The Determinants of Bone Health in Children With Autism Spectrum Disorder: A Systematic Narrative Review

ABSTRACT

Autism spectrum disorder (ASD) is a complex disorder associated with social and communication impairments and repetitive and restrictive behavioural patterns. Children with ASD often present with concurrent conditions, including poor bone health, which affect long‐term health. Although there is compelling evidence to suggest that children with ASD have poorer bone traits than typically developing children, the primary factors associated with these differences are unclear. This review will explore the potential role that factors such as physical activity, nutrition (calcium, protein, vitamin C, vitamin D) and lifestyle (sleep, medication) play on bone health in children with ASD. Having a greater understanding of the influencing factors of low BMD and how these might interact in a synergistic manner in ASD children will provide an opportunity to develop targeted interventions to improve bone health aiming to avert attainment of suboptimal peak bone mass which may lead to early onset osteoporosis, fracture and muscle deconditioning in this paediatric population.

Summary.

• Children with autism spectrum disorder are disproportionately at risk of poor bone health which could result in early onset osteoporosis later in life.

• Although there is clear evidence of low bone mass in children with autism spectrum disorder when compared to typically developing children, the primary factors associated with these differences remain unclear.

• Evidence suggests that the low bone mass associated with children with autism spectrum disorder may be a result of multiple factors at play simultaneously. Thus, the development of interventions to help support low bone mass in these children should include a multifactorial approach.

1. Introduction

Musculoskeletal conditions are a significant health burden, affecting approximately 30% of the Australian population [1]. Suboptimal peak bone mass can lead to osteoporosis, fractures, muscle deconditioning, subsequent frailty and loss of independence [2]. Osteoporosis is often regarded as an older person's disease, but an estimated 26% of peak bone mineral content (BMC) is laid down in early childhood [3] and up to 90% of final bone mass is acquired by 18–20 years of age [4]. Much literature has reviewed measures to prevent or delay osteoporosis, with recommendations including optimising peak bone mass in childhood and adolescence.

While suboptimal bone health remains a concern for the general population, there are clinical populations disproportionately at risk of poor bone health and these are often understudied. One such group is children with autism spectrum disorder (ASD), a persistent developmental disorder with symptoms including difficulty with social interactions, impaired communication skills and restricted or repetitive behavioural patterns [5, 6]. The 2015 Australian Bureau of Statistics Survey of ‘Disability, Ageing and Carers in Australia’, reported that approximately one in 150 people have ASD [5]. Children with ASD are known to have several co‐morbidities less frequently encountered in typically developing children (TDC). In the past two decades, research has focused on finding an association between reduced bone health in children with ASD as compared to in TDC. Seminal work by Hediger et al. using hand‐wrist radiographs showed that boys with ASD had lower bone cortical thickness when compared to normative values, providing the basis for what is now the gold standard measurement, dual energy X‐ray absorptiometry (DEXA), for evaluating bone health in this population [7]. Since then, several cohorts of paediatric patients with ASD have been studied focusing on fracture, BMD, bone microarchitecture and bone strength. A number of reviews have consistently concluded that children with ASD have lower BMD than TDC [8, 9, 10]. We reviewed data from retrospective cross‐sectional analyses and prospective interventional studies and found consistent demonstration of lower BMD in children with ASD when compared to TDC with DEXA imaging at the lumbar spine and/or femoral neck, total hip, whole body less head and whole body. Four reviewed studies reported BMD z‐scores of DEXA lumbar spine scans that were significantly lower in children with ASD (z‐scores ranging from −1.1 to −0.56) when compared to TDC (z‐scores ranging from −0.21 to 0.32), (p < 0.05) [11, 12, 13, 14]. Despite this consensus, few studies have sought to determine the cause of low BMD in children with ASD. Neumeyer et al.'s 4‐year follow‐up of participants showed that boys (aged 11‐ to 14‐years‐old) with ASD accrued bone at a rate similar to TDC, however, it was found that BMD remained overall lower in children with ASD throughout puberty [12]. This may suggest determinants of low BMD seen in children with ASD lay their foundations before puberty. Furthermore, as DEXA measures areal bone density expressed as cm², BMD is size dependent. Consequently each individual child's BMD z‐score needs to be adjusted for their height, using an adjusted z‐score [15]. Only two reviewed studies reported BMC, which is a recommended measurement in the paediatric population due to the potential for bias from height in BMD from DEXA [14, 16, 17]. Of the studies reviewed, one utilised the advanced scanning method of high‐resolution peripheral quantitative computed tomography (HRpQCT), giving a three‐dimensional measure of volumetric BMD (vBMD), bone strength and bone microarchitecture [4, 18, 19, 20]. Significant differences in vBMD and bone microarchitecture at ultradistal radius and ultradistal tibia were reported between children with ASD and TDC [17]. This indicates that children with ASD have not only low bone mass but also compromised bone strength which may indicate increased risk of osteoporosis and fracture.

