement in some 36-Item Short Form Health Survey subdomains, no difference in others, and statistically significantly worse scores for bodily pain).³⁷ One RCT assessing a behavioral counseling intervention for women with 1 or more risk factors (IPV, depression, smoking, environmental tobacco exposure) for adverse pregnancy outcomes reported on outcomes among the subgroup who had IPV at baseline (n = 306); women in the intervention group had fewer recurrent episodes of IPV during pregnancy and postpartum (adjusted odds ratio, 0.48 [95% CI, 0.29 to 0.80])³¹ and fewer very preterm neonates (≤33 weeks) (2 vs 9 women; P = .03); there was no statistically significant difference between groups in rates of low-birth-weight neonates (<2500 g), very-low-birth-weight neonates (<1500 g), or rates of preterm birth (<37 weeks).³⁴ One study (n = 54) of interpersonal psychotherapy (4 sessions) was not associated with significant improvement in any outcome.³⁸

Six RCTs (n = 5712) enrolled populations for whom perinatal status was not an inclusion criterion; all assessed brief counseling interventions (Table 3).^39-43,45 Results are shown in Figure 6. All reported on IPV, and 4 (n = 5308) found no significant difference between groups in rates of overall IPV^39,42 or combined physical and sexual violence^40,41; measures of IPV were either similar between groups or slightly higher in the intervention group. One RCT (n = 200) reported on subtypes of violence only and found benefit for psychological aggression but not for physical assault or sexual coercion.⁴⁵ Two RCTs (n = 475) measured quality of life; both found similar scores among the intervention and control groups, and differences were not statistically significant.^39,45 Three RCTs (n = 676) reported on depression outcomes; 2 (n = 472) found benefit in favor of the intervention group (although 1 found a difference below the threshold considered clinically meaningful)^39,45 and 1 (n = 204) found similar scores between groups.⁴³ One RCT (n = 272) found no difference between groups in the percentage of women who had anxiety at 6 and 12 months.³⁹

RCTs of IPV interventions were limited by high overall attrition (20% or higher in 7 of 11 RCTs), potential measurement bias (eg, recall bias or variation in comfort with self-reported measures of violence frequency), and heterogeneity in outcome reporting (particularly for IPV outcomes). Usual care and use of a co-intervention (eg, provision of an IPV resource sheet) in control groups varied across studies and sometimes was not described. In the behavioral counseling intervention targeting multiple risk factors (smoking, environmental tobacco smoke exposure, depression, and IPV),^31,34 improvement in birth outcomes among women reporting IPV at baseline may not be attributable to IPV counseling. In the subgroup of women reporting IPV at baseline, 62% reported being depressed, and those randomized to the intervention were also offered counseling for depression (in addition to IPV)³¹; the improvement in outcomes may be attributable to counseling for depression as opposed to IPV counseling.

Key Question 5. What are the harms of interventions for IPV in adults and adolescents?

Five RCTs (n = 1413) assessing interventions for IPV reported on harms (Table 3)^{36,37,39,42,45}; all are included in KQ4. One RCT³⁹ (n = 272) assessing brief counseling surveyed women at 6 and 12 months about potential harms of study participation (eg, whether they felt negatively judged by practice staff based on being a participant, whether their partner abuse had increased as a result of participation); there was no difference between groups in reported harms, and authors concluded that no harms were associated with the intervention. In 1 RCT (n = 110),³⁷ set in a prenatal clinic in Hong Kong, women were asked whether violence frequency had increased with study participation; no adverse events related to participation were reported by women in either group.³⁷ Three other RCTs (n = 1031) reported that no harms were associated with the intervention but did not comment on how harms were measured and assessed.^36,42,45

No studies of elder abuse or abuse of vulnerable adults were found pertinent to KQs 1, 3, 4, and 5, and only 1 study of elder abuse was found relevant to KQ2. One study (n = 139) assessing the accuracy of screening for elder abuse among individuals 65 years or older presenting for routine dental care was included.⁴⁶ Eligible participants received caregiver assistance and scored 18 or more on the Mini Mental Status Examination. Screening was conducted using the 15-item Hwalek-Sengstock Elder Abuse Screening Test (H-S/EAST) and compared against the CTS. The enrolled population had a relatively high prevalence of elder maltreatment based on CTS violence or verbal aggression scales (41%). Compared with the CTS (violence and verbal aggression scales combined), the H-S/EAST tool had a sensitivity of 46% (95% CI, 32% to 59%) and specificity of 73% (95% CI, 62% to 82%) for detecting elder abuse.

