Breastfeeding and the risk of dental caries: a systematic review and meta‐analysis

To synthesise the current evidence for the associations between breastfeeding and dental caries, with respect to specific windows of early childhood caries risk.


INTRODUCTION
Dental caries (tooth decay) is a major public health problem affecting 60-90% of school-aged children (1), with increased prevalence in children from lower socio-economic groups (2). It is caused by multi-factorial and complex interactions between cariogenic bacteria in the mouth with dietary carbohydrates that produce acids and demineralise the teeth (2). The pain and infection caused by dental caries can be extremely distressing and can impact on quality of life and ability to function (3), lead to lost productivity and involve high health care costs (4) including general anaesthesia for treatment of severe cases. This accounts for one of the most common causes of child hospitalisation in industrialised countries (5) and is among the most common causes of avoidable child hospitalisations (6). Early loss of deciduous dentition can lead to ongoing dental problems in the permanent dentition.
The evidence concerning infant feeding as a risk factor for dental caries is inconsistent. Dental caries risk is related to the carbohydrate content of breast milk or formula along with factors which determine the length of contact between breast milk or formula and the erupted dentition (i.e. frequency of feeding, and feeding practices which result in pooling of breast milk or formula around the teeth surfaces, such as feeding babies to sleep). The central determinant of caries risk, however, is the age of colonisation and levels of cariogenic bacteria (e.g. Streptococcus mutans) (7) in an infant's mouth. Earlier and denser oral colonization by cariogenic bacteria are related to increased caries risk (8). Breast milk, in contrast to formula, contains breast-specific Lactobacilli and substances, including human casein and secretory IgA, which inhibit the growth and adhesion of cariogenic bacteria, particularly oral Streptococci (9,10). The risk of dental caries is also dependent on the presence of teeth and rises with increasing number of teeth. Risk also changes as the infant's diet starts to include foods and drinks other than breast milk or formula, depending on the carbohydrate content, acidity and consumption frequency of the introduced diet.
The important aspect of timing of tooth eruption for our systematic review is that the deciduous teeth most at risk of early childhood caries (eight upper and lower central and lateral incisors) start to erupt at 6 months and are fully erupted by 12 months. The next most vulnerable deciduous teeth (four upper and lower first molars) erupt between 13 and 19 months, the remainder are erupted by 33 months (11).
Current WHO breastfeeding guidelines recommend exclusive feeding for the first 6 months of life and complementary breastfeeding up to 2 years (12). Although the UNICEF calculated global prevalence of breastfeeding at 12 months from 62 countries is 74%, this figure hides the underlying heterogeneity between countries (13). As opposed to low income countries, the duration of total breastfeeding in high/middle income countries is shorter with only 21% of US mothers breastfeeding at 12 months (14) and similar rates in the UK (13), Canada (5) and Australia (15). National guidelines in high/middle income countries, where the risk of infant morbidity and mortality from gastrointestinal disease is relatively low, recommend breastfeeding for at least 12 months (16). Thus, investigating windows of exposure before and after 12 months of age is relevant to breastfeeding guidelines and practices as well as timing of tooth eruption.
The relationship between breastfeeding and dental caries has been systematically (17) and narratively reviewed (18)(19)(20) with conflicting results between studies. There is controversy about what constitutes the best form of infant feeding to prevent dental caries and promote optimal dental health (21). Consequently no definitive optimal weaning times or breastfeeding practices have been determined to specifically address the risk of dental caries.

AIM
To summarise the current evidence for the association between breastfeeding and dental caries with specific reference to exposure windows and breastfeeding practices.

