Epidemiology and risk factors of hypovitaminosis D in a cohort of internationally adopted children: a retrospective study

Background Predictors of hypovitaminosis D were extensively studied in the adult population, leading to an approximately complete understanding of them, while there is a lack of studies in the pediatric population, especially in migrant and internationally adopted children. In this retrospective study, we tried to identify the major laboratory predictors of hypovitaminosis D in a cohort of internationally adopted children. Methods Data were extracted from the database of the “Ethnopediatrics Outpatient Clinic” of the “A. Gemelli” Foundation University Hospital in Rome, Italy. Our study included 873 children evaluated from March 2007 to May 2016. Analysis of variance, chi square test, t test and multivariate logistic regression were performed, a “p” value < 0.05 was considered significant, with a confidence interval of 95%. Results We did not find any significant correlation between Vitamin D and Calcium, Phosphates or Magnesium levels within the population we examined. Moreover, parathyroid hormone is not a good predictor of Vitamin D Status. Conclusions Considering the strong influence Vitamin D status has not only on bone health, but also on general well-being, it is due to perform a vitamin D assessment as soon as possible, especially in internationally adopted children. Electronic supplementary material The online version of this article (10.1186/s13052-018-0527-4) contains supplementary material, which is available to authorized users.


Background
Vitamin D (VD) is an essential nutrient with hormone-like activity, which regulates calcium and bone metabolism. Moreover, VD has been recently associated with immune status, autoimmune disorders, infectious diseases, decreased response to corticosteroids and psychiatric diseases [1][2][3][4][5].
These associations highlighted the need for a definition of VD optimal values, and this is particularly true for the pediatric population, because of the importance of VD for the bone health and the immune status during the tricky phase represented by the growth period.
Serum 25-hydroxy-vitamin D (25(OH)D) concentration is currently the marker of choice, as it is the major circulating form of VD and its value reflects both the amount produced in the skin after sun exposure and that assumed with food [6].
Limited data on the prevalence of hypovitaminosis D among internationally adopted children, a population which could be considered "at risk" for geographical origin and a housing system not always subdued to an optimal surveillance, are available, and there is no study assessing parathyroid hormone (PTH) levels in these groups. The relationship between serum concentrations of 25(OH)D and PTH has been considered as a possible way to define VD status, since PTH increases when vitamin D and, consequently, calcium decrease. Studies performed in adults, and very few studies in pediatric subjects, have showed conflicting results about this inverse relationship [27].
Therefore, it would be of utmost importance to define VD levels under which PTH begins to increase, and if it could be considered a marker of bone health or immune status.
We hypothesized that the prevalence of biochemical rickets and hyperparathyroidism in a population of internationally adopted children was higher than in other children subpopulations. Hence, we evaluated the prevalence of hypovitaminosis D in internationally adopted children. Moreover, we tested if some of the population characteristics, such as age, BMI, sex, housing solution before adoption, season and month of observation, country of origin, could be considered risk factors for hypovitaminosis D in this particular cohort, aiming to find a model applicable to all our patients.
In order to define the prevalence of biochemical rickets, we considered the relationship existing between Vitamin D levels and serum calcium, serum phosphates and serum alkaline phosphatase levels. We also tried to define the relationship between VD status and PTH levels in internationally adopted children.

Characteristics of the study population
We retrospectively gathered data provided by medical records of the Ethnopediatrics outpatient clinic of the "A. Gemelli" University Hospital in Rome, Italy. Inclusion criteria of our study were: being an internationally adopted child (age [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15]  Missing data about physical characteristics, due to the refusal of the child to undergo physical evaluation at the time of the first visit, were treated as "missing", as well as missing data about clinical history of the child, due to a lacking anamnesis, were. Retrieving the data: Biochemical analysis methods and normal values Blood samples were collected by venipuncture in the morning and routinely analyzed at the Hormone Laboratory ("A. Gemelli" University Hospital).
