Maya G Rowland

Low-Dose Aspirin for Prevention of Morbidity and Mortality From Preeclampsia: A Systematic Evidence Review for the U.S. Preventive Services Task Force

Preeclampsia is a leading cause of maternal death, affecting 2% to 8% of pregnancies globally (1, 2). It affected 3.8% of U.S. deliveries in 2010, and the rate of severe preeclampsia has increased over the past 3 decades (3). Perinatal mortality is nearly 2 times higher in pregnancies affected by preeclampsia (4), with 12% of maternal deaths due to the condition (5). Serious illness is more common, with more than one third of serious maternal morbidity and 15% of preterm births related to preeclampsia (6, 7). Preeclampsia is defined as hypertension (blood pressure 140/90 mm Hg) and proteinuria (presence of 0.3 g of protein in a 24-hour period) observed during the second half of pregnancy (>20 weeks of gestation) (8, 9). It is also classified as having severe features with any of the following: blood pressure above 160/110 mm Hg, thrombocytopenia, impaired liver function, renal insufficiency, pulmonary edema, or cerebral or visual disturbances (9). Preeclampsia with or without severe features can evolve rapidly into eclampsia or the hemolysis, elevated liver enzymes, and low platelets syndrome, sometimes leading to systemic complications and maternal death (10, 11). Poor perinatal health outcomes are associated with preeclampsia, primarily due to increased risk for intrauterine growth restriction (IUGR) or medically initiated preterm delivery. Once preeclampsia develops, the only effective treatment is delivery, with serious neonatal harms when remote from term (<34 weeks of gestation). Current understanding of preeclampsia pathophysiology suggests that it may be a collection of syndromes with different precipitating factors and outcomes (12). Early in pregnancy, aberrations in placental development can result in placental ischemia and release of inflammatory and oxidative stress factors into the maternal bloodstream. In addition, even with normal placentation, preexisting hypertension, diabetes, and other inflammatory conditions (such as lupus) may activate systemic inflammatory and oxidative stress processes, as can twin or higher-order pregnancies. Accurate prediction of who will develop preeclampsia and have serious complications is not currently possible (1315). The most consistent predictors of high risk are previous preeclampsia, certain medical conditions (diabetes, chronic hypertension, renal disease, autoimmune diseases, and the antiphospholipid syndrome), and multifetal pregnancy (16). Moderately elevated risk for preeclampsia is associated with nulliparity (first birth), advanced maternal age (40 years), between-pregnancy interval of more than 10 years, high body mass index (35 kg/m2), and family history of preeclampsia (mother or sister). Risk factors with less consistent evidence include changes in paternity between pregnancies, history of migraine headaches (17, 18), and asthma (17, 1922). Predictive models combining various biomarkers, patient risk factors, and clinical readings hold promise but are not yet sufficiently validated for clinical use (10, 2325). Previous comprehensive systematic reviews have found antiplatelets (primarily low-dose aspirin) to be beneficial for the prevention of preeclampsia among women at heightened risk (26, 27). We conducted this systematic review to support the U.S. Preventive Services Task Force (USPSTF) in updating its 1996 recommendation, which is no longer active. Methods Detailed methods are outlined in our full evidence report (28). This review addressed 3 key questions (Appendix Figure 1). First, is low-dose aspirin effective for reducing adverse maternal and perinatal health outcomes among women at increased risk for preeclampsia? Second, is low-dose aspirin effective for preventing preeclampsia among women at increased risk for the condition? Third, are there harms to the woman and fetus associated with aspirin use during pregnancy? Appendix Figure 1. Analytic framework and key questions. ARDS = acute respiratory distress syndrome; HELLP = hemolysis, elevated liver enzymes, and low platelets. *Abbreviated list of health outcomes. See Appendix Table 2 for a full list. Data Sources and Searches In addition to considering all studies from the previous USPSTF review, we performed a comprehensive search of MEDLINE, PubMed, the Database of Abstracts of Reviews of Effects, and the Cochrane Central Register of Controlled Trials for studies published between January 2006 and 1 June 2013. We also examined the reference lists from existing systematic reviews to identify potentially eligible studies, including an individual-patient data (IPD) meta-analysis published by the Perinatal Antiplatelet Review of International Studies (PARIS) Collaboration (27) and a 2007 Cochrane review (26). We searched ClinicalTrials.gov for ongoing trials (May 2013). Between the last search date and this publication, we actively monitored published literature for potentially important new trials or other large observational studies directly relevant to