Multiple modifiable and non‐modifiable factors are known to play a role in the accrual of BMD in the paediatric population. When considering modifiable factors in children with ASD, reduced levels of physical activity (PA) [21], food selectivity [22, 23], restricted eating patterns [24], reduced nutritional intake [8], altered microbiome [25], sleep disturbances [26, 27] and hormonal abnormalities [28] have been found compared to TDC. These factors individually or in combination may influence the reduced BMD observed in children with ASD.

Although these factors are associated with bone health in the general paediatric population, there is no consensus on a connection between their presence and subsequent influence on BMD in children with ASD. This review aims to explore the potential role factors such as PA, nutrition (calcium, vitamin C, vitamin D), lifestyle (sleep, medication) and the microbiome play when determining why impaired bone health is observed in children with ASD. This review includes a comprehensive search of studies investigating children with ASD, bone health and the associated factors influencing bone properties between the years of 2000–2022 using Medline, Embase and Scopus databases (Table 1). Based upon predetermined inclusion criteria, a total of nine key studies were included and discussed within this review (Figure 1, Table 2).

TABLE 1.

Search strategies and Boolean operators utilised across the three key databases.

Database	Search strategy
Medline	#1 exp. Child/ #2 exp. Child, Preschool/ #3 exp. Adolescent/ #4 exp. Young Adult/ #5 (child* or girl or girls or boy or boys or youth* or adolescen* or teen* young adult*).mp. #6 1 or 2 or 3 or 4 or 5 #7 exp. Autistic Disorder/ #8 exp. Autism Spectrum Disorder/or exp. Child Development Disorders, Pervasive/ #9 (autistic or autism or pervasive development disorder*).mp. #10 7 or 8 or 9 #11 6 and 10 #12 exp. Bone Density/ #13 exp. Bone Development/ #14 exp. Bone Remodelling/ #15 exp. Osteoporosis/ #16 exp. Skeleton/ #17 ((bone or bones) adj3 (densit* or develop* or health* or quality or qualities or loss or losses or turnover* or structure* or geometr* or (peak* and mass*))).mp. #18 (fracture or osteoporosis or skeleton or skeletal development or skeletal mineralization or skeletal mineralization).mp. #19 12 or 13 or 14 or 15 or 16 or 17 or 18 #20 Nutritional Status/ #21 exp. Diet/ #22 Energy Intake/ #23 exp. Proteins/ #24 exp. Dietary Proteins/ #25 Milk/ #26 exp. Milk Proteins/ #27 exp. Dairy Products/ #28 exp. Eggs/ #29 exp. Nutrients/ #30 exp. Micronutrients/ #31 exp. Vitamins/ #32 exp. Calcium/ #33 Calcium, Dietary/ #34 exp. Fatty Acids, Omega‐3/ #35 exp. Dietary Carbohydrates/or exp. Carbohydrates #36 exp. Fats/ #37 exp. Dietary Fibre/ #38 exp. Carbonated Beverages/ #39 (nutritional status or diet or energy intake or protein or protein intake or milk or dairy or egg or eggs or macronutrient* or micronutrient* or vitamin* or “25(OH)D” or calcium or Omega‐3 fatty acids or carbohydrate* or fat or fats or fibre or fibre or carbonate beverage* or cola).ti,ab. #40 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38 or 39 #41 exp. Exercise/ #42 exp. Motor Activity/ #43 exp. Locomotion/ #44 exp. Sports/ #45 exp. Recreation/ #46 (exercis* or physical activit* or motor activit* or physical function or motor function or locomotor activit* or sport* or recreation).ti,ab. #47 41 or 42 or 43 or 44 or 45 or 46 #48 exp. Sleep/ #49 Gastrointestinal Microbiome/ #50 exp. Growth Hormone/ #51 exp. Pharmaceutical Preparations/ #52 exp. Potentially Inappropriate Medication List/or exp. “Medication Review”/ or exp. Preanesthetic Medication/or exp. Medication Reconciliation/or exp. Self Medication/or exp. Medication Systems/or exp. Medication Errors/or exp. Medication Adherence/or exp. Medication Therapy Management/ #53 (sleep or gastrointestinal microbiome* or gut microbiome* or growth hormone* or medic* or drug*).ti,ab. #54 48 or 49 or 50 or 51 or 52 or 53 #55 40 or 47 or 54 #56 11 and 19 and 55 #57 limit 56 to yr. = “2000—Current”
