The Effect of Vitamin D Supplementation for Bone Healing in Fracture Patients: A Systematic Review

While most literature on vitamin D supplementation in fracture patients focuses on fracture prevention, the effect of vitamin D on bone healing is a much less studied concept. The primary aim of this systematic review was to assess whether vitamin D supplementation in fracture patients improves clinical or radiological union complications. The secondary aims were to assess supplementation effect on patient functional outcome scores and bone mineral density (BMD). A systematic search of all relevant articles was performed using the following databases: MEDLINE, Embase, Google Scholar, and Web of Science. The population selection included human patients with a fresh fracture treated conservatively or operatively. The intervention included any form of vitamin D supplementation, compared to no supplementation or a placebo. The primary outcomes assessed were clinical or radiological union rates or complications arising from the nonunion. The secondary outcomes assessed were functional outcome scores, BMD scores after treatment, and pain scores. A total of fourteen studies, assessing a total of 2734 patients, were included. Eight studies assessed the effect of vitamin D on clinical or radiological union. Five studies reported no significant difference in complication rates when supplementing fracture patients. Alternatively, three studies reported a positive effect with supplementation between the groups. One of these studies found a difference only for early orthopaedic complications (<30 days), but no differences in late complications. The other two studies found significant differences in clinical union; however, no changes were observed in radiological union. Six studies investigated functional outcome scores after supplementation. Four of these studies found no significant differences between most functional outcome scores. Only three studies reported BMD outcomes, one of which found limited effect on total hip BMD. The overall findings are that vitamin D alone does little to influence fracture healing and subsequent union rates or functional outcome. The studies suggestive of a positive effect were generally of a lower quality. Further high quality RCTs are needed to justify routine supplementation at the time of fracture.


Introduction
Te role of vitamin D in calcium homeostasis and subsequent bone mineralisation is well established [1]. Despite this, the prevalence of vitamin D defciency has reached epidemic proportions [2]. Over the past decade, a combination of increased awareness and testing has brought vitamin D defciency to the forefront, leading to a signifcant increase in the amount of literature available [3]. Much of the literature in relation to fractures focuses on the efect of vitamin D in preventing risk. Two landmark meta-analyses by Bischof-Ferrari et al. have showed that vitamin D supplementation appears to reduce fracture risk in the elderly [4,5]. Tese recommendations have formed the basis of multiple guidelines on routine vitamin D supplementation despite being contested by some authors [6].
However, the efect of vitamin D on bone healing following a fracture is a much less studied concept. Studies performed in animal models have shown promising efects that adequate supplementation could enhance bone healing [7,8]. From a biochemical aspect, vitamin D appears to be involved in every phase of the fracture healing process by mobilizing calcium. However, there is conficting data showing varying levels of metabolites during the healing stage and the mechanism is poorly understood. Te systematic review by Gorter et al. in 2014 was one of the frst to broadly examine the efect of vitamin D on bone healing [9]. At the time, despite a total of 105 studies, no clinical studies focusing solely on vitamin supplementation were found. Only three human studies assessed the efect of vitamin D on bone healing, two of which had calcium cosupplementation and one showed an increase in the callus area at the fracture site [10][11][12]. Sprague et al. conducted a similar review in 2017, where they found that there is a high prevalence of hypovitaminosis D in a fracture cohort, however only identifed a conference abstract which found a positive efect on bone healing [13]. Since then, this abstract has been published in its entirety by Haines et al. who concluded no diference with vitamin D supplementation [14]. Calcium and vitamin D supplements are highly cost-efective and could save almost up to €6 billion annually in the EU on fracture prevention [15]. Fracture complications are associated with high patient morbidity and increased hospital expenses [16]. If vitamin D was to show a positive efect on reducing fracture complications, this would be a highly costefective option to decrease both morbidity and healthcare resource expenditures [17].
With the increase in published research since the work by Gorter and Sprague, a repeated systematic review to assess more current literature would shed more light on this topic. In this review, importance was placed solely on the efect of vitamin D, without the confounding efect of combined calcium supplementation. Te primary aim of this systematic review was to assess whether vitamin D supplementation in fracture patients improves clinical or radiological union complications. Te secondary aims were to assess supplementation efect on patient functional outcome scores and bone mineral density (BMD).

