Systematic Evidence Review
The foundation of the systematic evidence review performed in support of the guideline development process was a series of critical questions related to cardiovascular disease (CVD) risk and prevention in youth. The questions encompassed defined risk factors, predefined outcome measures for each risk factor, and, most importantly, measures of CVD and target organ damage (TOD). Each of these elements was developed and refined through a review of the existing evidence by the Expert Panel. Additional information on the Critical Questions can be found in Table A-1 in Appendix A of the original guideline document.
To inform the identification of studies related to the critical questions, the Expert Panel held an inservice training with the contractor staff members who would be involved in overseeing the literature review to initiate the evidence review process. In addition, a series of group training sessions was held with the contractor staff at appropriate points throughout the process to clarify the scope of the review and expectations for supporting the production of high-quality, evidence-based guidelines.
Based on the critical questions, risk factors, and types of CVD TOD of interest, search parameters were developed to identify published studies relevant to pediatric cardiovascular (CV) risk reduction. This process involved determining appropriate databases, dates, terms, and limits for the search, as described below.
Searches were performed in the following databases:
- Cochrane Database of Systematic Reviews
- National Guideline Clearinghouse (NGC)
Searches were first conducted in PubMed/MEDLINE. Only unique studies from subsequent searches in the Cochrane database and NGC were retained for consideration (i.e., those studies that were not already captured in the initial PubMed/MEDLINE search).
In addition to these databases, a preliminary search of EMBASE was conducted. The great majority of studies identified in this preliminary search were also found in the other databases; therefore, it was determined that proceeding with a complete EMBASE search would not contribute significant additional information to the review.
The literature search allowed for further input by the Expert Panel to ensure that in-scope studies were not overlooked. Members of the Expert Panel contributed additional relevant studies based on their routine scanning of the literature. A supplementary literature search was also conducted to identify potentially relevant studies authored by members of the Expert Panel. Additional studies identified by these supplementary methods were included only if they met the same criteria for inclusion established for the primary evidence review.
Original searches in PubMed/MEDLINE, Cochrane, and NGC captured studies published between January 1, 1985, and December 31, 2006. Recognizing the timelag inherent in screening a large body of literature and developing evidence tables, the Expert Panel then called for an update of these searches to be conducted for the period between January 1, 2007, and June 30, 2007.
The Expert Panel established June 30, 2007, as the closing publication date for literature to be entered into the evidence review for these Guidelines. The Expert Panel recognized that, given the scope of these Guidelines and the nature of ongoing research in relevant areas, research findings might appear thereafter with the potential to have a material impact on one or more recommendations in the Guidelines. Therefore, to optimize the currency of the Guidelines, the Expert Panel sought, prospectively, to enable consideration of directly relevant, significant peer-reviewed evidence that might appear after the closing date. During a conference call convened on January 21, 2008, the Expert Panel Science Team established the following criteria to guide the full Expert Panel's consideration of studies published after the closing date:
- Any peer-reviewed published study identified by a member of the Expert Panel, as part of his or her routine surveillance of the literature, that is directly relevant to the recommendations of the Expert Panel will be considered for inclusion.
- To be included by the Expert Panel as evidence, the corresponding Risk Factor Team, or the full Expert Panel if applicable at a broader level, must judge that the findings of such recently published studies have the potential for a material impact on the content or strength of the recommendations of the Expert Panel.
- Such studies must meet the same basic criteria for inclusion established for the primary evidence review.
- If there is a difference of opinion about inclusion of a study, a final decision will be made by the Expert Panel Chair.
- Studies that are selected for inclusion will undergo abstraction and full text review by the process established for the primary evidence review. To distinguish it from the body of evidence assembled via the systematic literature search conducted through the closing date of June 30, 2007, the body of evidence from any such more recent studies will be documented separately from the evidence tables comprising data from the original search.
To explore the most appropriate search strategy and examine the sensitivity and specificity of particular search terms, an initial search was done in PubMed/MEDLINE. This search used broad medical subject heading (MeSH) terms and text words for the concepts of pediatric/young adult populations, CVD/TOD, and the risk factors. Terms were combined using the Boolean operators "AND," "OR," and "NOT," which are described briefly in Table A–3 in Appendix A of the original guideline document.
