Introduction

Although type 1 diabetes mellitus (T1DM) can be diagnosed at any age, it is one of the most prevalent diseases in children [1]. It is reported that about 79,000 children aged 14 or under the age of 14 years are suffering from T1DM worldwide annually [2]. The peak of T1DM occurrence is between 5 and 7 years of age and at or near puberty [3]. T1DM is correlated with major chronic disease risk factors and increases morbidity and mortality [4, 5]. Death under 30 years usually results due to acute complications of diabetes, including diabetic ketoacidosis and hypoglycemia [6]; and cardiovascular disease is the major reason for death later in life [5, 6]. The basis of diabetes care is the identification of the risk factors and then using the appropriate treatment [7]. Genetics and exposure to environmental factors [8] may play a major role in the occurrence of type 1 diabetes. Among environmental risk factors, a significant one that may modify the risk of type 1 diabetes in early childhood is dietary quality. Limited breastfeeding duration and early exposure to complex dietary proteins are identified as risk factors for advanced beta-cell autoimmunity or type 1 clinical diabetes [6]. Supplementing breast milk with a highly hydrolyzed milk formula would reduce the cumulative incidence of diabetes-associated autoantibodies in such children [9]. All the above are factors that researchers will also investigate in order to evaluate the function of diet during early infancy and development of T1D, as they may intervene and serve as confounders as possible. However, the correlation of dairy products with health outcomes in children is not well-understood [10]. Also, there are inconsistencies about the impact of dairy on diabetes [11]. Therefore, the purpose of this research was to explore this main question, Can dairy intake increase the risk of T1DM? We aim to determine the association between dairy products intake and T1DM risk using observational and cohort evidences.

Method

Search strategy

We systematically searched PubMed and Scopus up to 22 February 2020 with the following keywords: ((dairy product[tiab] OR milk[tiab] OR yogurt[tiab] OR cheese[tiab] OR kefir[tiab] OR butter[tiab] OR cream[tiab] OR “Dairy Products”[Mesh] OR Milk[Mesh] OR Yogurt[Mesh] OR Cheese[Mesh] OR Kefir[Mesh] OR Butter[Mesh] OR “Ice Cream”[Mesh]) AND (“type 1 diabetes”[tiab] OR “type 1 diabetic”[tiab] OR “type 1- diabetes mellitus”[tiab] OR “diabetes mellitus”[tiab] OR “diabetic patients”[tiab] OR DM[tiab] OR T1DM[tiab] OR TIDM[tiab] OR “insulin dependent”[tiab] OR IDDM[tiab] OR T1D[tiab] OR “Diabetes Mellitus, Type 1”[Mesh] OR “Diabetes Mellitus”[Mesh]) AND (“population-based” OR prospective OR “case control” OR longitudinal OR follow-up OR cohort OR retrospective OR nested OR “Longitudinal Studies”[Mesh] OR “Prospective Studies”[Mesh] OR “Case-Control Studies”[Mesh] OR “Cohort Studies”[Mesh] OR “Retrospective Studies”[Mesh])). Besides, a hand search of references of the published papers was done to detect other relevant articles. In Fig. 1, the details of the search strategy are illustrated. We also performed the systematic search for the second time in order to find any recent possible articles on 20 May 2020.

Fig. 1
figure 1

Flow chart of the number of studies identified and selected into the meta-analysis

Full size image

Eligibility criteria

Two trained reviewers (MSH and LSH) separately examined the eligibility of the studies twice. Studies were included if they met the following criteria: (a) observational study design; (b) involved subjects with <18 years old; (c) adequate information about any types of dairy products and the risk of type 1 diabetes mellitus; (d) publications that had provided estimates of relative risks (RRs) (odds ratios (ORs), hazard ratios (HRs), or rate ratios) with corresponding 95% confidence intervals (CIs); and (e) published in English.

Exclusion criteria

Totally, 1457 articles were found in our initial search. We excluded 856 articles by reading the title and abstract. The other 194 papers were excluded because of the following reasons: age > 18 (n = 84), type 2 mellitus (n = 3), review studies (n = 4), and the article without sufficient data for outcomes (n = 103).

Data extraction

Two independent reviewers (MSH and FDj) extracted the data. Any disagreements and differences were resolved by the study supervisor (SS-b), if necessary. The following information of studies was extracted: the first author’s last name, date of publication, country, participants’ age range, gender, sample size, number of cases, duration of follow-up, method of measurement of exposure and outcome, comparisons, and ORs or RRs for type 1 diabetes mellitus.