Serum 25(OH)D was measured by ChemiLuminescence ImmunoAssay (CLIA, Siemens) with an intra-and inter-assay variation of 5%. This method does not allow to differentiate D 2 and D 3 forms. The lower detection limit of the assay was 7 ng/mL. We entered in the database every value indicated as < 7 ng/mL within the test results as 6 ng/mL. Normal values are 31-100 ng/mL. Serum intact parathyroid hormone (PTH) was assayed by ElettroChemiLuminescence ImmunoAssay (ECLIA, Roche) with an intra-and interassay CVs of 5% (normal values 9-65 pg/mL). Serum calcium (Ca) and phosphates (P) were determined by colorimetric assay (COBAS, Siemens). Serum calcium normal values are 8.8-10.8 mg/dl for all ages, while serum P values are 4-7 mg/dL when the child is younger than 10 years, or 2.5-4.5 mg/dL when the child is older than 10 years. Intra-and interassay CVs were 2.7 and 2.9%, respectively. Alkaline phosphatase (ALP) was measured by kinetic before 2011, normal values < 750 IU/L for children younger than 10 years, < 1000 IU/L for children older than 10 years. A kinetic photometric test IFCC (AMP) after 2011 (Siemens) was used instead after 2011, normal values were 40-300 IU/L for children aged < 13-years-old, while for children older than 13 years it is necessary to differentiate between males (normal values 40-309 IU/L) and females (normal values 40-187 IU/L). CVs intra-and interassay was 4% respectively.
Missing data, due to unavailability of reagents at certain time in our laboratory, are treated as "missing".

Assessment of 25-OH-D status
We chose to apply the Endocrine Society clinical practice guidelines cut-offs [28,29]. Therefore, VD status was defined as severe deficiency when 25(OH)D was lower than 10 ng/ml, moderate deficiency when 25(OH)D value was between 10 and 20 ng/ml, mild deficiency when 25(OH)D value was between 20 and 30 ng/ml, or sufficiency when 25(OH)D was above 30 ng/ml.
Biochemical rickets was defined according to the following criteria: VDD (all categories) associated with increased ALP levels, normal or decreased calcium levels and normal or decreased phosphate levels.

Statistical analyses
The population in study was divided into three climatic zones on the basis of their birth country: temperate zone (from circles to the latitude of 40°N/S), subtropical zone (from 40°N/S to the tropic of cancer/capricorn) and tropical zone (from Cancer's tropic to Capricorn's tropic).
Skin color was categorized into three groups: very fair/ fair, intermediate, olive/brown. The Von Luschan's chromatic scale was used to differentiate skin colors (0-13 on Von Luschan's scale was categorized as "very fair/ fair"; 14-20 as "intermediate"; 21-36 as "olive/brown"). Descriptive and inferential statistics were carried out using the Statistical Package of Social Sciences (Chicago, IL, USA) for Windows software program version 22.0. Frequencies and percentages (%) were used to describe categorical variables (gender, macro-area, climatic zones, skin color, season at blood sampling, housing solution before adoption, category of BMI, VD status, PTH status, ALP status, Ca status, P status, biochemical rickets), while mean and standard deviation (mean ± SD) were used to express continuous variables (age on arrival in Italy, age at blood draw, time between the arrival and the blood draw, duration of the institutionalization, height, weight, BMI, 25(OH)D, PTH, ALP, Ca, P). We determined the normal distribution of continuous variables with a Kolmogorov-Smirnof test: the result permitted us to use parametric tests.
Differences among the categorical variables were tested for statistical significance using the Chi-square test. A p value < 0.05 was considered significant, Confidence Interval (CI) was 95%. Differences among mean values of continuous variables were tested for statistical significance using the Analysis of Variance (ANOVA) test and T test. A p value < 0.05 was considered significant, with a CI = 95%.