Embase	#1 exp. child/ #2 exp. preschool child/ #3 exp. adolescent/ #4 exp. young adult/ #5 (child* or girl or girls or boy or boys or youth* or adolescen* or teen* young adult*).mp. #6 1 or 2 or 3 or 4 or 5 #7 exp. autism/ #8 (autistic or autism or pervasive development disorder*).mp. #9 7 or 8 #10 6 or 9 #11 exp. bone density/ #12 exp. bone development/ #13 exp. bone remodelling/ #14 exp. osteoporosis/ #15 exp. skeleton/ #16 ((bone or bones) adj3 (densit* or develop* or health* or quality or qualities or loss or losses or turnover* or structure* or geometr* or (peak* and mass*))).mp./ #17 (fracture or osteoporosis or skeleton or skeletal development or skeletal mineralization or skeletal mineralization).mp. #18 11 or 12 or 13 or 14 or 15 or 16 or 17 #19 exp. nutritional status/ #20 exp. diet/ #21 caloric intake/ #22 protein/ #23 protein intake/ #24 exp. milk/ #25 milk protein/ #26 exp. dairy product/ #27 exp. egg/ #28 nutrient/ #29 exp. trace element/ #30 exp. vitamin/ #31 calcium/ #32 calcium intake/ #33 omega 3 fatty acid/ #34 exp. carbohydrate intake/ #35 exp. carbohydrate/ #36 fat #37 exp. dietary fibre/ #38 exp. carbonated beverage/ #39 (nutritional status or diet or energy intake or protein or protein intake or milk or dairy or egg or eggs or macronutrient* or micronutrient* or vitamin* or “25(OH)D” or calcium or Omega‐3 fatty acids or carbohydrate* or fat or fats or fibre or fibre or carbonate beverage* or cola).ti,ab. #40 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38 or 39 #41 exp. exercise/ #42 exp. motor activity/ #43 exp. locomotion/ #44 exp. sport/ #45 exp. recreation/ #46 (exercis* or physical activit* or motor activit* or physical function or motor function or locomotor activit* or sport* or recreation).ti,ab. #47 41 or 42 or 43 or 44 or 45 or 46 #48 exp. sleep/ #49 exp. intestine flora/ #50 exp. growth hormone/ #51 exp. drug/ #52 exp. drug therapy/ #53 (sleep or gastrointestinal microbiome* or gut microbiome* or growth hormone* or drug* or medic*).ti,ab. #54 48 or 49 or 50 or 51 or 52 or 53 #55 40 or 47 or 54 #56 10 and 18 and 55 #57 limit 56 to yr. = “2000—Current”
Scopus	((TITLE‐ABS‐KEY (autis* OR asd OR “pervasive development disorder” *)) AND (TITLE‐ABS‐KEY (child* OR girl* OR boy* OR paed* OR youth* OR adolescen* OR teen* OR “young adult” OR “young adults”) AND (TITLE‐ABS‐KEY (bone* OR bmd OR osteo* OR skelet*) AND ((TITLE‐ABS‐KEY (nutrition* OR diet* OR “energy intake” OR protein* OR milk* OR dairy AND * OR egg* OR nutrient* OR micronutrient* OR vitamin* OR calcium OR supplement* OR “fatty acids” OR omega‐3 OR carb* OR “carbonated beverages” OR fat* OR fib* OR “25(OH)D”) OR (TITLE‐ABS‐KEY (exercis* OR “physical activity” OR “motor activity” OR “physical function” OR “motor function” OR “locomotor activity” OR locomotion OR sport* OR recreation*) OR (TITLE‐ABS‐KEY (sleep OR “gastrointestinal microbiome” OR “gut microbiome” OR “growth hormone” OR medic* OR drug*) AND (LIMIT‐TO (PUBYEAR, 2022) OR LIMIT‐TO (PUBYEAR, 2021) OR LIMIT‐TO (PUBYEAR, 2020) OR LIMIT‐TO (PUBYEAR, 2019) OR LIMIT‐TO (PUBYEAR, 2018) OR LIMIT‐TO (PUBYEAR, 2017) OR LIMIT‐TO (PUBYEAR, 2016) OR LIMIT‐TO (PUBYEAR, 2015) OR LIMIT‐TO (PUBYEAR, 2014) OR LIMIT‐TO (PUBYEAR, 2013) OR LIMIT‐TO (PUBYEAR, 2012) OR LIMIT‐TO (PUBYEAR, 2011) OR LIMIT‐TO (PUBYEAR, 2010) OR LIMIT‐TO (PUBYEAR, 2009) OR LIMIT‐TO (PUBYEAR, 2008) OR LIMIT‐TO (PUBYEAR, 2007) OR LIMIT‐TO (PUBYEAR, 2006) OR LIMIT‐TO (PUBYEAR, 2005) OR LIMIT‐TO (PUBYEAR, 2004) OR LIMIT‐TO (PUBYEAR, 2003) OR LIMIT‐TO (PUBYEAR, 2002) OR LIMIT‐TO (PUBYEAR, 2001) OR LIMIT‐TO (PUBYEAR, 2000)