Quality assessment

The quality of included studies was evaluated by means of the Newcastle-Ottawa Scale [12]. For cohort and case-control studies that were included in the analysis, we used their own specific methods. The NOS allocates a maximum of nine points to each study: four for selection, two for comparability, and three for assessment of outcomes (nine represented the highest quality). Any inconsistencies were set by discussion.

Statistical analysis

Random effects models were applied in the current meta-analysis. Therefore, the overall effect size was calculated. Between-study heterogeneity which refers to the variation in study outcomes among studies was assessed using Cochran’s Q test and I2 statistic. Sensitivity analysis was conducted to examine changes in pooled effect size when one study had been removed. Publication bias was assessed by visual inspection or using Egger’s regression asymmetry test. All statistical analyses were done through Stata software, version 12.0 (Stata Corp LP, College Station, TX, USA). A p value of 0.05 was considered as statistical significance.

Results

Study selection

We found 1457 publications in our initial search. Of those, 400 records were duplicates. We reviewed titles and abstracts of all remaining publications, and as a result, another 856 publications were excluded, yielding 201 studies for full text assessing (Fig. 1). Of those, 7 publications provided sufficient information for the present meta-analysis. Thus, 6 case-control studies [13,14,15,16,17,18] and 1 cohort [19] with 5067 participants were included in the final analyses.

Characteristics of the included studies

Most included studies reported the association of dairy products consumption and risk of type 1 diabetes in children, and only one study assessed the risk of beta-cell autoimmune as an outcome.

General characteristics of the studies are provided in Table 1. These studies were published between 1991 and 2012 and were conducted in Finland [16,17,18,19], Germany [15], Sweden [13], and the UK [14]. All of the studies included both genders. The exposure assessment tool in six studies was the dairy products consumption questionnaire, and only one study was relied on 3-day food recall. All studies were at high quality (≥7 stars) on the basis of the Newcastle-Ottawa Scale (Table 2).

Table 1 The characteristics of included studies of dairy product intake and type 1 diabetes risk1
Full size table
Table 2 Application of methodology criteria to case-control studies⁋
Full size table

Our meta-analysis was performed on 7 studies. We extracted 14 effect sizes for type 1 diabetes risk. The overall effect of an association between dairy products consumption and risk of T1DM is illustrated in Fig. 2. The analysis indicated that dairy products consumption increased the risk of T1DM (RR: 1.04, 95% CI: 1.01, 1.08), with high heterogeneity, (I2 = 76.6%, P heterogeneity < 0.001). Based on the visual inspected funnel plot (Fig. 3) and Egger test, the studies did not have publication bias (p = 0.161). Sensitivity analysis results showed that excluding any of the studies was not significantly changed.

Fig. 2
figure 2

Forest plot of the association between dairy product consumption with risk of type 1 diabetes in children. OR stands for odds ratio

Full size image
Fig. 3
figure 3

Funnel plot for dairy product intake and type 1 diabetes risk

Full size image

Discussion

The current meta-analysis revealed a significant linkage on the consumption of dairy products and increased risk of T1DM in children with considerable heterogeneity. In 2006, Multinational Project for Childhood Diabetes (DIAMOND) was indicated that Finland, Sweden, and the UK are the first, third, and fifth, respectively, with the very high incidence of childhood type 1 diabetes, and Germany was classified as a high-risk group [20]. Culturally, racial variation in communities of European origin has shown a higher incidence compared to non-Europeans but may cause these differences, particularly among long-term immigrants to European countries [21]. The most important risk factors for type 1 diabetes as autoimmune diabetes are genetic, family history, and environment [22]. Although the evidences proved that the diet is a potential risk factor for the induction of diabetes autoantibodies in children [17, 23], whether dairy products protect against T1DM or not is a controversial issue. Surely, kind of milk consumption pattern and specially fresh cow milk in the north European population (Finland, Sweden, the UK) [24] is an important factor in diabetes incidence and prevalence. Previous epidemiological data have shown the relationship between dairy product consumption and risk of T1DM. In line with our study, Suvi et al. found that high intakes of dairy products during childhood may be diabetogenic in children with type 1 diabetes [16]. In 2018, Canada Clinical Practice published guidelines about reducing the risk of developing diabetes, and suggested nonlinear inverse associations were observed for total dairy products and yogurt, with most of the benefit being observed when increasing the intake of total dairy products from little to no dairy up to 300 to 400 g/day or yogurt up to 120 to 140 g/day, above which there was no further benefit [25].