Post hoc analysis was performed for χ 2 tests significant results, when a contingency table bigger than 2*2 was generated Bonferroni adjusted p-values were considered to determine which couple of categorical variables significantly differed.
A multivariate logistic regression was performed for exploring possible predictors of low 25(OH)D levels at a cut-off point of 30 ng/ml.
Variables to be entered in the multivariate model were chosen on the basis of the univariate analysis results (p < 0.25). To identify the best model a backward approach was developed, based on Likelihood Ratio test. The final model was assessed in terms of calibration and discrimination through Hosmer-Lemeshow and C-statistics.
ORs for proportions below these cut-offs and their confidence intervals are reported.

Results
Characteristics of the population, season at blood drawing and mean values of the analytes investigated (25(OH)D, PTH, ALP, Ca, P) are resumed in Table 1. Detailed informations about post-hoc analyses are included in Additional file 1.
Vitamin D mean value and status and characteristics of the population (Table 2) We did not find any significant difference regarding differences between mean values in females and males (p = 0.594) regarding mean 25(OH)D values. On the other hand, we found a slightly significant influence of sex on Vitamin D status (p = 0.043). Post hoc analysis determined that sex was significantly related to VD status only when testing a severe versus a moderate VDD, with female children having an increased risk of developing severe VDD than male children (OR 0.55, 95% CI 0.36-0.86).
We did not find any statistically significant difference between 25(OH)D mean values and BMI categories (p = 0.477). The difference is not statistically significant either when applied to Vitamin D status (p = 0.521).
A statistically significant difference was found for skin color (p < 0.001) regarding mean 25(OH)D values. Mean 25(OH)D value was 18.00 ng/mL (SD ± 10.78 ng/mL) for children included in the group "olive to brown"; 20.86 ng/mL (SD ± 10.12 ng/mL) for children included in the group "very fair to fair"; 23.24 ng/mL (SD ± 11.76 ng/mL) for children included in the group "intermediate". This difference remains statistically significant when considering Vitamin D Status (p < 0.001). However, post hoc analysis highlighted that statistically significant differences only exist with regards to intermediate skin color when compared to olive/brown skin color. As a matter of fact, the character "olive/ brown" skin resulted to be predictive of a worse VD status both when the severe VDD status is compared to mild VDD status (OR 0.27, 95% CI 0.151-0.489) and More detailed informations about the results above illustrated can be found in Table 2 and in Additional file 1.
Time from the arrival to the first evaluation was not found to be statistically related with 25(OH)D mean values (p = 0.388) and Vitamin D Status (p = 0.912). Age at the arrival in Italy was significantly associated both with 25(OH)D mean values (p < 0.001) and Vitamin D status (p < 0.001). Age at the blood draw association with 25(OH)D mean values (p = 0.049) was marginally significant, but it had a highly statistically significative relationship with Vitamin D status (p < 0.001). Table 3 shows mean time from the arrival to first evaluation, age at the arrival and age at first evaluation in relation to Vitamin D Status. Table 4 shows age at the arrival in Italy in years (mean ± SD) per Vitamin D Status, and its relations with sex, macro-area, season at blood draw, housing solution and BMI status. Figure 1 shows mean age at the arrival, in years, and its relationship with Vitamin D Status, while fig. 2 shows  Details about the results above described can be found in Table 5 and Additional file 1.
Climatic-zones of origin were also found to have a significant influence over 25(OH)D mean values (p < 0.001) and Vitamin D Status (p = 0.002). Mean 25(OH)D value in children coming from a tropical area was 22.27 ng/mL (SD ± 12.24 ng/mL), in children coming from a subtropical area was 17.83 ng/mL (SD ± 8.51 ng/mL), while in children coming from a temperate zone was 21.00 ng/mL (SD ± 10.28 ng/mL).  Post hoc analyses highlighted that statistically significant differences existed only for those children coming from subtropical areas when compared to children coming from tropical areas. As a matter of fact, children coming from a tropical area tend to have higher VD values than children coming from a subtropical area when comparing the severe VDD status with normal VD status (OR 5.329, 95% CI 2.512-11.41).