our key questions; none were identified. Study Selection Two investigators independently reviewed abstracts and full-text articles for inclusion according to predetermined criteria. We resolved discrepancies through consensus with a third investigator. To evaluate benefits of aspirin prophylaxis, we included any study that used a risk selection approach aimed at achieving a sample of women at high risk for preeclampsia. The trials could define risk on the basis of medical history, pregnancy characteristics, or clinical measurements known to be associated with risk for the condition. Although preeclampsia occurs more often in first births than in subsequent ones, prevalence rates are relatively low (approximately 4%) compared with other high-risk groups. Because aspirin treatment based only on this risk factor has not been supported, trials with nulliparity as the sole risk factor were not included for evaluation of benefits. We used broader inclusion criteria to identify possible harms of aspirin exposure during pregnancy. The trials of women at high risk were combined with trials of women at low or average risk exposed to daily low-dose aspirin. Large prospective observational studies were also included to assess harms but were not included in pooled analyses. We included interventions that compared patients receiving 50 to 150 mg of aspirin with a placebo or no treatment group and excluded studies of nonaspirin antiplatelet medications or aspirin combined with another active substance. We also excluded studies that we rated as poor-quality on the basis of the USPSTF quality rating standards (29) and studies not published in English. Data Extraction and Quality Assessment Two investigators critically appraised all included studies independently using the USPSTFs design-specific criteria (29), which we supplemented with the National Institute for Health and Care Excellence methodology checklists (30) and the Newcastle-Ottawa Scale (31). According to the USPSTF criteria, a good-quality study met all prespecified standards. A fair-quality study did not meet (or it was unclear whether it met) at least 1 criterion, but it also had no known limitation that could invalidate its results. A poor-quality study had a single fatal flaw or multiple important limitations that could seriously bias its results. Discrepancies were resolved through discussion of identified limitations and consultation with a third investigator, if necessary. One investigator extracted study details and results, and a second investigator reviewed the abstracted information. Data Synthesis and Analysis We used the metan procedure in Stata, version 11.2 (StataCorp, College Station, Texas), for all reported meta-analyses and the metaan procedure for sensitivity analyses (32). For dichotomous outcomes, we entered the number of events and nonevents and estimated pooled random-effects risk ratios by using the DerSimonianLaird method for all outcomes, except those in which fewer than 10% of the participants had the event (33), for which we used a fixed-effects MantelHaenszel model (34). We also included prediction intervals in forest plots of random-effects models, which provided an estimate of where the effect size from 95% of newly conducted trials would fall, assuming that the between-study variability in the included trials held for new trials (35). The prediction intervals are shown on the forest plots by the horizontal lines that extend from the diamond representing the 95% CI of the pooled estimate. Potential sources of heterogeneity in effect size by aspirin timing, dosage, and preeclampsia risk determination were identified a priori and explored using meta regression and visual inspection of sorted forest plots. We used the I 2 and chi-square statistics to assess statistical heterogeneity. To evaluate small-study effects, we examined funnel plots and used the Begg or Peter test depending on the outcome distribution (36, 37). We used profile likelihood estimation to conduct sensitivity analyses for the pooled effects because the DerSimonianLaird method can overestimate CI precision in meta-analysis, particularly when fewer than 10 studies or when smaller studies with few events are pooled (38). Role of the Funding Source This study was funded by the Agency for Healthcare Research and Quality (AHRQ) under a contract to support the work of the USPSTF. Members of the USPSTF and the AHRQ medical officer assisted in the development of the reviews scope. Approval from AHRQ was required before the manuscript could be submitted for publication, but the authors are solely responsible for its content and the decision to submit it for publication. Results Our literature search yielded 544 unique citations. From these, we reviewed the full text of 75 articles. Twenty-three studies (27 articles) met our inclusion criteria (Appendix Figure 2 and Appendix Table 1). Appendix Figure 2. Summary of evidence search and selection. The diagram excludes 51 RefMan (Thomson Reuters, Philadelph

Behavioral Counseling to Promote a Healthy Lifestyle in Persons With Cardiovascular Risk Factors: A Systematic Review for the U.S. Preventive Services Task Force