FIGURE 1.

Flow chart of studies included within this systematic review.

TABLE 2.

Description and analyses of the key information of the identified studies in alphabetical order of author.

Authors	Title	Year	Type of study	Population	Location BMD measurement	BMD outcomes	Investigated factors possibly influencing
							Nutrition	Calcium	Vitamin‐D	Physical activity	Medications	Hormones
Barnhill et al.	Bone Mineral Density in Boys Diagnosed with Autism Spectrum Disorder: A Case–Control Study	2017	Retrospective study	ASD n = 40 TDC n = 40	DEXA lumbar spine	ASD BMD z‐score = −0.56 (−0.88, −0.23) TDC BMD z‐score = 0.32 (−0.11, 0.74) p = 0.002	✓	✓	✓
Clapham et al.	Effectiveness of surf therapy for children with disabilities	2020	Prospective study, 8 weeks	ASD n = unspecified Total children = 71	DEXA unspecified location	Did not report raw values of BMD p = 0.004				✓
Goodarzi and Hemayattalab	Bone mineral density accrual in students with autism spectrum disorders: Effects of calcium intake and physical training	2012	Randomised control trial, 6 months	ASD n = 60	DEXA femoral neck	Pre‐interval BMD (g/cm2) mean: Ex+Ca+ = 0.625, SD = 0.020 Ex+Ca− = 0.625, SD = 0.018 Ex−Ca+ = 0.624, SD = 0.017 Ex−Ca− = 0.625, SD = 0.022		✓		✓
						Post‐interval BMD mean: Ex+Ca+ = 0.643, SD = 0.017 Ex+Ca− = 0.633, SD = 0.017 Ex−Ca+ = 0.628, SD = 0.016 Ex−Ca− = 0.624, SD = 0.021
						Mean difference: Ex+Ca+ = −0.0190* Ex+Ca− = −0.0080* Ex−Ca + = −0.0040* Ex−Ca− = −0.0003
Neumeyer et al.	Bone density in peripubertal boys with autism spectrum disorders	2013	Retrospective study	ASD n = 18 TDC n = 19	DEXA lumbar spine	ASD BMD z‐score = −1.13 (−1.41, −0.85) TDC BMD z‐score = −0.21 (−0.46, 0.04) p = 0.020	✓	✓	✓	✓		✓
					DEXA femoral neck	ASD BMD z‐score = −1.64 (−1.85, −1.43) TDC BMD z‐score = −0.52 (−0.76, −0.28) p: 0.001
					DEXA hip	ASD BMD z‐score = −0.92 (−1.16, −0.68) TDC BMD z‐score = 0.14 (−0.15, 0.43) p = 0.002
Neumeyer et al.	Bone microarchitecture in adolescent boys with autism spectrum disorder	2017	Retrospective study	ASD n = 16 TDC n = 18	DEXA lumbar spine	ASD BMD z‐score = −0.62 (−0.89, −0.35) TDC BMD z‐score = −0.11 (−0.30, 0.08) p = 0.120	✓	✓	✓	✓		✓
					DEXA femoral neck	ASD BMD z‐score = −1.19 (−1.41, −0.97) TDC BMD z‐score = −0.41 (−0.62, −0.20) p = 0.010
					DEXA total hip	ASD BMD z‐score = −0.73 (−0.96, −0.5) TDC BMD z‐score = −0.23 (−0.43, −0.03) p = 0.094
					DEXA whole body	ASD BMD z‐score = −1.25 (−1.43, −1.03) TDC BMD z‐score = −0.49 (−0.67, −0.31) p = 0.005
					DEXA whole body less head	ASD BMD z‐score = −1.12 (−1.34, −0.9) TDC BMD z‐score = −0.48 (−0.67, −0.29) p = 0.029
					HRpQCT ultradistal radius	ASD log‐transformed total area (mm2) mean = 5.80 Control log‐transformed total area (mm2) mean = 5.80 p = 0.710
						ASD log‐transformed cortical area (mm2) mean = 1.88 Control log‐transformed cortical area (mm2) mean = 2.80 p = 0.076
						ASD log‐transformed trabecular area (mm2) mean = 5.70 Control log‐transformed trabecular area (mm2) mean = 5.70 p = 0.750
						ASD log‐transformed cortical thickness (mm) mean = −2.50 Control log‐transformed Cortical thickness (mm) mean = −1.50 p = 0.076
						ASD total volumetric BMD (mgHA/cm3) mean = 222.80 Control total volumetric BMD (mgHA/cm3) mean = 239.30 p = 0.50
						ASD cortical volumetric BMD (mgHA/cm3) mean = 639.10 Control cortical volumetric BMD (mgHA/cm3) mean = 680.90 p = 0.043
						ASD trabecular volumetric BMD (mgHA/cm3) mean = 175.50 Control trabecular volumetric BMD (mgHA/cm3) mean = 186.20 p = 0.720