The preliminary, broad search of PubMed/MEDLINE identified in excess of 1 million citations, signaling the need to refine the search terms to identify the most relevant ones. In consultation with the Expert Panel, key refinements in the search strategy were made, including (1) the use of major MeSH terms rather than MeSH terms, where appropriate; (2) the use of title and abstract terms rather than text words; (3) a reduction of the number of terms for each concept, leaving only the most central and essential terms; and (4) the application of excluded concepts to the search (in the form of "NOT" terms).
In the final search strategy, a combination of MeSH terms, major MeSH terms, and title and abstract terms was employed to identify the full range of relevant literature. Search terms were identified to capture studies in the pediatric and young adult target populations (ages 0–21 years) that also addressed CVD/TOD and/or at least one of the risk factors. Specific search terms are provided in Table A–9 in Appendix A of the original guideline document.
A set of limits was applied to the search to help refine the results to the most useful types of studies. The first level of basic search limits included:
- Publication date: published between January 1, 1985, and June 30, 2007
- Language: English language abstract or full text
- Publication type: no editorials, letters, comments, case reports, or non-systematic reviews
Search terms and field tags used to apply these limits to the search are provided in Table A–9 in Appendix A of the original guideline document.
In addition to these basic limits, search terms were used to exclude studies examining certain out-of-scope conditions. For example, during a preliminary review of the literature, many studies were identified that focused on various pediatric conditions such as Kawasaki disease, otitis media, or congenital heart diagnoses. Through consultation with the Expert Panel, search terms were developed for the most commonly observed out-of-scope conditions. Studies containing these terms were prospectively excluded by using the Boolean operator "NOT."
Systematic Reviews, Meta-Analyses, Randomized Controlled Trials, and the Guidelines
After applying the initial limits and using terms to eliminate out-of-scope concepts, the number of results returned from the original literature searches was still in excess of 60,000 citations. Given the size of these literature results, the Expert Panel determined that part of the review would focus on certain study types that would be most useful to the Expert Panel during Guidelines development: systematic reviews (SRs), meta-analyses (MAs), randomized controlled trials (RCTs), and guidelines. Search terms and field tags identifying study type were used to select these studies in the PubMed/MEDLINE database.
Secondary studies, such as SRs and MAs, compile results from primary analyses and, in some cases, review of these study types may lessen the need to examine primary evidence on a topic. However, given the breadth of this evidence review, there were no instances in which an SR or MA captured the entire scope of interest of a critical question; therefore, this review depended on RCTs as an important source of primary data on relevant interventions.
The schematic in Figure A–1 in Appendix A of the original guideline document presents a simplified, high-level depiction of how the key search concepts were combined to achieve the overall search strategy.
In childhood, much of the evidence linking risk factors to atherosclerosis comes from epidemiologic studies. Therefore, in addition to including SRs, MAs, RCTs, and guidelines in the review, the Expert Panel determined that it was necessary for the evidence review to include major epidemiologic studies selected by the Expert Panel. These studies represent landmark longitudinal and natural history studies and other sentinel work that have provided important information and insight about atherosclerosis and CV risk in children. The major observational studies that were included are listed in Table A–4 in Appendix A of the original guideline document.
A separate targeted search of PubMed/MEDLINE was conducted to identify literature relevant to these major studies related to the risk factors for the inclusive period of the evidence review from January 1, 1985, to June 30, 2007. The observational literature was also updated by the Expert Panel using the same criteria developed for the classic evidence review. Terms used to conduct this search are provided in Table A–10 in Appendix A of the original guideline document. National Heart, Lung and Blood Institute (NHLBI) staff reviewed the titles and then the abstracts for studies to be included based on the 14 risk factors under review. When a longitudinal study reported results of the same variables at increasing intervals from the beginning of the observational period, the most recent report detailing the longest period of observation was selected for inclusion. Duplicate reports of the same results were excluded. The observational studies to be included in the evidence review were selected by the Expert Panel Risk Factor Teams.