Decreases in cardiovascular mortality rates in recent decades have been attributed, in part, to improvements in modifiable risk factors (1). A substantial portion of the U.S. population has at least one modifiable risk factor for cardiovascular disease (CVD) (such as hypertension, dyslipidemia, impaired fasting glucose, the metabolic syndrome, and cigarette smoking) (27). Despite convincing evidence that healthy diet and physical activity are associated with important health outcomes, including reduction in cardiovascular events and mortality rates (817), U.S. adults are not meeting recommendations for healthy diet and physical activity (1820). Likewise, nutrition and exercise counseling practices in primary care remain suboptimal, even for persons at high risk for CVD (2124). In 2012, the U.S. Preventive Services Task Force (USPSTF) recommended that clinicians consider selectively providing or referring adults without preexisting CVD or risk factors for intensive behavioral counseling interventions (C recommendation) (25). The USPSTF subsequently recommended that clinicians screen all adults for obesity and offer or refer obese patients to intensive, multicomponent behavioral interventions (B recommendation) (26). This systematic review was designed to complement the existing reviews that supported the 2012 USPSTF recommendations and to support the USPSTF in updating its 2002 and 2003 recommendations on healthy diet and physical activity counseling in persons with known cardiovascular risk factors (27, 28). To conduct this review, we developed an analytic framework with 4 key questions (Supplement 1) that included the effect of dietary or physical activity counseling on patient health outcomes (question 1), intermediate cardiovascular diseaserelated outcomes (question 2), behavioral outcomes (question 3), and the harms of counseling (question 4). Supplement 1. Analytic Framework Methods Detailed methods, including search strategies; detailed inclusion criteria; and excluded studies are publically available in our full evidence report (29). Data Sources and Searches We searched MEDLINE, PubMed, PsycINFO, the Database of Abstracts of Reviews of Effects, and the Cochrane Central Register of Controlled Trials from January 2001 to October 2013. We supplemented our searches with suggestions from experts and reference lists from other relevant systematic reviews. Study Selection Two investigators independently reviewed 7218 abstracts and 553 full-text articles against a priorispecified inclusion criteria (Supplement 2). We included studies in adults who had at least 1 cardiovascular risk factor, including hypertension, dyslipidemia, impaired fasting glucose or glucose tolerance, the metabolic syndrome, and cigarette smoking. We excluded studies limited to persons with known diabetes (considered a CVD risk equivalent), coronary artery disease, cerebrovascular disease, peripheral artery disease, or severe chronic kidney disease. We also excluded populations at increased risk for CVD (such as those who are obese, physically inactive, and prehypertensive) but without other CVD risk factors because these bodies of evidence were considered in previous reviews (30, 31) and USPSTF recommendations (25, 26). We included behaviorally based counseling interventions to promote a healthy diet or physical activity, delivered alone or as part of a multicomponent intervention. We excluded interventions that provided controlled diets or supervised exercise, as opposed to interventions aimed at evaluating whether counseling could change behavior. Supplement 2. Literature Flow Diagram We limited studies of efficacy or effectiveness to fair- or good-quality randomized, controlled trials or controlled clinical trials that had at least 6 months of follow-up, were done in developed countries, and published their results in 1990 or later. Included trials had to have a control group (such as usual care, a minimal intervention, or attention control). We examined health outcomes (such as morbidity or mortality related to CVD), intermediate health outcomes (such as physiologic measures of blood pressure, lipid and glucose, and weight; diabetes incidence; medication use; and composite CVD risk scores), and behavioral outcomes (such as self-reported dietary intake and physical activity or objectively measured markers of behavior change [such as VO2max or urinary sodium]). We also included observational studies that reported serious harms (that is, adverse events resulting in unexpected or unwanted medical attention). Data Extraction and Quality Assessment One reviewer extracted population characteristics, study design elements, intervention and control characteristics, and study results into standardized evidence tables. A second reviewer checked the data for accuracy. Articles that met our inclusion criteria were critically appraised by 2 reviewers independently using the USPSTF and National Institute for Health and Care Excellence criteria (32, 33). We rated articles as good-, fair-, or poor-quality. Good-quality studies