Methods
Te systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [18]. Te review was registered on the PROSPERO database (CRD: 42022306990). A systematic search of all relevant articles was performed by using the following databases: MEDLINE, Embase, Google Scholar, and Web of Science from incepeption till 28th February 2022. Te search strategy, included keywords, search terms, and MeSH headings, found in Appendix. Boolean operators AND/OR were used to expand and refne the searches where the databases allowed. Conference abstracts published by the Orthopaedic Research Society, the Orthopaedic Trauma Association, the European Federation of National Associations of Orthopaedics and Traumatology (EFORT), and the Orthopaedic Proceedings Supplement issued by Te Bone & Joint Journal were also searched to include gray literature. Te latter contains abstracts of articles presented at scientifc meetings or congresses organised by various orthopaedic associations. A manual citation search of all included studies was performed.

Study Eligibility
Criteria. Te PICOS model was used to formulate the eligibility criteria for the included studies [19]. Te population selection included human patients with a fresh fracture treated conservatively or operatively. Te intervention included oral, liquid, or intramuscular form of vitamin D supplementation, compared to no supplementation or a placebo, regardless of dosing regimen. Studies where vitamin D was cosupplemented were only included if the calcium regimen was standard across all intervention groups including the control, or if the efect of vitamin D alone could be compared to control groups with no treatment. Observational studies with no intervention, which correlated vitamin D levels on admission and fracture outcomes, were excluded. Te primary outcomes assessed were clinical and radiological union rates, or complications arising from the subsequent nonunion. Secondary outcomes assessed were functional outcome scores, BMD scores after treatment, and pain scores. All study types were included except for case series or case reports. Tere were no date or language restrictions. Studies involving animals, maxillofacial fractures, and pathological fractures or those involving fracture healing following elective primary surgery (e.g., osteotomy or ankle fusion) were excluded.

Study Selection and Data Collection.
Te study selection was frst screened by title and then by abstract. Te screening was performed by two authors (T.G and A.G), with a third author (H.A) to settle any disagreements following a discussion; however, this was not required. Cohen's kappa coefcient was 0.82 for title screening and 0.86 for abstract screening. Te selection process is summarized in the PRISMA fow diagram (Figure 1). A total of 12,899 titles were identifed throughout the screening process, of which 60 abstracts were identifed. Of the abstracts excluded during this phase, 23 were studies assessing the efects of vitamin D status at the time of fracture on healing outcomes, with no intervention given. A total of eleven studies focused on alternative outcomes such as mortality, vitamin D metabolite levels, or bone turnover markers. Another eight studies had an inappropriate intervention which included vitamin D combined with calcium, which could not be compared separately. Tree studies focused on outcomes after elective surgery such as ankle fusion or elective osteotomy, which were also excluded.
Data was extracted by the primary author (T.G) using the data template developed by the Cochrane Consumers and Communication Review Group. Attempts to obtain any missing data were made by contacting the corresponding author of the respective study through email. Any duplicate studies in the selection process were removed manually during abstract screening.

Quality Assessment.
Risk of bias assessments was performed in duplicate using the Risk of Bias Version 2 (ROB-2) [20] tool for randomised control trials, the Risk of Bias in Nonrandomised Studies of Interventions (ROBIN-I) [21] tool for nonrandomised control trials, and the Newcastle-Ottawa Scale (NOS) [22] for observational studies by two authors (T.G & A.G). For RCTs, the quality assessments included bias assessments for intervention deviation, missing outcomes, outcome measurements, and result selection. For NRCTs, bias assessment for confounding, intervention classifcation, and patient selection was also performed in addition to the assessments for RCTs. For cohort studies, the quality assessment domains included cohort selection, cohort comparability, and outcome reporting.