A summary of the evidence for IPV is provided in Table 4. Consistent evidence from 3 RCTs (n = 3759) found no benefit of screening adult women for IPV over 3 to 18 months of follow-up.^12,13 RCTs of IPV screening were limited by heterogeneity in enrollment settings and screening processes; however, trials measured similar outcomes and found consistent results. Potential harms of screening asymptomatic populations for abuse include labeling, stigma, and increased violence. No included RCT found harms associated with screening. However, only 1 RCT actively monitored harms; these were assessed 14 days after screening, which may not be an adequate duration.¹⁵ Other potential harms include false-positive test results that lead to more in-depth inquiry or referrals from health professionals that would not lead to benefit and may cause labeling.

Screening tools are available for clinical practice that may reasonably identify women experiencing past-year IPV. Most tools were assessed in only 1 study. In the 5 studies assessing accuracy of tools for detecting past-year IPV in women (HARK, HITS, E-HITS, PVS, and WAST), sensitivity ranged from 65% to 87% and specificity from 80% to 95%. When limiting to studies enrolling participants from non–emergency department settings (ie, primary care or community samples only), sensitivity was slightly higher (range, 75%-87%) and specificity was unchanged. Estimates of screening test accuracy for detecting past-year IPV were derived from populations with an IPV prevalence of 14% to 27%. This is similar to the prevalence reported by the KQ1 RCT enrolling women from US primary care settings (15%). In a population of 100 000 women with 15% prevalence of IPV, use of the HARK tool (80% sensitivity and 95% specificity) would result in 81 000 true-positive test results and 5000 false-positive results (positive predictive value, 83%). Use of the WAST tool, with slightly higher sensitivity (87%) but lower specificity (89%) than the HARK, in a population with the same IPV prevalence would result in 87 484 true-positive results and 11 000 false-positive results (positive predictive value, 56%). The meaning of false-positive test results is unclear. This may indicate a misunderstanding of the screening question; alternatively, women experiencing IPV may choose to answer the reference standard negatively, since disclosure of violence may be uncomfortable.

Evidence from 11 studies (n = 6740) evaluating interventions for women with screen-detected IPV was heterogeneous, imprecise, and often inconsistent. Two RCTs assessed home visiting interventions^33,36 and 1 found a significantly lower rate of any IPV among women in the intervention group than in the control group.³⁶ One RCT assessing a behavioral counseling intervention in pregnant women found benefit for IPV and very preterm birth.⁴⁷ The 6 RCTs enrolling nonpregnant women (all focused on brief counseling) generally found no benefit or mixed results.

Conclusions of this review for IPV interventions may differ slightly from those from the prior (2012) review for the USPSTF.⁴⁸ In addition to including several recently published studies, 1 trial (Mothers’ Advocates in the Community [MOSAIC]) included in the prior report was excluded because it enrolled women referred to the intervention based on symptoms of abuse or self-disclosure of IPV status.⁴⁹ Women in the MOSAIC trial randomized to weekly home visits by trained nonprofessional peer supporters had lower mean abuse scores than women in the control group at 1 year.

Few RCTs reported on harms of interventions. No trial found increased IPV among the intervention group or other harms attributed to the intervention.

To better understand the benefits and harms of screening asymptomatic populations in primary care settings for abuse, future studies could assess whether screening specific groups results in improved health outcomes. Included RCTs of screening enrolled women of childbearing age, but none enrolled women from prenatal settings or reported outcomes separately for women screened during pregnancy. Screening studies should report on harms over a sufficient period after screening to assess potential psychosocial harms. In addition, research is needed to assess the accuracy of screening tools in men, as well as the benefit and harms of interventions for men who experience IPV.

