standards.11 For example, there have been multiple quality improvement efforts to reduce rates of bloodstream infections in the NICU in which multicenter groups work collaboratively to standardize and implement best practices identified through review of care at the centers with the lowest reported rates of bloodstream infections.12–14 CER, on the other hand, can be thought of as research directed at identifying best practices among the wide variation present in medical care. This is especially true when there are multiple treatments or processes available to treat the same condition. 834 Lagatta et al In contrast to the quality improvement examples cited, a CER study on preventing bloodstream infection would compare outcomes after use of differing handwashing agents, types of catheters, or approaches to line placement, while using similar risk adjustment measures that are used for benchmarking outcomes in quality reporting.11,15,16 For an outcome that is clearly positive or negative, such as line infection rates, the goal of CER is to provide evidence that can be implemented to optimize outcomes, and not necessarily directed at standardizing practice. For outcomes in which family or clinician preferences vary, or optimal treatments differ by patient subgroup, a goal of CER is to provide enough evidence to allow clinicians and families to tailor their approach to achieve the outcomes that are best for the patient. In neonatology, tailored approaches to care based on parent preferences are discussed most commonly in the context of resuscitation for extremely preterm infants,17 but could be applied to other medical decision making as well. For example, a NICU that allows more home nasogastric feedings, places gastrostomy tubes earlier, or prescribes home oxygen or apnea monitors more readily might have a shorter length of stay, but this shorter length of stay is not necessarily preferable to a family with more limited home resources. This is the rationale behind measuring a broad array of outcomes for any given intervention, as well as focusing on outcomes that are directly interpretable to families. STUDY DESIGNS FOR COMPARATIVE EFFECTIVENESS RESEARCH Any traditional clinical study design can be used to answer research questions that can be categorized as CER. Neonatal literature contains examples that highlight CER principles across categories of study designs and data sources. This article highlights 3 major categories: Randomized controlled trials (RCT), systematic reviews of existing literature, and observational studies. These examples are not meant as an exhaustive review of the literature, but rather to highlight existing neonatal research as examples of CER. Randomized, Controlled Trials RCTs are traditionally designed to evaluate the efficacy of therapies in a specific population of subjects, analyzing an intervention by comparing randomized groups to receive either a treatment or a blinded placebo under standardized conditions. RCTs are also used to determine effectiveness by comparing treatments or procedures. Successful randomization and blinding minimizes confounding, and is the reason that RCTs are considered the gold standard of clinical research. The media controversy after the SUPPORT trial makes it seem as though effectiveness trials in neonatology are new,18 but numerous neonatal RCTs are examples of effectiveness trials. Although comparisons of oxygen saturation targets are the most publicized recent clinical trials in this category,19–21 many other studies fit this description. Trials comparing modes of ventilation for preterm infants can be considered comparative effectiveness trials, because there could be no ethical or practical placebo group receiving no respiratory support. Examples include prophylactic intubation and surfactant versus nasal continuous positive airway pressure (NCPAP) at resuscitation, or noninvasive ventilation comparisons, such as synchronized nasal intermittent positive pressure ventilation versus NCPAP or high-flow nasal cannula versus NCPAP.22–24 Even in evaluations of drug therapies, neonatal literature has relevant examples, such as comparisons of bevacizumab versus laser therapy for retinopathy of prematurity,25 dopamine versus epinephrine for hypotension,26 and high versus low amino acid levels in parenteral nutrition for extremely low birth weight Neonatal Comparative Effectiveness 835 infants.27 Developing new therapies and testing them against a placebo is obviously crucial to major progress in care for critically ill infants; initial evaluations of surfactant and antenatal steroids led to enormous improvements in neonatal mortality and morbidity.28,29 However, effectiveness trials such as optimizing the timing and preparation of surfactant and antenatal steroids,30–32 and improving key components of neonatal intensive care management such as ventilation, blood pressure, and nutritional support, are needed to continue the significant improvements in mortality and morbidity seen since 2000.33,34 RCTs have limitations in the context of CER, which focuses on identifying the best treatment of available options individualized to relevant subgroups. “Standardized conditions” may have differing effects on the generalizability of trial results to a broader population. A larger sample size is often required to power a