Data Synthesis.
Te patient and study characteristics were reviewed by the two authors (T.G and A.G). Tere was a high degree of heterogeneity noted amongst the studies included. When assessing union rates, the studies focused on a variety of age groups, with diferent fracture types, diferent supplementation regimens, and diferent time endpoints for the primary outcome. Te secondary outcomes assessing functional scores were also very heterogenous, with only two studies reporting the same functional outcome. As a result of this, a meta-analysis was not possible and a descriptive review of the collated data was instead presented throughout the study.

Results
A fnal total of 14 studies, assessing a total of 2734 patients, were included to assess fracture outcomes following vitamin D supplementation [10,14,[23][24][25][26][27][28][29][30][31][32][33][34]. Nine of these were controlled trials of which seven were randomised and two were nonrandomised. Five studies were cohort trials of which one was prospective and four were retrospective. One study identifed in abstract form during the search was later published in full and also included in the analysis [34]. Table 1 summarizes the characteristics of the included studies, in order of the outcome investigated, and outlines the main fndings.

Fracture Union.
Eight studies assessed the efect of vitamin D on clinical or radiological union [14,24,27,[31][32][33][34]. Tere were two studies which did not distinguish between the union types [28,32]. Of the remaining studies, four reported both radiological and clinical outcomes [14,31,33,34], while two studies reported only radiological outcomes [24,27]. Tere were fve studies which reported no signifcant diference in complication rates when supplementing fracture patients [14,24,27,32,34]. Te overall quality of these studies was rated as "low risk of bias" for the controlled trials and "fair quality" for the observational studies (Figures 2 and 3 and Table 2). Tere were three studies which reported a positive efect with supplementation between the groups [28,31,33]. One study found a diference only in early orthopaedic complications (<30 days) but did not fnd any diferences in late complications [28]. Another two studies found statistically signifcant diferences in clinical union, however no changes in radiological union [31,33]. Te quality of these studies ranged from "poor" to "fair" for the two observational trials, Reports not retrieved (n =0 ) Reports sought for retrieval (n =4 ) Reports not retrieval (n =0 ) Records excluded (n =12838) Records identified from: Websites (n =13) Organisations (n =178) Citation searching (n = 5) etc.
Records removed before screening: Identification of studies via other methods Figure 1: PRISMA fowchart summarizing screening process.       Advances in Orthopedics and a "serious risk of bias" was observed in the nonrandomised control trial (Figures 2 and 3 and Table 2).
Another study found that although no diferences in the initial rehabilitation phase were found, supplementation slowed the decline in EQ (5D) scores after six months [25]. Te remaining four studies found no signifcant diferences between the majority of functional outcome scores [10,26,27,29]. Te study by Ko et al. was graded as "fair," while the control trials ranged from "low risk of bias" to "serious concerns" [27]. Te latter refers to Hoikka et al.'s study which assessed the grip strength [10]. One study found no improvement on gait velocity (p � 0.490), but a decrease in pain scores at 28 weeks when supplementation was used (p � 0.037).