generally met all criteria, whereas fair-quality studies did not meet all criteria but had no known important limitation that could invalidate its results. Poor-quality studies had important limitations that were considered fatal flaws (for example, more than 40% attrition with or without differential attrition between intervention groups; lack of randomization with biased assignment of participants to intervention groups, often with differences in baseline characteristics or no reporting of baseline characteristics; per protocol analyses only; and description of methods that did not allow adequate assessment of quality). These studies were excluded from this review. Data Synthesis and Analysis Because of the clinical heterogeneity across this body of evidence, we stratified our analyses according to the type of intervention (that is, a focus on dietary counseling alone, physical activity alone, or combined diet and physical activity counseling) and according to how study populations were targeted or defined (that is, dyslipidemia, hypertension, impaired fasting glucose or glucose tolerance, or mixed risk factors). We did random-effects meta-analyses for 5 or more studies using the DerSimonianLaird method to estimate the effect size of counseling on intermediate health outcomes (that is, systolic and diastolic blood pressure; total, high-density lipoprotein, and low-density lipoprotein cholesterol; triglycerides; fasting blood glucose; diabetes incidence; and weight or body mass index) (34). We did qualitative synthesis for health outcomes, behavioral outcomes, and harms. Outcome analyses were also stratified by length of follow-up after randomization (short term was less than 12 months, intermediate term was 12 to 24 months, and long term was greater than 24 months). We used stratified analyses, visual inspection of forest plots arranged by effect size, and/or meta-regressions to examine the effect of a priorispecified primary sources of heterogeneity on effect size: study population, intervention type, overall intervention intensity (low was less than 30 minutes of total contact, medium was 30 to 360 minutes, and high was more than 360 minutes), number of intervention contacts, duration of intervention, length of follow-up, overall study quality, year of publication, country setting, type of control group, and population risk (including average age; percentage of persons who smoke or have hypertension, dyslipidemia, or diabetes; average systolic blood pressure; average low-density lipoprotein cholesterol level; average body mass index; and use of medications). We assessed the presence of statistical heterogeneity among the studies using standard chi-square tests, and the magnitude of heterogeneity was estimated using the I 2 statistic (35). In instances of 10 or more studies, we formally assessed for publication bias and whether the distribution of the effect sizes was symmetrical with respect to the precision measure by using funnel plots and the Egger linear regression method (36, 37). We did all analyses using Stata, version 11.2. Role of the Funding Source Agency for Healthcare Research and Quality staff oversaw the project and assisted in external review of the companion draft evidence synthesis. Liaisons for the USPSTF helped to resolve issues about the scope of the review but were not involved in the conduct of the review. Results Description of Included Trials Seventy-four fair- or good-quality healthy lifestyle counseling trials in persons with cardiovascular risk factors met our inclusion criteria (Supplements 3 and 4). Forty-nine trials evaluated combined lifestyle counseling interventions, 18 diet-only interventions, and 10 physical activityonly interventions. Of the interventions evaluated, only 2 were low-intensity, 48 were medium-intensity, and 37 were high-intensity. Medium-intensity interventions had a median of 5 contacts (interquartile range [IQR], 3 to 8 contacts) and a median duration of 9 months (IQR, 4 to 11 months). High-intensity interventions had a median of 16 contacts (IQR, 9 to 31 contacts) and a median duration of 12 months (IQR, 8 to 18 months). Counseling interventions included didactic education as well as individualized care plans, problem-solving skills, and audit and feedback. Many trials included weight loss or weight goals for participants who were overweight. Some counseling interventions included cointerventions (such as smoking cessation counseling when applicable, protocols for medication adjustment, and provision of free or low-cost exercise options). Interventions were delivered by dietitians, nutritionists, physiotherapists, exercise professionals or consultants, or trained interventionists (such as health educators, psychologists, nurses, or case managers). Supplement 3. Trial Results in Health Outcomes (Key Question 1) Supplement 4. Included Studies and Outcomes, by Intervention

Primary care behavioral interventions to prevent or reduce illicit drug use and nonmedical pharmaceutical use in children and adolescents: a systematic evidence review for the U.S. Preventive Services Task Force.

