Page 5 - Layout 1

Methods Study Selection

Two reviewers screened abstracts and full-text reports
Input From Stakeholders
with conflicts resolved by consensus or third-party
We formulated the population, intervention, adjudication. Studies were included if they met the
comparator, outcome, timing, setting (PICOTS) following criteria: evaluated pregnant women between
conceptual framework and Key Questions in 24 and 36 weeks’ gestation having had acute preterm
consultation with key informants during a topic labor arrested with primary tocolytic therapy; contained
refinement stage. The public was invited to provide at least one group that was administered the SQ
comments on the Key Questions. During the review terbutaline pump; and assessed one of the specified
process, we followed a research protocol we developed outcomes listed in the key questions or described a
with the clinical and methodological input of a long-term childhood outcome. Noncomparative studies
technical expert panel. The protocol followed the (i.e., case series) were assessed only for pump-related
Effective Health Care Program’s Methods Guide for harms outcomes, such as incidence of pump failure,
Effectiveness and Comparative Effectiveness Reviews. 7 missed doses, or overdose. Non-English records
without an English abstract were excluded. We also
Data Sources and Searches excluded case reports, but in a post hoc decision sought
FDA summaries of postmarketing data highlighting
We developed a peer-reviewed search strategy and
serious harms.
searched the following databases: MEDLINE In-
Process & Other Non-Indexed Citations and MEDLINE Data Extraction and Risk of Bias Assessment
(1950 to April 1, 2011); Embase (1980 to April 1,
2011); Cumulative Index to Nursing and Allied Health One reviewer extracted data into a standardized
Literature (CINAHL) via EBSCOhost (1985 to electronic form and assessed study risk of bias and
December 7, 2009), the Cochrane Library via the Wiley applicability. Extraction items included general study
interface (April 1, 2011) (including CENTRAL, characteristics (e.g., year of publication, study design),
Cochrane Database of Systematic Reviews, Database of population characteristics (e.g., inclusion/exclusion
Abstracts of Reviews of Effects – DARE, Health criteria, age, race, level of activity), intervention
Technology Assessment – HTA, and the National characteristics (e.g., dose, duration, details about
Health Service Economic Evaluation Database – NHS comparators, level of care), and outcomes with their
EED), and the Centre for Reviews and Dissemination estimates. A second reviewer verified outcomes data
(CRD) databases (January 2, 2010). Appendix A and study risk of bias assessments. Ratings for level of
provides details of the search strategies. We hand- activity, level of care, and assessments of applicability
searched the bibliographies and text of review articles, were verified by a clinical expert. Level of activity and
letters to editors, and commentaries and the reference level of care were rated based on composite
lists of included studies for additional references. We assessments across preidentified variables.
also reviewed grey literature sources and information
We assessed study risk of bias given the study design,
received from pharmaceutical companies (see
by outcome, using generic items to assess confounding
Appendixes B and C), and sought unpublished
and various types of bias (e.g., selection, performance,
information from Matria (now called Alere) Healthcare
detection bias, attrition bias). Selected items from the
about their perinatal program and associated database.
McMaster Quality Assessment Scale of Harms were
In February 2011, the FDA issued new warnings also incorporated into the risk of bias assessment for
8
against the use of terbutaline to treat preterm labor, so harm-related outcomes. Certain criteria were specific
we also accessed a summary of the FDA postmarketing to particular study designs (e.g., allocation generation
surveillance results. This decision was made post hoc. and concealment applied only to RCTs). We rated each
relevant outcome in a study with an overall risk of bias
rating designated as high, medium, or low. Outcomes
were rated as high risk of bias if there was an apparent

5