Discussion
Te overall impression of these fndings is that vitamin D does very little to infuence fracture healing and subsequent union rates. Two of the three studies which concluded a positive efect of supplementation were generally of a lower quality [31,33]. Te nonrandomised study by Behrouzi et al. saw the authors dividing the groups based on an undocumented vitamin D level and only supplemented the defcient patients with a bolus dose. In a retrospective study by Gorter et al., the groups were misbalanced with nonsupplemented patients (n � 368) outnumbering the supplemented patients (n � 141). Te main diferences in this study are seen only in the smaller cohort of thirty patients which remained defcient despite treatment. Tese patients may either have required higher supplementation in view of severe defciency or were not fully compliant to the treatment.
Te phase II pilot RCT by Slobogean et al. in 2022 serves as a high quality benchmark for future work on the topic [34]. Te authors investigated four groups to assess not only efect of vitamin D on healing but also the efect of its dose, which was a persistent problem amongst the other studies due to a wide variation. Tis study is also unique in which the authors have utilised validated scoring systems to assess clinical (FiX-IT) [35] and radiological union (RUST) [36]. Tis may aid in reducing observation bias for what may be a challenging outcome to assess, as well as provide a numerical outcome which may be pooled in future studies on the topic. Slobogean and colleagues found no improvement on radiological or clinical diference in fracture healing at 3 or 12 months when comparing high dose with low dose, high loading dose with high daily dose, and low dose with placebo. Te only signifcant diference found was in post-hoc analysis for clinical union scores at 3 months when comparing high dose to placebo (p � 0.16), however required confrmation with a larger trial, especially in keeping with the rest of the negative fndings.
When assessing the primary outcomes of clinical and radiological union, the decision was made to include two studies which assessed our primary outcome in combination with other orthopaedic complications such as peri-implant fractures, dislocations, wound infections, and reoperations [24,28]. While these outcomes may be infuenced by factors other than bone healing, such studies were included to not lose quality evidence on the topic. Te justifcation for inclusion was that incidence of these additional complications was low, except for wound infection which may be identifed by the timing of complication manifestation. Regardless of this, results from these studies should be interpreted with caution. For instance, the Ingstad et al. study reports a difference in early complications (<30 days) of borderline statistical signifcance (p � 0.044). Early complications would favour surgical skin infections as opposed to the bone healing complications which is the scope of this review. Furthermore the late complications at 12 weeks (p � 0.242) and 1 year (p � 0.079) were not found to be signifcant and would typically include union complications.
Te included studies in this review had a large degree of heterogeneity in both patient characteristics and study design. Diferent cohorts of patients were assessed, ranging from young trauma patients with tibial or femur shaft fractures to middle aged women with distal radii fractures and to elderly patients with neck of femur fractures. Tis heterogeneity was also seen in supplementation strategy. Numerous organisations have issued varying recommendations on the optimal advised intake, as well as cutofs for vitamin defciency [37][38][39]. Tese vary based on their geographical prevalence, as well as whether the guidelines focus on bone or pleiotropic efects. Tere is no standard recommended daily intake dosing, and the advice is that supplementation should be patient-specifc and preferably adhering to regional guidelines [40]. Some of our included studies made use of bolus dosing which may improve compliance and achieve greater production of vitamin D3 faster, when compared to daily dosing. Te latter, however, ofers more predictable and long-lasting efects [41]. Tere 8 Advances in Orthopedics thus appears to be no widely accepted supplementation strategy yet, and no positive correlation with a particular regimen was identifed in this review. With respect to secondary outcomes, most functional scores showed no beneft with supplementation. Tere was a wide amount of variety in the scoring systems reported, with six studies looking at eleven functional outcome scores. One of these studies, although fnding no signifcant difference in their outcome scores, reported an improvement in the pain component of EuroQoL scoring system at 26 weeks [26]. Tis was not replicated in the rest of the scoring systems, the majority of which included pain components. Te rationale of a presumed positive impact of vitamin D on functional outcome may be due to its efects on muscle strength performance rather than bone healing. Tis correlation has been studied extensively in the literature on both athletes and frail elderly patients to assess fall risk reduction [42][43][44]. A review of this literature shows a number of studies which support this positive correlation; however, difculties with data aggregation and a number of studies conversely not corroborating these fndings mean that the data here are still conficting [45,46]. For instance, in a study by Lee et al., vitamin D levels correlated with improved grip strength in the unafected hand of distal radius fracture patients [47]. Yet in our included studies, active supplementation in neck of femur fractures found no signifcant improvement in grip strength at one, three, or six months [10,26] nor in gait velocity [26].
Hypovitaminosis D appears to be highly prevalent in fracture patients [48]. A study by Gorter et al. demonstrated that vitamin D status at the time of fracture may impact fracture healing [49]. Tis study only found a diference in clinical union, with no diference in radiological union. Our review excluded several recent observational studies assessing vitamin D status on outcomes, which were beyond the scope of our aims as they lacked an intervention. A more recent systematic review of this literature could be the scope of further research. It may well be that although commencing supplementation at the time of a fracture is too late, supplementing to normal vitamin D levels before would improve outcomes if a fracture was to occur.