Drug use among adolescents is a serious public health problem in the United States. The 2012 National Survey on Drug Use and Health reported that 9.5% of children aged 12 to 17 years reported illicit drug use during the past month. Marijuana and prescription psychotherapeutics (including pain relievers) are the most commonly used drugs among children and adolescents. Seven percent of children aged 12 to 17 years reported current use of marijuana, and an estimated 5% used marijuana for the first time within the past year. In 2012, 2.8% of children aged 12 to 17 years reported using a prescription drug for nonmedical reasons, and 2.2% reported nonmedical use of opioid pain relievers. Illicit drug use was approximately 15 times higher among young persons who smoked cigarettes and drank alcohol during the past month (61.1%) than among those who neither smoked cigarettes nor drank alcohol during the past month (4.0%) (1). Drug and alcohol use are the primary health risk behaviors that contribute to unintentional injuries, homicide, and suicidethe leading causes of morbidity and mortality among adolescents and young adults (2). Even infrequent drug or alcohol use increases the risk for serious adverse events by increasing risk-taking behaviors in intoxicated or impaired persons. The Substance Abuse and Mental Health Services Administration and the American Academy of Pediatrics recommend that universal screening, brief intervention, and referral to treatment (SBIRT) for substance use should be a part of routine health care as a method to reduce the health burden associated with substance use (3, 4). Although SBIRT is appropriate for all levels of risk, it is a particularly useful early intervention approach to identifying and intervening with persons with nondependent substance use before they require extensive or specialized treatment. Among children and adolescents, primary care interventions can include positive feedback for nonusers as primary prevention; brief advice for those at low risk for abuse (secondary prevention); or a motivational intervention directed at high-risk patients for reducing use, reducing associated high-risk behaviors, or accepting a referral to treatment. In 2008, the U.S. Preventive Services Task Force (USPSTF) concluded that the evidence was insufficient to recommend for or against screening adolescents, adults, and pregnant women for illicit drug use (5). We undertook the current review to synthesize the evidence on the benefits and harms of primary carerelevant behavioral interventions designed to prevent or reduce illicit drug use or the nonmedical use of prescription drugs among children and adolescents only. The USPSTF used this review to update its recommendation for this population. Methods With input from the USPSTF, we developed an analytic framework and 3 key questions (KQs) to guide our review (Appendix Figure 1). The proposed analytic framework and KQs were posted on the USPSTFs Web site for public comment for 4 weeks. On the basis of this input, we made appropriate revisions and received final approval for publication from USPSTF liaisons. The full report provides details on our methods and results, including search strategies and all evidence tables (www.uspreventiveservicestaskforce.org/uspstf13/drugmisuse/drugmisusedraftrep.htm). Appendix Figure 1. Analytic framework and key questions. Data Sources and Searches We searched for English-language publications in PubMed, PsycINFO, and the Cochrane Central Register of Controlled Trials from January 1992 through 4 June 2013 and in MEDLINE through 31 August 2013. We also assessed the 2 trials that were specific to children and adolescents and were included in the 2008 review (6). We examined the reference lists of 6 relevant published reviews and meta-analyses (712), as well as the reference lists of included studies. We considered gray literature sources and recommendations from experts. Study Selection Two investigators independently reviewed abstracts against prespecified eligibility criteria. We dually reviewed all full-text articles for potential inclusion. We included randomized, controlled trials (RCTs) or controlled clinical trials designed to prevent or reduce drug use in children and adolescents (aged <18 years [no lower age restriction]) who were not diagnosed with a substance use disorder or seeking treatment for substance misuse. We included trials conducted in primary care or those that tested interventions we judged feasible for conduct in primary care that had a link to a health care setting or system, with or without referral to specialty treatment services. This included interventions employing the full SBIRT model and other approaches to primary prevention (to prevent initiation of use) or tertiary prevention (to prevent continued use and adverse effects in those already using). We also included interventions delivered exclusively through electronic media (such as the Internet or CD-ROMs) that were not