					HRpQCT ultradistal tibia	ASD log‐transformed total area (mm2) mean = 6.60 Control log‐transformed total area (mm2) mean = 6.70 p = 0.390
						ASD log‐transformed cortical area (mm2) mean = 4.10 Control log‐transformed cortical area (mm2) mean = 4.40 p = 0.031
						ASD log‐transformed trabecular area (mm2) mean = 6.50 Control log‐transformed trabecular area (mm2) mean = 6.60 p = 0.500
						ASD log‐transformed cortical thickness (mm) mean = −0.57 Control log‐transformed cortical thickness (mm) mean = −0.35 p = 0.085
						ASD total volumetric BMD (mgHA/cm3) mean = 247.90 Control total volumetric BMD (mgHA/cm3) mean = 252.20 p = 0.72
						ASD cortical volumetric BMD (mgHA/cm3) mean = 737.70 Control cortical volumetric BMD (mgHA/cm3) mean = 753.20 p = 0.280
						ASD trabecular volumetric BMD (mgHA/cm3) mean = 186.80 Control trabecular volumetric BMD (mgHA/cm3) mean = 189.40 p = 0.940
Neumeyer et al.	Bone Accrual in Males with Autism Spectrum Disorder	2017	Prospective study	ASD n = 25 TDC n = 24	DEXA lumbar spine	Unadjusted baseline BMD z‐score ASD = −0.89 (−1.11, −0.67) Unadjusted baseline BMD z‐score TDC = −0.03 (−0.27, 0.21) p = 0.035	✓	✓	✓	✓		✓
					DEXA femoral neck	Unadjusted baseline BMD z‐score ASD = −1.54 (−1.73, −1.35) Unadjusted baseline BMD z‐score TDC = −0.59 (−0.79, −0.39) p = 0.012
					DEXA total hip	Unadjusted baseline BMD z‐score ASD = −0.66 (−0.88, −0.44) Unadjusted baseline BMD z‐score TDC = 0.09 (−0.16, 0.34) p = 0.060
					DEXA whole body	Unadjusted baseline BMD z‐score ASD = −1.36 (−1.55, −1.17) Unadjusted baseline BMD z‐score TDC = −0.44 (−0.65, −0.23) p = 0.130
					DEXA whole body less head	Unadjusted baseline BMD z‐score ASD = −1.24 (−1.46, −1.02) Unadjusted baseline BMD z‐score TDC = −0.45 (−0.69, −0.21) p = 0.410
Neumeyer et al.	Nutrition and Bone Density in Boys with Autism Spectrum Disorder	2018	Retrospective study	ASD n = 25 TDC n = 24	DEXA lumbar spine	BMD z‐score ASD = −0.94 (−1.16, −0.72) BMD z‐score TDC = 0.02 (−0.16, 0.20) p < 0.001	✓	✓	✓	✓
					DEXA femoral neck	BMD z‐score ASD = −1.56 (−1.75, −1.37) BMD z‐score TDC = −0.35 (−0.53, −0.17) p < 0.001
					DEXA total hip	BMD z‐score ASD = −0.92 (−1.17, −0.67) BMD z‐score TDC = 0.06 (−0.16, 0.28) p = 0.003
					DEXA whole body	BMD z‐score ASD = −1.29 (−1.49, −1.09) BMD z‐score TDC = −0.49 (−0.67, −0.31) p = 0.003
					DEXA whole body less head	BMD z‐score ASD = −1.19 (−1.41, −0.97) BMD z‐score TDC = −0.48 (−0.67, −0.29) p = 0.014
Roke et al.	Bone mineral density in male adolescents with autism spectrum disorders and disruptive behaviour disorder with or without antipsychotic treatment	2012	Retrospective study	ASD n = 102 APs = 56 No APs = 46	DEXA lumbar spine	BMD z‐score boys with ASD with AP‐induced hyperprolactinaemia = −0.18 (−2.36, 1.69) BMD z‐score boys with ASD without AP‐induced hyperprolactinaemia = 0.15 (−1.46, 1.51) p = 0.150					✓
						BMD z‐score boys with ASD not treated with AP = −0.026 (−0.46, 0.35) p = 0.740
					DEXA whole body	BMD z‐score boys with ASD with AP‐induced hyperprolactinaemia = 0.038 (−2.1, 2.84) BMD z‐score boys with ASD without AP‐induced hyperprolactinaemia = 0.083 (−1.24, 2.24) p = 0.790
						BMD z‐score boys with ASD not treated with AP = −0.20 (−0.67, 0.14) p = 0.800
Soden et al.	Nutrition, physical activity, and bone mineral density in youth with autistic spectrum disorders	2012	Retrospective study	ASD n = 26	DEXA	Mean BMD z‐score = −0.10 BMD z‐score range = −3.30 to 2.70	✓			✓	✓