Strengths and Limitations.
To our knowledge, this systematic review is the frst since the two major systematic reviews on this topic by Gorter et al. and Sprague et al. in 2014 and 2017, respectively. Our systematic review is strengthened by its comprehensive search through multiple databases and gray literature sources. It also includes most studies published after 2017, which had not been previously considered. It also excludes studies using cosupplementation of calcium, such as the Doetsch et al. study which had been included in other reviews [11]. Te included studies capture a broad international cohort of patients, of varying ages and fracture types, which was a fair representation of fracture patients.
Te main limitation of our study is the wide degree of heterogeneity in many of the study designs, including treatment supplementation strategies and outcomes measured. Diferent studies assessed union or functional outcomes at diferent time points or by diferent criteria. Only one study employed a numeric scoring system for union assessment [34]. As a result, the data for union rates between treatment and nontreatment groups were unable to be pooled, permitting only the assessment of trends. Several functional outcome scores were self-reported, as was compliance to treatment in certain studies. Both these factors may lead to response bias. Another limitation is the inclusion of nonrandomised control trials or retrospective studies. Tis evidence is generally of a lower quality and could again potentially increase the risk of bias. Had this systematic review been solely limited to RCTs, none would have shown any positive efect of supplementation on union or functional outcomes. A further limitation of this review is that the studies are underpowered. Incidence of fracture healing complications is generally low, so large numbers of patients would be required, incurring signifcant costs and limiting feasibility. Indeed, this was the case for several studies which halted recruitment early or did not proceed beyond the pilot trial. In this case, the risk of type 2 errors is increased and could alter results.

Conclusion
In summary, it appears that despite initial successes in animal studies, the supplementation of vitamin D in an efort to promote fracture healing and subsequent outcomes does not translate to humans in clinical practice. Tere has been signifcantly more literature on the topic in recent years although the evidence is still plagued by underpowered studies. Te standards for any future quality research on the topic should include power calculations and a feasibility study taking into consideration the regional research framework. Tis should then be followed by a multicentre, randomised control trial of fracture patients, supplemented with varying doses of vitamin D supplementation, and independently assessing union at regular intervals using validated scoring systems.
While hypovitaminosis D is prevalent in fracture patients, the administration of vitamin D alone at the time of diagnosis does not confrm any additional beneft. Of the studies which do show some positive efect for supplementation, these are undermined by nonrandomisation and retrospectivity. Reviewing the literature to determine the efect of vitamin D status at fracture diagnosis is warranted given the increase of recent published studies. Until more high quality and adequately powered trials emerge with evidence to the contrary, routine supplementation of vitamin D alone at the time of fracture diagnosis appears futile in improving bone healing. "fracture"/exp AND "adult"/exp OR "child"/exp OR "human"/exp) AND ("vitamin d"/exp OR "colecalciferol"/exp OR "ergocalciferol"/exp OR "calcitriol"/exp) AND ("fracture healing"/exp OR "bone defect healing" OR "bone fracture healing" OR "bone healing" OR "bone healing, fracture" OR "bone union" OR "consolidation, fracture" OR "fracture bone healing" OR "fracture consolidation" OR "fracture healing" OR "fracture union" OR "healing, fracture" OR "fracture nonunion"/exp OR "delayed fracture healing" OR "delayed union" OR "fracture healing impairment" OR "fracture malunion" OR "fracture nonunion" OR "fracture union delay" OR "fractures, malunited" OR "fractures, ununited" OR "malunion" OR "malunited fractures" OR "non-union" OR "nonunion" OR "nonunion (fracture)" OR "nonunion fracture" OR "ununited fracture" OR "ununited fractures" OR "pain"/exp) AND

Data Availability
Te data used to support the fndings of the study are obtained from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.