Detailed informations about the results above described can be found in Table 5 and Additional file 1.
Even housing solution had a significant relationship with 25(OH)D mean values (p = 0.016) and vitamin D status (p = 0.008). Mean 25(OH)D values in children who were guest of a foster house were 24.21 ng/ml (SD ± 16.13 ng/mL), in children housed in a foster family were 18.79 ng/mL (SD ± 6.72 ng/mL), while in children who were guest of an orphanage were 20.95 ng/mL (SD ± 10.92 ng/mL).
Post hoc analyses highlighted how actually the only significant difference can be found comparing children coming from foster homes with children coming from foster families. Children coming from foster families seem to be more at risk of a worst VD status, especially when comparing moderate VDD status with normal VD status (OR 0.137, 95% CI 0.049-0.453).
Detailed data about the results above described can be found in Table 5 and Additional file 1.

Vitamin D mean values and status and their relations with season of blood sampling
Season at blood sampling was significantly related both with 25(OH)D mean values (p < 0.001) and Vitamin D status (p < 0.001). Mean 25(OH)D value in children having their blood drawn in spring was 18.85 ng/mL (SD ± 10.39 ng/mL), in children undergoing blood draw in summer was 26.82 ng/mL (SD ± 11.08 ng/mL), in children who were firstly evaluated during fall was 22.28 ng/ mL (SD ± 10.10 ng/mL) and in children having their first examination during winter was 18.10 ng/mL (SD ± 11.32 ng/mL).
Post hoc analyses highlighted how during summer the risk of VDD is substanstially decreased, compared to all the other seasons. Interestingly, a statistically significant difference can be found between children observed during spring and children observed during fall, when severe VDD status is compared to mild VDD status. Children observed during fall tend to have a better VD status than children observed during spring (OR 3.319, 95% CI 1.656-6.715). Table 5 shows Vitamin D status in relation to macroarea, climatic zone of origin, housing solution and season at blood sampling. Detailed data about post hoc analyses can be found in Additional file 1. Fig. 1 Vitamin D status in relation to internationally adopted children mean age. Vitamin D status seems to be inversely related to internationally adopted children mean age. As a matter of fact, in our cohort a worst status can be observed in older children Post hoc analysis revealed that this statistically significant difference is only real when comparing severe VDD status with mild VDD status, with children having an increased ALP being at more risk of having a worse VD status (OR 0.4562, 95% CI 0.2737-0.7773).
Detailed data about post hoc analyses can be found in Additional file 1.
Ca status was not significantly related to 25(OH)D mean values (p = 0.338) or Vitamin D status (p = 0.296).
P mean values were significantly associated to 25(OH)D mean values (p = 0.004) and to Vitamin D status (p = 0.012). P status was equally significantly related to 25(OH)D mean values (p = 0.008) and to Vitamin D status (p = 0.014). 25(OH)D mean value in children having a normal P was 21.22 ng/mL (SD ± 11.14 ng/mL), in children having a decreased P was 23.06 ng/mL (SD ± 12.61 ng/mL). Table 6 shows Vitamin D status in relation to PTH status, ALP status, Ca status and P status.

Biochemical rickets
Only 811 children (94.4%) out of 859 had the complete blood tests panel for the definition of biochemical rickets. The analysis performed permitted to identify 129 children affected by biochemical rickets as defined in the "Subjects and Methods" section. 28 children (21.7%) had severe VDD and an increased ALP, associated with normal or reduced Ca and P. 61 children (47.3%) had moderate VDD and an increased ALP, associated with normal or reduced Ca and P. 40 children () had mild VDD and an increased ALP, associated with normal or reduced Ca and P. Table 7 resumes details about children affected by biochemical rickets.