Note: DEXA values for bone outcomes represent mean BMD z‐score (confidence intervals), HRpQCT values for bone outcomes represent mean log‐transformed trabecular area, log‐transformed trabecular area thickness and volumetric BMD (vBMD).

Abbreviations: AP = antipsychotic medication; ASD, autism spectrum disorder; BMD, bone mineral density; DEXA, dual energy X‐ray absorptiometry; Ex+Ca−, exercise intervention only; Ex+Ca+, exercise intervention and calcium intervention; Ex−Ca−, neither exercise nor calcium interventions; Ex−Ca+, calcium intervention online; HRpQCT, high‐resolution peripheral quantitative computed tomography; TDC, typically developing children.

^*Significance as per authors.

2. Modifiable and Non‐Modifiable Factors Associated With Low BMD in Children With ASD

Despite a broad search strategy, only nine studies were identified for analysis within this review. No studies pertaining to sleep or the microbiome were found within the literature. Across the studies, limitations were small sample sizes and a limited number of eligible DEXA scan results due to motion artefacts [11, 12, 13, 14, 17, 29, 30, 31, 32]. Four studies were published by the same research group, two of which used the same cohort of participants: this limits generalisability of findings and increases the risk of bias [11, 12, 13, 17]. Four studies did not include a TDC control group: this also limits generalisability of findings [29, 30, 31, 32]. Many studies reviewed only boys with ASD. Given the growth and developmental differences between genders, findings from these studies cannot be assumed to accurately apply to girls [11, 12, 13, 14, 17, 30, 31, 32].