Search strategy
We identified human English language studies through systematically searching electronic databases: PubMed Central, CINAHL and EMBASE from inception to the present. Our exposure of interest was breastfeeding as compared to formula or other feeding. Our outcome of interest was the development of dental caries in deciduous or permanent teeth. An extensive list of search terms was used and is reported in Table 1.
We checked reference lists of all primary studies and review articles for additional references. The titles and abstracts were independently reviewed for initial inclusion by two researchers (RT and GB). Disagreement was resolved by discussion and if consensus could not be reached, a third author (CL) made the final decision. #1 "Breast Feeding" #2 "Milk, Human" #3 Breast AND Feed* #4 Breast-fe* #5 Infant fe* #6 Infant nutrition* #7 S1 OR S2 OR S3 OR S4 OR S5 OR S6 #8 dental caries #9 tooth decay #10 early childhood caries #11 nursing bottle caries #12 S8 OR S9 OR S10 OR S11 #13 S7 AND S12 **For #13 limit to 'Human'

Eligibility criteria
We included observational and experimental studies published in full text. We included children and adolescents from both general and high-risk populations (e.g. low socioeconomic communities). Dental caries as reported by appropriately qualified practitioner/researchers, a parent or through health records databases were included. We excluded participants who were born prematurely (<36 weeks gestation) because these infants are often fed by other sources and can have complicated medical interventions.
Assessment of quality and risk of bias Two researchers (RT and GB) independently conducted a quality assessment of each study using the Newcastle-Ottawa Scale (NOS) (22). Study quality was graded on a scoring system (see Tables 2-5 for key criteria). Differences in assessment and grading were resolved by discussion with a third researcher (CL).
The assessment of risk of bias was guided by the GRADE system for rating the quality of the evidence of observational studies (23).
Literature review identified key confounders that should be controlled for in breastfeeding and dental caries studies: socio-economic status, age, mother's educational level, number of teeth, and exposure to sugar in the diet (food or other liquid).

Data extraction
We extracted: study design; study country; age range of children; number of children; exposure and outcome definitions; how the outcome data were measured; effect estimates; confounders included in analysis; sub-group analysis; interactions; and findings.
Assessment for meta-analysis Exposure and outcome definitions and effect estimates (odds ratios (OR), relative risks, prevalence ratios) with 95% Confidence Interval (95%CI) were abstracted where available for inclusion in a meta-analysis. Given the biological plausibility of the potential associations, we aimed to assess exposure to breastfeeding in two specific time windows: (i) Up to 12 months of age (upper and lower incisors present) and (ii) Beyond 12 months of age (other teeth erupting up to 33 months-increased risk of caries). As there were very few mothers who exclusively breastfed infants until 12 months or beyond, within these time windows we categorized studies into: (i) Never breastfed compared to any breastfeeding and (ii) More versus less breastfeeding. This category was created to include all  We performed meta-analysis if there were three or more studies in each time window and category of breastfeeding. Random effects meta-analyses were performed if the heterogeneity (I 2 ) was >25%. Heterogeneity was considered to be high, and results unreliable if I 2 values were >75%. We were unable to quantitatively assess for publication bias as no group contained more than 10 studies. Studies not meeting these criteria were qualitatively assessed.

Search results
Electronic literature search (2 October 2014) and manual search found 480 peer-reviewed scientific articles after duplicate papers were removed. Of these, 366 were excluded after abstract review for failing to meet the eligibility criteria. A large number of these papers were not related to breastfeeding or dental caries, were not in English or were not original research. Of the remaining 114 full text articles, 51 were excluded as: (i) they did not assess the relevant exposure (breastfeeding) and outcome (dental caries) or (ii) all feeding types were analysed together or (iii) data were duplicated in more than one paper or (iv) no analysis was reported or studies lacked control or comparator groups [ Fig. 1 (24)]. In total 63 papers were included.