Multivariate analysis
We entered the statistically significant variables (p < 0.25) in relation to Vitamin D status (sex, skin color, macroarea of origin, climatic zone of origin, season at first evaluation, housing solutions, ALP status, P status, Age at arrival in Italy, Age at first visit) in a multiple logistic regression model. At the end of the assessment, skin color (p = 0.011), season at first blood draw (p < 0.001), housing solution (p < 0.001), ALP status (p = 0.025) and the age at the first blood draw (p < 0.001) had a statistically significant association with Vitamin D Status. Table 8 shows details about the multiple logistic regression analysis final results. Figure 3 shows the risk of developing hypovitaminosis D during the different seasons when using spring as baseline.

Discussion
Migrant children are a population of great interest nowadays, and their health problems-infectious ones, but also not-infectious ones-are a source of interesting discussions for pediatricians all over the world. Internationally adopted children are a controlled sub-population of the migrant one, thus it could offer an insight on a larger populationthough with a lot of differences regarding travelling, housing solutions and dietary habits and deprivations. Vitamin D is an important pro-hormone, classically associated with bone health, but recently found to be involved in immune system and neuro-psychological disorders, and cardiovascular risk. It can be seen how a D hypovitaminosis could put at risk the entire well-being of a child [1][2][3][4][5].
Our study, carried on for nine years, tried to test some of the well-known predictors for the Vitamin D deficiency in the internationally adopted children population and the relationships existing between vitamin D and other factors implied in the bone health, such as parathyroid hormone, alkaline phosphatase, free calcium and phosphates.
From January 2007 to December 2015, 29,866 children arrived in Italy through an international adoption, from Eastern Europe countries above all, followed by Latin America, Asia and Africa [30]. The cited report offers only details about 2014 and 2015, but it specifies that the percentages reported are similar to those previously observed. During this period, adopted male children were more than the female ones (58.3% vs 41.7%; ratio = 1.4:1), with a mean age at the arrival of 5.9 years (SD not specified).
The population we observed had a similar composition to that of the entire "internationally adopted children" Italian population: male to female ratio was 1.3:1, with a prevalence of children coming from Eastern Europe countries and a mean age at the arrival of 5.31 years (SD ± 2.92 years).
We did not find a statistical difference of mean 25(OH)D values between sexes. However, a slight statistical significance was found with regards to the Vitamin D status, with a slight higher percentage of females affected by VDD than males. Female children are, as a matter of fact, more interested by severe VDD and mild  Decreased Ca c 0 0 ALP Alkaline phosphatase; Ca Serum calcium; P Serum phosphate; 25-OH-D 25-hydroxylate-Vitamin D a Normal values, depending on the date of the blood draw and the age/sex of the patient, are accurately described within the section "Subjects and Methods"; b Normal values, depending on the age of the patient, are accurately described within the section "Subjects and Methods" c Normal values can be found in Table 3 and within the section "Subjects and Methods" VDD than male children, while male children seem to be more affected by moderate VDD than females (Table 2 and Additional file 1). A reason for this evidence could be a higher BMI in female children than males, a justification often found by other studies, highlighting how a higher BMI is directly related to the risk of developing VDD, as VD is a lipophilic vitamin and a higher fat mass increase its distribution volume [31][32][33][34][35].
Our results, though, are opposite to those showed by other studies in which VDD was found to be more prevalent among males than in females, or it showed no significant relationship with sex at all [11,[36][37][38][39].
In our population, prevalence of VDD is significantly related to the age at the arrival in Italy and to the age at first evaluation. Figure 1 shows the mean age per group of Vitamin D status, highlighting that in our population a higher mean age is related to a worst Vitamin D status.