2.1. Nutrition and Bone Health in ASD Children

Children with ASD have several dietary co‐morbidities such as selective eating patterns and restricted diets that influence the nutritional value of their oral intake [22, 33]. Seven studies reviewed the modifiable factor of ‘nutrition’, and reported inconsistent findings with respect to the relevance in causing low BMD in children with ASD [11, 12, 13, 14, 17, 30, 32]. Neumeyer et al. reviewed caloric intake by food‐type and concluded that boys with ASD consumed 16% fewer calories, 37% less animal protein and 20% less fat than TDC [13]. However, when controlling for PA, these differences were not statistically significant. Despite accounting for PA, estimated caloric deficits still ranged from 21% to 30% in the ASD group as compared to TDC [13]. This may suggest that despite an insufficient caloric intake, this modifiable factor is not the sole driver of low BMD in children with ASD. Soden et al. studied serum 25(OH)D and exposure to sunlight in children with ASD and found no significant correlation with BMD [32]. Investigations of oral intake of vitamin D and calcium provided conflicting results. There was a consensus that low intake of vitamin D and calcium through diet alone was negatively correlated with BMD in children with ASD [11, 12, 13, 14, 17]. Three studies by Neumeyer and colleagues, found that serum levels of vitamin D and calcium were the same between children with ASD and TDC [12, 13, 17], however one study by the same research group found that children with ASD had lower serum levels of vitamin D (p = 0.03) but similar levels of serum calcium compared to TDC [11]. Serum calcium levels were likely maintained by increased parathyroid hormone (PTH) secretion. Barnhill et al. found that children with ASD had statistically higher levels of serum vitamin D and magnesium as compared to TDC [14]. This study also measured supplemental calcium and vitamin D intake, observing 29/40 and 27/40 ASD subjects took calcium and vitamin D supplements, respectively, compared to only 2/40 TDC [14]. Despite the nutritional differences identified within the studies (including supplements), BMD z‐scores in children with ASD were not correlated with any biochemical markers, except for boron, which was negatively associated with BMD, or with nutrient intake, and remained significantly lower compared to TDC [11, 12, 13, 14, 17]. Intact PTH was measured only in two of these studies [14, 32], and found only to be increased in one participant, consistent with Stage II vitamin D deficiency (25(OH)D of 20 ng/mL) [32]. Taken together, the literature suggest that dietary intake or restrictive dieting alone do not cause clinically significant nutritional deficiencies or low BMD in children with ASD.

2.2. Physical Activity (PA) and Bone Health in ASD Children

PA has been determined as an important modifiable factor for bone health. In this review, several studies by Neumeyer and colleagues found that light levels of PA were negatively correlated with BMD in children with ASD: 72.2%–79.2% of TDC participants, but only 11.1%–27.3% of ASD participants reported vigorous levels of PA, defined by METs [11, 12, 13, 17]. However, Soden et al. found that neither time spent undertaking PA nor use of electronic media were correlated with BMD, despite four participants having BMD z‐scores meeting the criteria for ‘at‐risk’ BMD levels [32]. It is well‐established that bone loading from weight‐bearing exercise induces biochemical signals from osteocytes that stimulate osteoblastic bone formation and bone remodelling [34, 35, 36, 37]. Consequently, an absence of investigation into weight‐bearing exercise calls into question the relevance of these BMD findings. Not all exercise is weight‐bearing—swimming and cycling are non‐weight‐bearing for example, gymnastics and running are weight‐bearing. It is possible that the literature observed no correlations between PA and BMD because the exercise reported although vigorous was predominantly non‐osteogenic. The five studies reviewing PA used self‐reporting questionnaires, which may result in imprecision and bias [11, 12, 13, 17, 32, 38]. These tools are limited in their ability to capture specific bone‐related PA [12, 13, 17, 39]. Clapham et al. reported a significant increase in the BMD of children with a range of disabilities (including ASD) after an 8‐week surfing programme (which saw the participants paddling, balancing, kneeling and standing on a surfboard) indicating that structured exercise may improve BMD in the studied children [29]. However, Clapham et al. did not document specific bone‐related activities associated with BMD, included multiple disabilities and did not detail BMD changes specifically in participants with ASD.

2.3. Other Associated Factors

Many medications are reported to reduce BMD. Antipsychotics and antidepressants have specifically been researched regarding their use in children because of concern for bone health [40, 41, 42, 43]. Roke et al. found that boys with ASD and antipsychotic‐induced hyperprolactinaemia had significantly lower vBMD at the lumbar spine and higher total percentage body fat compared to boys with ASD treated with antipsychotics without hyperprolactinaemia [31]. While this study demonstrates the potential risk for children with ASD on antipsychotic medication, other literature has established that individual susceptibility to hyperprolactinaemia varies, and that states of hyperprolactinaemia can spontaneously resolve, perhaps negating long‐term risk to bone health [43, 44]. This study included children using concomitant medication that may induce hyperprolactinaemia, reducing generalisability of the results. Soden et al. also allowed participants to continue concomitant medications, and found no association between BMD and class or number of medications [32]. Much of the literature regarding BMD in children with ASD excludes those who are taking medications known to influence BMD.