Quality assessment
Tables 2 3, 4, and 5 detail the NOS score assigned to each included study. The cohort and cross-sectional studies that were embedded in RCTs of a range of breastfeeding promotion interventions (25-30) scored highly as the study designs overcame many sources of bias and reporting limitations that were apparent in the other cohort, case-control and cross-sectional studies. Other cohort studies were weakened by the method used to ascertain infant feeding practices (self-report) which subjected them to recall bias, recruitment of children through oral health services (selection bias), lack of reporting of the absence of caries at the commencement of the study (ascertainment bias), loss to follow -up and accounting for these participants (attrition bias), and lack of controlling for confounders. Case-control study designs were inherently subject to recall bias when ascertaining infant feeding practices. Furthermore, cases and controls were not representative of the broader population as they were recruited in settings where children were likely to have caries. Selection bias was also a problem as the selection of controls was not clearly described. Crosssectional studies were the weakest but most common study design. The studies which scored <4 were classified as unsatisfactory due to major limitations in study design and reporting. Studies that scored 4 were classified as satisfactory, however, all of these studies lacked consideration of key confounders. In the higher quality studies (≥5) there were limitations in how exposure was ascertained as many studies used self-report questionnaires (recall bias).

Meta-analysis
We meta-analysed the small number of studies which included statistical effect measures.
Breastfeeding up to 12 months of age One prospective cohort (34) and four cross-sectional studies (48,52,59,70) reported odds ratios for the association between children who were exposed to more versus less breastfeeding up to 12 months (OR 0.50; 0.25-0.99, I 2 86.8%) (Fig. 2). There were not enough studies to perform metaregression for formal investigation of this heterogeneity. There appeared to be differences, however, based on the comparison groups of the included studies. The two studies which compared ever breastfeeding in the first 12 months with never breastfeeding (48,70), both showed a marked protective effect of breastfeeding on dental caries compared with other feeding. Whereas the three studies which compared a longer duration of breastfeeding in the first 12 months to a comparison group which included children who had had some exposure to breastfeeding did not (34,52,59). A meta-analysis on this three study subgroup found an OR of 0.92; 0.69-1.23, I 2 0% (Fig. 3).
Breastfeeding after 12 months of age Two cohort studies (33,34), one case control study (40) and four cross-sectional studies (52,65,75,78) reported odds ratios for the association between more or less breastfeeding after the age of 12 months and dental caries. The comparison groups for these studies included both those who had never been breastfed and those who had been breastfed for shorter durations. The pooled estimate was OR 1.99; 1.35-2.95, I 2 69.3% (Fig. 4).  Only two studies (26,80) reported prevalence ratios so these could not be meta-analysed.
Nocturnal breastfeeding in those breastfed longer than 12 months One cohort (36), one case-control (40) and three crosssectional studies (67,84,86) reported odds ratios for the association between more versus less nocturnal breastfeeding and the risk of dental caries amongst the subgroup of children breastfed longer than 12 months. The pooled estimate was OR 7.14; 3.14-16.23, I 2 77.1% (Fig. 5).

Narrative synthesis
The majority of studies (n = 46) were not included in the meta-analyses due to methodological differences in the measures of exposure and outcomes, or reporting of correlational analyses only.

Studies embedded in randomised controlled trials (RCTs)
It is not ethical to conduct randomized trials assigning participants to breastfeeding and non-breastfeeding groups in order to more definitively assess the association between breastfeeding and dental caries. However, a number of RCTs have been conducted that investigated the impact of breastfeeding promotion programmes (25)(26)(27)(28)(29)(30). In a RCT of an intervention that provided monthly advice on healthy feeding practices over 12 months via home visits in Brazil (25,27,30) the intervention group demonstrated a lower incidence of caries at 12 months (OR = 0.52, 0.27-0.97, p = 0.03) and 4 years (RR = 0.78, 0.65-0.93, p = 0.004). Investigating breastfeeding frequency at 12 months, the study also found a doubled risk of caries when feeding frequency was 3-6 times/day (RR = 2.04, 1.22-3.39, p = 0.000) and ≥7 times/day (RR = 1.97, 95%CI 1.45-2.68, p = 0.000) compared to 0-2 times/day. Analyses were adjusted for maternal schooling level, daily meals, bottle use for fruit juice/soft drinks, consumption of high density sugar and number of teeth. Another birth cohort study nested in an intervention conducted through maternal health centres in Brazil (26) found that, in adjusted regression models, as breastfeeding continued beyond 6 months the prevalence ratio of caries in breastfed children increased (compared to breastfeeding <6 months) but was only significant when still breastfeeding at ≥24 months: 6-11 months (PR = 1.45, 95% CI 0.83-2.53); 12-23 months (PR = 1.39, 95%CI 0.73-2.64); ≥24 months (PR = 1.85, 95%CI 1.11-3.08). A birth cohort study nested in a breastfeeding promotion intervention in Belarus found no significant difference in caries incidence or prevalence in the intervention group when children were aged 6.5 years (28,29).