In an evidence-based review, studies included are generally limited to RCTs, SRs, and MAs. In addition to the epidemiologic studies described above, Expert Panel members also included studies that provided important information for each risk factor, defining the context in which the Guidelines' recommendations were developed. These references are not part of the evidence tables but are identified sequentially throughout the text and will be listed in Appendix B of the original guideline document by section in numeric order, as identified in the text. Of particular importance were studies of genetic conditions impacting CV risk status and natural history studies of specific diseases known to be associated with accelerated atherogenesis.
Inclusion and Exclusion Criteria
In preparation for review of the literature, inclusion and exclusion criteria were developed by the Expert Panel. These criteria outlined additional boundaries for the review. Certain criteria were applied only by the Expert Panel, given that judgments regarding the application of these criteria required relevant clinical expertise.
- Pertained to at least one of the specified risk factors and measured at least one of the predetermined outcomes
- Related to at least one of the critical questions
- Focused on the target population (ages 0–21 years)
- For longitudinal studies and other studies with extended followup periods, the population was required to be in this age range at initiation, and this subcohort could be identified in subsequent analyses.
- For the Guidelines, the target population was required to include at least part of this age range.
- Conducted in Europe, North America, Australia, New Zealand, Japan, or Israel
- In 2004, an NHLBI-appointed Task Force published The Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents. This report included a complete review of the current evidence on this subject and detailed recommendations for managing blood pressure (BP) throughout childhood. These recommendations were used as the basic recommendations for BP management for these Guidelines and are considered complete until 2003 when the review for the report ended. The literature review for BP, therefore, was limited to January 1, 2003, through June 30, 2007; selected studies from 2008 identified by the Expert Panel that met all the criteria for inclusion in the evidence review were also included.
- Any study not meeting the above requirements was excluded from the review.
- Studies that otherwise met inclusion criteria but that were found, upon examination, to have measured risk factors in only an incidental way or as part of assessing the safety of an intervention were excluded. For example, a study of an asthma medication might measure BP to ensure that there were no adverse effects of the medication. Such studies that measured in-scope outcomes that were not linked to a risk factor condition were excluded from the review.
- Duplicate reports of findings based on the same original studies were generally excluded. For instances in which a series of studies (typically longitudinal studies or large RCTs) reported results for the same outcome measures over a period of time, the most recent studies and main results of trials were typically retained and older studies were excluded. These determinations were made individually during the review of each study.
- Studies that did not meet basic internal/external validity standards (e.g., as a result of narrowly defined patient population) were excluded.
- Studies that addressed the target population, often as part of a broader age range, but did not provide findings specific to patients in the target age range were excluded.
- Studies were excluded that on closer inspection were found not to be SRs, MAs, RCTs, guidelines, or reports from the selected epidemiologic studies.
- Studies were excluded that had an insufficient number of patients at followup to draw meaningful conclusions.
- Studies conducted in patients with diabetes focused on interventions that were related exclusively to glycemic control were excluded.
- For studies that focused on smoking as a risk factor, those that reported on interventions related to policymaking or merchant behavior were excluded.
During the review process, inclusion and exclusion criteria were modified to account for topics identified as irrelevant and certain included topics were clarified. Throughout this process, abstractors and the Expert Panel were in close contact to resolve questions regarding the application of inclusion/exclusion criteria to individual studies.
Literature Review Process
After completing electronic searches in each database, a total of 11,231 SRs, MAs, RCTs, guidelines, and major observational studies were identified for review. The distribution of search results by database and study type is presented in Table A–5 in Appendix A of the original guideline document.
Abstracts and citations for these studies were compiled and organized using Reference Manager.
Figure A–2 in Appendix A of the original guideline outlines the phases of the literature review process and the number of studies excluded at each stage. Throughout each review phase the Expert Panel provided guidance regarding the appropriate application of inclusion and exclusion criteria.
Following the review of titles and abstracts, trained abstractors conducted a full-text review of the studies and excluded additional studies. NHLBI staff also reviewed the full text of these studies and identified additional studies to exclude. Following the full-text review phase, an additional 200 studies were excluded. Citations for studies excluded at the full-text level are provided online, along with the complete evidence tables.
In addition to a review of the studies captured through the literature search process (i.e., studies published between January 1, 1985, and June 30, 2007), the NHLBI and the Expert Panel identified an additional 29 relevant studies that were published after June 30, 2007, for inclusion in the review.