There is limited evidence of the hormonal influence on BMD in children with ASD. The steepest gain in bone mass occurs approximately 6 months after the initial pubertal growth spurt [3]. The levels of insulin‐like growth factor (IGF‐1) and gonadal hormones surge during puberty, playing an important role in the marked increase of bone accrual and subsequent skeletal development during this developmental stage of life [45, 46]. Reviewed studies measured serum levels of IGF‐1, oestrogen and testosterone [11, 12, 17]. While two of these studies by Neumeyer and colleagues did not find a difference in these hormone levels between children with ASD and TDC [11, 12], a third study by the same authors found children with ASD counterintuitively had significantly higher levels of IGF‐1 than TDC, with levels of testosterone and oestradiol showing no difference between groups [17]. Within the TDC group, BMD was positively associated with IGF‐1 as expected in pubertal children, but was not correlated with IGF‐1 for boys with ASD [17]. As IGF‐1 plays a role in stimulating bone apposition, these findings may suggest a lack of responsiveness to IGF‐1 or attenuation of IGF‐1 actions caused by other factors [47].

2.4. Synergistic Effect of Factors Influencing Bone Health

The synergistic effect of concurrent modifiable factors may influence bone health in children with ASD. Of the literature reviewed, there is evidence of the interplay of the simultaneous effects of PA, particularly weight‐bearing PA, and adequate nutrition such as high caloric intake, calcium and vitamin D on bone health in children with ASD [13, 30]. Goodarzi and Hemayattalab's randomised control trial (RCT) investigated the individual and cumulative influence of weight‐bearing PA and supplemental calcium on femoral‐neck BMD in boys with ASD. Findings showed that combined therapy (exercise and supplemental calcium) was 23% more effective than no intervention, 19% more effective than consuming calcium alone and 14% more effective than exercise alone in increasing BMD [30]. While this RCT complements studies undertaken within the paediatric population, it is limited by the lack of blinding of researchers and participants to the interventions, inclusion of vitamin D in the calcium supplementation (of an unspecified amount), lack of baseline calcium, PTH, and vitamin D testing and lack of measurement of serum levels of calcium, PTH and 25(OH)D after the 6‐month interventions. While the optimal length of time for PA to exert peak gains on BMD has not been established, it is known that this effect takes a minimum of 6–9 months to occur [48, 49]. Thus, findings of this study could have been limited by the short 6‐month duration [29, 30].

Similarly, the combined effect of appropriate pubertal hormones—IGF‐1, GH, testosterone and oestradiol—which stimulate bone growth in children is dependent on adequate nutrition as well as osteoblastic stimulation from PA [11, 12, 17]. The added influence of medications is briefly addressed in the literature, particularly noting that antipsychotic medication may have a negative impact on BMD as a result of hyperprolactinaemia‐induced hypogonadotropic hypogonadism [31] although this has not been synergistically investigated with other factors. The additive effects of the reviewed factors, in addition to other potentially influential factors on bone health such as sleep and the microbiome, are yet to be studied.

3. Conclusion

While it is clear that BMD is lower in children with ASD compared to TDC, there is limited data and inconclusive evidence to suggest which sole or cumulative factors influence this finding. This review found that current studies analysing the nutrition, PA, medications and hormones give no clear consensus about the extent to which these factors may have an impact on bone health in children with ASD. The evidence does suggest, however, that the low BMD associated with ASD children may be a result of multiple factors at play simultaneously. Given the heterogeneity of study designs currently in the literature, there is a need for further investigation of factors potentially influencing the reduced BMD observed in children with ASD, with prospective studies designed with appropriate control groups, the inclusion of females and practical primary outcomes.

Author Contributions

All authors conceived and conceptualised the review. R.T.L. conducted the literature search and R.L.D. and R.T.L. reviewed the manuscripts. R.T.L. drafted the manuscript. All authors have read and approved the final version and agree to be accountable for all aspects of the work.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data will be available upon request.