Breastfed versus formula fed
Studies that examined ever versus never breastfed children reported a range of findings. Six cross-sectional studies Articles assessed for meta-analyses = Breastfeeding at 12 months: n=5 Breastfeeding beyond 12 months: n=6 Breastfeeding on demand or nocturnally: n=6 reported no significant difference in the prevalence of caries between the two groups (49,61,(72)(73)(74)83); one cohort and one cross-sectional study reported significantly lower caries in breastfed children (32,57); one cross-sectional study found a lower adjusted caries risk in breastfed versus bottlefed children (OR = 0.61, 95%CI 0.39-0.97, p = 0.038) (70); one cohort study reported higher caries increment in breastfed children between 12 to 18 months but the association disappeared in the multivariate analysis (35); one cross-sectional study reported an increased risk of dental caries in ever breastfed children of borderline significance (p = 0.08) (77); and one cross-sectional study found a lower adjusted caries risk in breastfed versus bottlefed children.

Breastfeeding duration
Three of four cohort studies found that breastfeeding beyond 12 months was correlated or associated with increased caries Odds Ratio More vs Less Breastfeeding up to 12 months and risk of caries   prevalence compared with shorter durations of breastfeeding. Chaffee et al. (26) found that the adjusted prevalence ratio of caries in children breastfed ≥24 months was 2. Odds Ratio

More vs Less Breastfeeding up to 12 months and the risk of caries
Nocturnal breastfeeding beyond 12 months and risk of dental caries The few studies that controlled for confounding factors found decreased caries risk with shorter breastfeeding duration (6-12 months) compared to longer duration (>13 months) (26,34,45,76,80) and increased risk of caries if breastfed <6 months (31,48).

Breastfeeding on demand and nocturnally
In addition to the meta-analysed studies, a number of crosssectional studies reported significant correlations between infants/children breastfed during the night (44,67), on demand (68) or sleeping with a nipple in the mouth (60,71,76) and increased prevalence of dental caries. One cohort study found an increased adjusted risk of dental caries with increased daily breastfeeding frequency including nocturnal feeding (25).