No RCTs of screening or interventions for older and vulnerable adults were identified in this review. Studies of screening tools also are lacking. Screening and interventions for this population are likely to be different than those for IPV, given the nature of the abuse (eg, different relationship with the perpetrator). In addition, some older and vulnerable adults may not have sufficient physical, mental, or financial abilities to engage in screening. For these situations, instruments could be targeted toward caregivers. Additional challenges to this research may include legal requirements related to disclosure, underlying medical conditions of patients (eg, cognitive impairment), and dependence on the perpetrator for caregiving and access to medical care.

This review has several limitations. First, the scope focused on asymptomatic populations without signs or symptoms of abuse. The literature on whether certain physical or psychological symptoms should trigger an assessment of abuse (ie, “case finding”) was not assessed. Second, this review did not evaluate evidence on programs to prevent abuse victimization or studies that assess screening and interventions for abuse perpetrators. Third, for KQ3 (harms of screening), the review was limited to study designs with a concurrent control group; this excluded uncontrolled studies. The prior review for the USPSTF concluded that study populations and methods in uncontrolled studies varied widely.⁴⁸ Results from these studies did not show significant harm related to screening; some found that a minority of women indicated discomfort with screening (particularly those with prior IPV), infringement of privacy, worries about increasing abuse by disclosing IPV, and feelings of sadness or depression.⁴⁸

Although available screening tools may reasonably identify women experiencing IPV, trials of IPV screening in adult women did not show a reduction in IPV or improvement in quality of life over 3 to 18 months. Limited evidence suggested that home visiting and behavioral counseling interventions that address multiple risk factors may lead to reduced IPV among pregnant or postpartum women. No studies assessed screening or treatment for elder abuse and abuse of vulnerable adults.

Corresponding Author: Cynthia Feltner, MD, MPH, Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, 725 Martin Luther King Jr Blvd, CB#7295, Chapel Hill, NC 27599 (cindy_feltner@med.unc.edu).

Accepted for Publication: August 16, 2018.

Author Contributions: Dr Feltner had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Feltner, Wallace, Kistler, Jonas.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Feltner, Wallace, Berkman, Kistler, Middleton, Barclay, Higginbotham, Jonas.

Critical revision of the manuscript for important intellectual content: Feltner, Wallace, Kistler, Higginbotham, Green, Jonas.

Statistical analysis: Feltner, Higginbotham.

Obtained funding: Feltner, Jonas.

Administrative, technical, or material support: Feltner, Berkman, Kistler, Middleton, Barclay, Higginbotham, Green, Jonas.

Supervision: Feltner, Jonas.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This research was funded under contract HHSA-290-2012-00015-I, Task Order 5, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services, under a contract to support the USPSTF.

Role of the Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.

Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project, including AHRQ staff (Amanda Borsky, DrPH, MPP; Tracy Wolff, MD, MPH) and RTI International–University of North Carolina EPC staff (Carol Woodell, BSPH; Lynn Whitener, DrPH, MSLS; Sharon Barrell, MA; Loraine Monroe). USPSTF members, expert consultants, peer reviewers, and federal partner reviewers did not receive financial compensation for their contributions. Ms Woodell, Dr Whitener, Ms Barrell, and Ms Monroe received compensation for their role in this project.

Additional Information: A draft version of this evidence report underwent external peer review from 3 content experts (Adelita Gonzales Cantu, PhD, RN, The University of Texas Health Science Center at San Antonio; Joanne Klevens, MD, PhD, Centers for Disease Control and Prevention; and Timothy Platts-Mills, MD, MSc, University of North Carolina at Chapel Hill) and 2 federal partner reviewers from the Centers for Disease Control and Prevention. Comments from reviewers were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence review.

Editorial Disclaimer: This evidence report is presented as a document in support of the accompanying USPSTF Recommendation Statement. It did not undergo additional peer review after submission to JAMA.