linked to health care. We excluded trials among youths diagnosed with substance abuse or dependence because they represented specialty treatment only. We also excluded studies conducted among adolescents who were mandated or directly referred to substance abuse or dependence treatment via the juvenile justice system, social services, parents, or a similar referral system. In addition, we excluded interventions conducted in substance abuse treatment centers, schools, worksites, and other institutions (for example, juvenile detention centers). Included trials had control groups that offered minimal or no treatment and reported drug use or health or social outcomes at least 6 months after baseline. Data Extraction and Quality Assessment Two independent investigators rated the quality of all included trials as good, fair, or poor according to USPSTF standards (13). We excluded poor-quality trials. One reviewer abstracted data from studies that were rated fair or good. A second reviewer checked all abstracted data for accuracy and completeness. We resolved discrepancies through discussion. Data Synthesis and Analysis We summarized all included studies in narrative form and summary tables detailing the important features of the study populations, design, intervention, and results. We used the between-group differences that were reported by authors of included studies, when available. We identified too few trials to conduct any meta-analysis, as well as too much variability in several factors (such as population or intervention). As a result, we conducted a qualitative analysis for all KQs and stratified the results into 2 groups based on the intervention: primary carebased or computer-based. Primary carebased studies recruited directly from or were conducted in primary care clinics. Computer-based interventions were judged to be feasible for primary care because they used only electronic methods of delivery, although they did not recruit from or take place in primary care. Role of the Funding Source Agency for Healthcare Research and Quality (AHRQ) staff provided technical oversight for the project. Although USPSTF liaisons helped resolve issues around the reviews scope, they were not involved in the reviews conduct. Results We reviewed 2253 abstracts and 144 full-text articles for possible inclusion (Appendix Figure 2). We identified 6 trials (reported in 7 publications) that met our inclusion criteria (1420). The most common reasons for exclusion included settings (for example, not linked to or feasible for primary care [k= 45]), out-of-scope populations (for example, aged >18 years, seeking treatment, or diagnosed with substance abuse or dependence [k= 26]), and not reporting any relevant outcomes (k= 19) (Appendix Figure 2). Table 1 provides a summary of evidence for the benefits and harms of each included study by outcome (drug use behaviors [KQ 2], health and social outcomes [KQ 1], and harms [KQ 3]). Appendix Figure 2. Summary of evidence search and selection. Table 1. Summary of Evidence for Benefits and Harms of Drug Use Interventions Effects of Interventions on Drug Use Primary CareBased Interventions Three of the 6 studies were conducted in or recruited patients from primary care (Appendix Tables 1 and 2) (16, 17, 20) and tested 4 active treatment groups. We rated all 3 studies as fair-quality according to USPSTF standards (13), with various threats to internal validity (see the full report for more detail on study quality). The smallest study had 41 participants (17), and the largest had more than 2500 (16). Ages ranged from 12 to 20 years, and girls were overrepresented60% to 68% of participants were girls. All 3 studies took place in the United States, and 1 of them (16) also included a sample of adolescents in the Czech Republic. Two of the studies were conducted among a general primary care population (16, 20), whereas the remaining study was conducted among a sample of young persons diagnosed with asthma (17). One of the studies screened adolescents for drug use before enrollment; only those who reported any marijuana use in the past year were randomly assigned (20). In this trial, marijuana use was the primary focus; the other studies targeted drug, alcohol, or tobacco use. Appendix Table 1. Study Characteristics of Included Trials Appendix Table 2. Intervention Characteristics of Included Trials All 3 studies took place during 1 office visit. Three of the interventions included brief counseling (2 to 40 minutes) by the primary care physician (16), family nurse practitioner (17), or trained research therapist (20), and all included a computer-based, self-administered educational component. The study by Walton and colleagues randomly assigned adolescents to a therapist-led brief intervention, computer-based brief intervention, or usual care control group (20). Interventions provided information and advice about substance use along with a decision-making exercise. One of the trials (16) was consis