DISCUSSION
Qualitative assessment of studies investigating breastfeeding up to 12 months of age suggested that children who were exposed to more breastfeeding (longer duration) compared to less or no breastfeeding were protected from dental caries. Meta-analysis of five studies also found reduced risk of dental caries in children breastfed more versus less up to 12 months, however, the heterogeneity between studies was too high to make the estimate reliable. In contrast, children who were breastfed beyond 12 months had an increased prevalence of dental caries. Amongst those who continued to be breastfed after 12 months, there was a further increased risk of caries in children who were breastfed nocturnally.
Three elements are essential for dental caries to occur: a tooth, cariogenic bacteria (e.g. Streptococcus mutans) and substrate for the bacteria (sugar) (2). The risk of developing dental caries changes as factors associated with each element change. The first tooth usually erupts in an infant's mouth between 6 and 12 months of age. As each tooth erupts the risk of developing dental caries increases, hence age and number of teeth increases risk. Cariogenic bacteria are transmitted to the child via close contact with the mother's saliva (88) but their levels and cariogenicity vary between individuals (2) depending on maternal bacterial levels, maternal caries prevalence, oral hygiene practices and exposure to dietary sugars (21). Breast milk is known to contain immunomodulatory factors along with a rich microbiome which is responsible for establishing normal intestinal flora (89). Initial protection from dental caries may be mediated through establishment of a healthy oral microbiome in infants through exposure to breastfeeding and contact with skin and breast milk microbiomes. Additionally, the child's oral microbiome changes over time with the emergence of new teeth. The essential substrates for cariogenic bacteria are simple carbohydrates (sugars) which can be in a range of forms (e.g. lactose, sucrose, glucose). The longer these sugars are in contact with teeth, the higher the risk of dental caries. The amount of carbohydrate (cariogenicity) contained in the different milks and formulas may also help to explain the different results we found before and after 12 months of age. The cariogenicity of human breast milk has not been extensively examined under in vivo conditions, however animal studies suggest that at high frequency exposures, human breast milk has greater cariogenicity compared to bovine milk but less than infant formula (90,91). Relative cariogenicity of breast milk will also depend on the comparison group. Below 12 months it is usual to feed infants either breast milk or formula which have around the same carbohydrate content. After 12 months, however, children in high income countries are often weaned onto cow's milk which has half the carbohydrate content of human milk. However, each element is subject to modification by risk factors such as socio-economic status, maternal educational level, maternal oral health, maternal smoking status, position in birth order, sugars in diet, oral hygiene and exposure to fluoride (2).
Breastfeeding duration, frequency of breastfeeding and nocturnal breastfeeding during sleep are most often analysed as separate breastfeeding behaviours, however they are inter-related. Nocturnal breastfeeding is often used to comfort an infant or child who may then fall asleep with the nipple in their mouth. In this position, the tongue fills the mouth and holds the breast milk against the surfaces of the teeth, thereby prolonging the exposure of the substrate to the cariogenic bacteria that are attached to the teeth surfaces and hence increasing the risk of dental caries. It is possible that children breastfed beyond 12 months are also engaging in nocturnal breastfeeding but the modification of dental caries risk by infant feeding practices has not been examined in depth in any of the studies included in this review. In addition, children >12 months are no longer being exclusively breast or bottle fed and the diet is expanding to include other fluids and solids. It has been reported that children who are breastfed for longer durations also have more frequent cariogenic food intakes (25,53,58). Oral hygiene practices to remove bacterial plaque are important as more teeth erupt to reduce the risk of dental caries. Only a few studies included in this review controlled for key confounding factors and this may have resulted in an overestimation of the role of prolonged, frequent and nocturnal breastfeeding in the development of dental caries. Until the dietary and oral hygiene details of these children are controlled for we cannot be certain whether prolonged, frequent or nocturnal breastfeeding can be principally associated with early childhood caries. This is the first systematic review of breastfeeding and dental caries that includes critical exposure windows, limited meta-analyses and a range of study types. We provide quantitative evidence that is suggestive of the potentially protective effects of breastfeeding from dental caries up to 12 months, but higher risk of dental caries in children breastfed beyond 12 months, frequently, and/or nocturnally. However, there is high heterogeneity between the studies included in the meta-analyses (possibly due to differing comparison groups) and lack of controlling for key confounders (e.g. other foods/drinks in the diet, oral

CONCLUSION
Breastfeeding up to 12 months of age is not associated with an increased risk of dental caries and in fact may offer some protection compared with formula. However, children breastfed beyond 12 months, a time during which all deciduous teeth erupt, had an increased risk of dental caries. This may be due to other factors which are linked with prolonged breastfeeding including nocturnal feeding during sleep, cariogenic foods/drinks in the diet, or inadequate oral hygiene practices. Further research with careful control of pertinent confounding factors is needed to elucidate this issue and better inform infant feeding guidelines. As per recommendations from previous reviews (17,19), the introduction of food sources to infants should be low in simple carbohydrates (sugars) and oral hygiene practices should start with the eruption of the first tooth so that bacterial plaque is removed from tooth surfaces to reduce the risk of dental caries.

CONFLICT OF INTEREST STATEMENT
Preparation of the manuscript was assisted by funding from the WHO, which had no part in determining the outcomes or presentation of findings. None of the authors has any conflicts of interest to declare. Incidence density ratio = ratio of incidence density of those exposed to those not exposed to the particular independent variable concerned.