As our knowledge, accumulating evidence-supported dairy products intake can increase the risk of type 1 diabetes [14, 15, 17, 18, 26]. Numerous mechanistic pathways suggested about the correlation between dairy products consumption and T1DM risk [27]. Most dairy products are milk based [28]; therefore, we have tried to point the mechanisms for the impact of milk on T1DM risk. Moreover, a series of studies have shown that children with newly diagnosed type 1 diabetes had elevated levels of antibodies, in particular to food antigens, lactose intolerance, and cow’s milk proteins [17, 29,30,31,32]. The role of genetics, as one of other risk factors in T1DM occurrence, has been of recent interest. T1DM as a chronic immune-mediated disease with a subclinical prodromal period is characterized by selective loss of insulin-producing β-cells in the pancreatic [16] islets in genetically susceptible subjects. Auto-reactive T cells, CD4 and CD8 cells, and a series of auto-antigens like glutamic acid decarboxylase (GAD) have been implicated as active players in β-cell destruction [19, 33]. The issue of whether there is any primary autoantigen in T1D has remained controversial. Given that there are two major HLA haplotypes conferring disease susceptibility, i.e., the DR3–DQ2 haplotype and the DR4–DQ8 haplotype, one may assume that there will be at least two primary antigens in T1D [16, 22, 34]. The first signs of β-cell autoimmunity might appear already during the first year of life [33]. The enormous studies reported that a strong connection between the longitudinal consumption of cow milk in the children and the development of advanced β-cell autoimmunity is related to several previous case-control and cohort results with endpoints varied from early pre-type 1 diabetes to clinical disease [16, 19, 35]. Lamb et al., clearly explained that higher intakes of cow’s milk may promote progression to type 1 diabetes in children with autoimmune islets [36], an effect that could be induced by certain fatty acids in cow’s milk and meat, such as myristic [34]. Also, lipid-mediated signals can play an important role in lipotoxicity induced by fatty acids [37]. A nested case control determined that there was no relationship between the consumption of sour milk products and cheese with advanced β-cell autoimmunity [17]. In this article, the mechanism of the effect of sour milk and cheese on T1DM was mentioned kind of protein that is partly hydrolyzed to smaller peptides and amino acids. Familial history is another risk factor for type 1 diabetes [38]. Parkkola’s article indicated the children with familial type 1 diabetes have an autoantibody profile, implying similar pathogenic disease mechanisms [39] Further, the duration of breastfeeding has the potential for being T1DM [40].

The recent trial published that cow’s milk does not play a critical role in the development of type 1 diabetes [41]. It was reported that camel milk is safe and efficacious in improving long-term glycemic control, with a significant reduction in the doses of insulin in type 1 diabetic patients [42]. To dispose the controversy in evidence with about effect of cow milk on diabetes mellitus [16, 19, 35, 41], a systematic review published in 2017 supported that consumption of camel milk can decease blood sugar and insulin resistance [43]. Another hypothesis was discussed by Sørensen et al. that found an association between higher serum 25-hydroxyvitamin D in late pregnancy and lower risk of type 1 diabetes in offspring [44]. Dairy products are known to be rich in vitamin D. This has been demonstrated that vitamin D changes the balance of the T cell response in the body towards downregulation of the T-helper-1 immune response [45] and beneficial factor as it plays an important function in regulating the immune system, as well as diabetes-relevant metabolic pathways.

Our meta-analysis has several strengths; for example, our search strategy was very accurate and covered multiple databases. Further, our statistical examinations indicated no evidence of publication bias in our analyses, and, finally, to the best of knowledge, this is the first meta-analysis to be performed in this regard, although some limitations exist. At first, these studies that are selected to be included have heterogeneous risk measurement methods. The high heterogeneity is another limitation that could impact our findings powerfully. Second, owing to the small number of studies, there was no ability to better examination of the association between childhood dairy products consumption with risk of T1DM. Third, most of the studies presented some form of bias, and thus, it is hard to reach a certain conclusion. Another limitation that is necessary to mention is the quantity of milk that is different in various studies.

Conclusion

We found a positive association between dairy products consumption and odds of type 1 diabetes mellitus in children. Our team suggested that these findings need to be confirmed by larger trials in order to conclusively determine any relationship between dairy products intake and T1DM.