The negative association of age and serum 25(OH)D in children has already been described in literature. According with Mansbach et al., this might be because of  Fig. 3 Risk of finding a vitamin D (VDD) deficiency according to season of blood draw. When using spring as a baseline, the only significative association is a lower risk of finding a VDD when the blood draw is performed during summer. Autumn is associated with a non-significant slightly lower risk, while Winter with a non-significant slightly higher risk the difference in duration of sun exposure, while Gustafson et al. sustain that it is caused by the rapid growth spurt, or "catch-up growth" the children reach out in the first year after adoption [7,40]. Catch-up growth is a process similar to the rapid growth experienced by infants and toddlers, when use of VD supplement is customary for the prevention of rickets [41]. Therefore, during this period internationally adopted need larger amounts of dietary or supplemental iron and VD than those a child of the same age would normally need.
Interestingly, in our population high 25(OH)D concentrations were observed in children aged < 1 year regardless of the area of origin (Fig. 1). This is most likely attributable to the use of fortified formula milk during the period spent in an institution and by pediatricians' increasing awareness of the hypovitaminosis D issue once they arrive in Italy. As reported by other studies, this relation might lead to nullify the expected effects of an adoptee's birth country on VD level [35,40].
Our study showed a strong association between skin color and Vitamin D Status, that only remained significant for the value "olive-brown skin" after the multivariate analysis (logistic regression). Our data showed that children with olive to brown skin had a statistically significant 2-fold higher risk of having hypovitaminosis D than children with fair skin (OR 2.387, Table 8). These results agree with those reported by an Italian study by Franchi et al. evaluating 1.374 children living in the north of Italy. The study observed that ethnicity was a strong predictor of 25(OH)D levels, reporting a high prevalence of hypovitaminosis D in African, North African, and Indian children living in Italy (74.8, 81.2, 89.7, and 76.0%, respectively) [25]. Other studies reported the same results. A severe 25(OH)D deficiency associated to a proximal myopathy was observed in recent immigrants from Palestine, Pakistan and India to Northern Europe develop due to the limited effect of sunshine and a low dietary VD intake [37]. Hintzpeter [35] showed that Turkish and Arab-Islamic participants of both sexes as well as Asian and African girls were found to be at high risk for VDD, with the highest OR for African girls (7.8) [35]. Similarly, data from NHANES III showed that adolescents of African origin had an increased risk of VDD (OR 8.6) compared with white adolescents [15].
These consistent observations can lead us to conclude that a spontaneous recovery of VD status is unlikely in an adopted child arrived in Italy with VDD, independently from their ethnicity.
In our study macro area of origin was significantly associated with VD status (p < 0.001). Basing our observation on percentages alone, VDD seems to have been more common in Africa than in Latin America or Europe and Russian Federation. If statistically significant, our results would differ from those showed by Chiappini et al., whose study highlighted the highest prevalence of VDD in children coming from the Russian federation [26]. However, post hoc analyses highlighted that the real statistically significant differences are only between children coming from Asia and children coming from Africa. See Additional file 1 and the "results" section for a more detailed description of this difference. Thus, it is not possible to conclude that VDD is more common in children coming from Africa than in children coming from Latin America or Europe and Russian Federation.
As previously said with regards to skin color, it is thought that occasional exposure to sunlight provides most of the VD requirement of the human population [42]. However, skin synthesis of VD may not compensate for the low nutritional intake in Europe, because of the relationship occurring between its latitude and the rake angle of solar radiations. As a matter of fact, Webb et al. [43] demonstrated that the photosynthesis of pre-VD is nearly impossible during winter months at latitudes similar to those encasing Europe. A high prevalence of hypovitaminosis D combined with high PTH levels and lower bone mineral density has recently been reported in Finnish girls and Italian children and adolescents, suggesting that Italian children are at risk for this condition, as well as other European populations [44][45][46]. Moreover, another Italian study by Vierucci et al., showed that no VD is produced in central Italy during late fall, winter and the beginning of spring because of sun exposure [38,39].
The prevalence of VDD, even at southern latitudes, was remarkable, supporting other reports suggesting that living in sunny areas of the world does not necessarily protect children from VDD [39,43,[47][48][49][50].
