Association between pacifier use and breast-feeding, sudden infant death syndrome, infection and dental malocclusion

To critically review all literature related to pacifier use for full-term healthy infants and young children. The specific review questions addressed are: What is the evidence of adverse and/or positive outcomes of pacifier use in infancy and childhood in relation to each of the following subtopics: • breast-feeding; • sudden infant death syndrome; • infection; • dental malocclusion.

Specific criteria were used to determine which studies would be included in the review: (i) the types of participants; (ii) the types of research design; and (iii) the types of outcome measures. To be included a study has to meet all criteria.

The participants included in the review were healthy term infants and healthy children up to the age of 16 years. Studies that focused on preterm infants, and infants and young children with serious illness or congenital malformations were excluded. However, some total population studies did include these children. Types of research design: It became evident early in the review process that very few randomised controlled trials had been conducted. A decision was made to include observational epidemiological designs, specifically prospective cohort studies and, in the case of sudden infant death syndrome research, case-control studies. Purely descriptive and cross-sectional studies were excluded, as were qualitative studies and all other forms of evidence. A number of criteria have been proposed to establish causation in the scientific and medical literature. These key criteria were applied in the review process and are described as follows: (i) consistency and unbiasedness of findings; (ii) strength of association; (iii) temporal sequence; (iv) dose-response relationship; (v) specificity; (vi) coherence with biological background and previous knowledge; (vii) biological plausibility; and (viii) experimental evidence. Studies that did not meet the requirement of appropriate temporal sequencing of events and studies that did not present an estimate of the strength of association were not included in the final review. Types of outcome measures: Our specific interest was pacifier use related to: • breast-feeding; • sudden infant death syndrome; • infection; • dental malocclusion. Studies that examined pacifier use related to procedural pain relief were excluded. Studies that examined the relationship between pacifier use and gastro-oesophageal reflux were also excluded as this information has been recently presented as a systematic review.

The review comprised published and unpublished research literature. The search was restricted to reports published in English, Spanish and German. The time period covered research published from January 1960 to October 2003. A protocol developed by New Zealand Health Technology Assessment was used to guide the search process. The search comprised bibliographic databases, citation searching, other evidence-based and guidelines sites, government documents, books and reports, professional websites, national associations, hand search, contacting national/international experts and general internet searching. ASSESSMENT OF QUALITY: All studies identified during the database search were assessed for relevance to the review based on the information provided in the title, abstract and descriptor/MeSH terms, and a full report was retrieved for all studies that met the inclusion criteria. Studies identified from reference list searches were assessed for relevance based on the study title. Keywords included: dummy, dummies, pacifier(s), soother(s), comforter(s), non-nutritive sucking, infant, child, infant care. Initially, studies were reviewed for inclusion by pairs of principal investigators. Authorship of articles was not concealed from the reviewers. Next, the methodological quality of included articles was assessed independently by groups of three or more principal investigators and clinicians using a checklist. All 20 studies that were accepted met minimum set criteria, but few passed without some methodological concern.

To meet the requirements of the Joanna Briggs Institute, reasons for acceptance and non-acceptance at each phase were clearly documented. An assessment protocol and report form was developed for each of the three phases of review. The first form was created to record investigators' evaluations of studies included in the initial review. Those studies that failed to meet strict inclusion criteria were excluded at this point. A second form was designed to facilitate an in-depth critique of epidemiological study methodology. The checklist was pilot tested and adjustments were made before reviewers were trained in its use. When reviewers could not agree on an assessment, it was passed to additional reviewers and discussed until a consensus was reached. At this stage, studies other than cohort, case-control and randomised controlled trials were excluded. Issues of clarification were also addressed at this point. The final phase was that of integration. This phase, undertaken by the principal investigators, was assisted by the production of data extraction tables. Through a process of trial and error, a framework was formulated that adequately summarised the key elements of the studies. This information was tabulated under the following headings: authors/setting, design, exposure/outcome, confounders controlled, analysis and main findings.

With regard to the breast-feeding outcome, 10 studies met the inclusion criteria, comprising two randomised controlled trials and eight cohort studies. The research was conducted between 1995 and 2003 in a wide variety of settings involving research participants from diverse socioeconomic and cultural backgrounds. Information regarding exposure and outcome status, and potential confounding factors was obtained from: antenatal and postnatal records; interviews before discharge from obstetric/midwifery care; post-discharge interviews; and post-discharge postal and telephone surveys. Both the level of contact and the frequency of contact with the informant, the child's mother, differed widely. Pacifier use was defined and measured inconsistently, possibly because few studies were initiated expressly to investigate its relationship with breast-feeding. Completeness of follow-up was addressed, but missing data were not uniformly identified and explained. When comparisons were made between participants and non-participants there was some evidence of differential loss and a bias towards families in higher socioeconomic groups. Multivariate analysis was undertaken in the majority of studies, with some including a large number of sociodemographic, obstetric and infant covariates and others including just maternal age and education. As might be expected given the inconsistency of definition and measurement, the relationship between pacifier use and breast-feeding was expressed in many different ways and a meta-analysis was not appropriate. In summary, only one study did not report a negative association between pacifier use and breast-feeding duration or exclusivity. Results indicate an increase in risk for a reduced overall duration of breast-feeding from 20% to almost threefold. The data suggest that very infrequent use may not have any overall negative impact on breast-feeding outcomes. Six sudden infant death syndrome case-control studies met the criteria for inclusion. The research was conducted with information gathered between 1984 and 1999 in Norway, UK, New Zealand, the Netherlands and USA. Exposure information was obtained from a variety of sources including: hospital and antenatal records, death scene investigation, and interview and questionnaire. Information for cases was sought within 2 days after death, within 2-4 weeks after death and in one study between 3 and 11 years after death. Information for controls was sought from as early as 4 days of a nominated sudden infant death syndrome case, to between 1 and 7 weeks from the case date, and again in one study some 3-11 years later. In the majority of the studies case ascertainment was determined by post-mortem. Pacifier use was again defined and measured somewhat inconsistently. All studies controlled for confounding factors by matching and/or using multivariate analysis. Generally, antenatal and postnatal factors, as well as infant care practices, and maternal, family and socioeconomic issues were considered. (ABSTRACT TRUNCATED)