Our work highlighted that the risk of developing hypovitaminosis D was significantly higher in children coming from foster families and orphanages, compared to those coming from foster families. This relationship was still statistically significant after multivariate analysis.
It is a well-demonstrated knownledge that children in foster care experience high rates of child abuse, emotional deprivation, and physical neglect than children living with their families [51][52][53]. Moreover, it is plausible to think that children in orphanages likely do not spend much time outdoors and that they lack adequate sun exposure. Another reason for the association with VDD could be that as children grow up in institutional care, they shift from a diet of vitamin D-fortified formula milk to cooked food, which may not be fortified with vitamin D.
According to other studies, we observed how the season of blood sampling could be considered a predictor of the Vitamin D Status [26,54]. As a matter of fact, it is only natural to find out that a pro-hormone which synthesis is related to sun-exposure has higher levels during summer than during winter. It has been shown that even minimal exposure of skin to sunlight, is adequate to reach VD sufficiency in US and Central Europe, and a number of studies demonstrated that both serum concentrations of 25(OH)D and other biomarkers of bone health status increase during summer only to reduce during winter, while PTH behavior is the exact opposite [55][56][57][58][59][60][61][62][63].
Our data highlighted how drawing a blood sample during summer was a protective factor against hypovitaminosis D, when compared to spring (Fig. 1). OR for summer shows a reduction of the risk of approximately the 80%.
In our study, almost half the children had normal Ca, P, ALP and PTH levels. This data differs from other pediatric studies, which show a percentage way lower [23,64]. Normal serum calcium works as an inhibitory signal for parathyroids, stopping PTH secretion, as described by Steingrimsdottir and Aggarwal [65,66].
Our study differs from other studies on the same matter because of the interest taken in showing the relationship existing between vitamin D and parathyroid hormone. Only a very small percentage of our population was affected by hyperparathyroidism, but a significant data was that we could observe elevated PTH levels in children having 25(OH)D levels over 30 ng/mL and normal calcium, a finding deserving further studies: was it a primary hypersecretion or a signal of an altered bone metabolism?
We also tried to identify the PTH inflection point for this population. With this term, other studies indicated the mean 25(OH)D value at which PTH begins to increase. We did not succeed, probably because of the small number of children with an increased PTH included in our population and the extreme variability of values at which PTH was found to increase.
Biochemical rickets affected a significant percentage of our study population. Surprisingly, even children with only mild VDD were affected by it (Table 8). It is though that rickets is confined to children exposed to extreme undernourishment but, as a matter of fact, it can be also observed in western countries, because of the growing epidemia of childhood obesity and the previously discussed relationship of high body mass index and fat mass with VDD [67].

Limits -Strenghts
Our study was limited by the lack of data on dietary intake of calcium and vitamin D. Type of nutrition in the first few months of life can deeply influence values of VD. Some of these infants and toddlers lived with their biological families prior to adoption and might have been breast-fed (low VD intake, limited Ca absorption); however, many international adoptees spent most of their life in orphanages or with foster families and were primarily formula fed (high VD intake, maximized Ca absorption).
Additionally, we could not retrieve any information about sunshine exposure, use of sun protections, medical conditions (i.e renal and/or hepatic failure), use of hypovitaminizing drugs, use of concealing clothing.
However, our study also has some strengths: it shows the irrelevance of PTH as a predictor of VDD, and identifies, through a multivariate analysis, some of the factors a pediatrician cannot miss when evaluating an internationally adopted child for the first time.

Conclusions
The study indicates a widespread global VD insufficiency compared with proposed threshold levels. This high prevalence implies that higher doses than currently recommended would be needed to reach country specific "desirable" levels.
Our results reinforce the importance of an early evaluation of serum 25(OH)D levels in internationally adopted children to start the adequate strategy of VD supplementation/treatment, avoiding further deterioration of VD status, particularly in older, dark skin children arrived in Italy during winter and spring period.