Windswept Deformity a Disease or a Symptom? A Systematic Review on the Aetiologies and Hypotheses of Simultaneous Genu Valgum and Varum in Children

Objective: The objective of this study is to create an overview of the possible aetiologies of windswept deformity and to emphasize the points of attention when presented with a case. Methods: A systematic search according to the PRISMA statement was conducted using PubMed, African Journals Online, Cochrane, Embase, Google Scholar, and Web of Science. Articles investigating the aetiology of windswept deformity at the knee in children, and articles with windswept deformity as an ancillary finding were included. The bibliographic search was limited to English-language articles only. The level of evidence and methodological appraisal were assessed. Results: Forty-five articles discussing the aetiology of windswept deformity were included. A variety of aetiologies can be brought forward. These can be divided into the following groups: ‘Rickets and other metabolic disorders’, ‘skeletal dysplasias and other genetic disorders’, ‘trauma’ and ‘descriptive articles without specific underlying disorder’. With rickets being the largest group. Interestingly, in the group without a specific underlying disorder, all patients were from African descent, being otherwise healthy and presented with windswept deformity between two and three years of age. Conclusion: We have presented an overview that may help identify the underlying disorder in children with windswept deformity. A step-by-step guide for clinicians who see a child with windswept deformity is provided. Even though, according to the Oxford level of evidence, most articles have a low level of evidence.


Introduction
In 1975, Oyemade made mention of windswept deformity (WSD), under the term "varovalga", characterised by the formation of a valgus deformity of one knee, and a varus deformity of the contralateral knee [1]. In 1976, the term windswept deformity was used by Fulford et al. to describe the general postural deformity acquired by children with cerebral palsy, in their first weeks of life [2]. Consequently, windswept deformity is used to describe the phenotypical presentation of a varus and valgus deformity, however, the location of this may vary, as well as the underlying pathology. In this article, we will focus on windswept deformity of the knee.
Although the long-term consequences of windswept deformity have not yet been described, it is presumable that it has a similar impact as other angular deformities of the knees. Untreated, it can lead to further deformity, gait abnormalities, limb shortening and osteoarthritis [3].
Apart from varovalgum at the knee, different forms of windswept deformity exist, including hip deformity in children with cerebral palsy [15]. The aim of this study is to perform a systematic search according to the PRISMA statement, in order to analyse which aetiologies of windswept deformity have been published, and assess their level and quality of evidence, as well as any bias. Our aim is to create an overview of the possible aetiologies for windswept deformity and emphasize the points of attention when presented with a case.

Materials and Methods
This systematic review was written according to the PRISMA statement for reporting systematic reviews and meta-analyses [16].

Eligibility Criteria
Types of studies: all articles investigating the aetiology of windswept deformity at the knee and all articles with windswept deformity as an ancillary finding. The language was restricted to articles written in English. There was no restriction on the date of publication.
Types of participants: all patients with a presentation of windswept deformity during childhood (0-18 years) were included. There were no restrictions on gender or race.

Information Sources
The following databases were searched on 16/10/2020: PubMed, African Journals Online, Cochrane, Embase, Google Scholar, and Web of Science. Articles were screened for eligibility, based on title, abstract and the full text. Additionally, the reference lists of the included articles were screened for further identification of any relevant articles and they were included where applicable.
An overview of the search can be found in Appendix A.

Study Selection and Data Extraction
Studies were selected based on the eligibility criteria. First, duplicates were removed, followed by the selection based on title and abstract. The remaining articles were screened for their eligibility based on the full text. The included articles were screened upon a second survey and consensus amongst the authors was reached. The relevant data were extracted and reviewed by the authors.

Level of Evidence and Quality Assessment
The level of evidence was scored according to the Centre for Evidence-Based Medicine (CEBM) by three authors, followed by an assessment of methodological appraisal. The latter was performed according to the Joanna Briggs Institute (JBI) for case reports, case series, cohort studies, case-control studies, and cross-sectional studies [17]. Cut-off values were obtained using a scoring system, where a "yes" answer scored 2 points, "unclear" 1 point, "no" 0 points, and "not applicable" was subtracted from the maximum obtainable score. For each article, the obtained score was divided by the maximum obtainable score which led to a percentage. An article was considered to be of high quality when the score was 75% or higher, moderate between 50% and 75%, and low when the score was lower than 50%. Articles classified as "literature review" did not meet the criteria for level of evidence scoring or critical appraisal. An overview of the methodological appraisal can be found in Appendix B.

Results
An overview of the results and the number of records retrieved from the final search can be found in Figure 1.
The three searches, performed in six databases, yielded a total of 773 records. After removing 193 duplicates, 580 records remained. Of these records, 424 were excluded based on title/abstract screening, leaving 156 records to be assessed based on the full text. From these 156 full-text records, 109 were excluded, as shown in Figure 1. This resulted in a total of 47 publications that were selected, of these, all the references (n = 1201) were screened from which four additional articles were selected. Upon second survey another six articles were excluded based on age > 18 years (n = 1), not being windswept deformity (n = 2) and no clear aetiology stated (n = 3). This resulted in a final number of 45 selected publications.
Although 45 articles is a substantial amount of included articles, only very few focus on the aetiology of windswept deformity, with most articles describing windswept deformity as an ancillary finding. The Oxford CEBM level of evidence regarding these publications is generally low, with most articles ranked as level IV. Additionally, the quality of the articles, assessed as described in the methods, varied greatly, ranging from a score of 50% to 100%.
In Table 1, general information can be found about the articles, including the country of study, study design, the aim of the study, and the main information related to windswept deformity, as well as the Oxford CEBM classification on the methodological quality score.
From the selected articles, a variety of aetiologies for windswept deformity can be brought forward. These can be divided into the following groups:

•
Rickets and other metabolic disorders; • Skeletal dysplasias and other genetic disorders; • Trauma; • Descriptive articles without the specific underlying disorder.  Epidermolytic hyperkeratosis (EHK) with rickets in a 6-year-old boy showed progressive WSD since age 3.
A 7-year-old patient with reported low birth weight and length. Onset of walking at age 4. At 7 years she had a short stature (-4SD) and WSD. A TRPV4 mutation was found. Determine the varieties of angular deformities of the knee in children, in/around Ilorin, Nigeria.
Ten patients with WSD presented in one year. Age distribution between 2 and 5 years.
One bone diseased, while the other appears to be compensating

IV moderate
Teotia et al. [46] India Literature review Report effects of endemic fluoride exposure on metabolic bone disease.
WSD as presentation of bony leg deformity due to high levels of fluoride exposure in drinking water.
Endemic chronic fluoride toxicity n.a. * One patient with metaphyseal dysplasia had WSD and underwent correction for both varus and valgus deformity.
Metaphyseal dysplasia Iib moderate * n.a.: not applicable, articles classified as "literature review" did not meet the criteria for level of evidence scoring or critical appraisal. Literature reviews did not provide the number of WSD patients included and therefore these numbers are not mentioned in the table for literature reviews.
In Table 2, a quantitative summary was made on the demographic data of windswept deformity. A total of 184 patients with windswept deformity were included in Table 2, in 69 cases the gender was described: 45 (65.2%) males and 24 (34.8%) females. Of the 170 patients where the race is described, 108 (63.5%) are of African descent, 11 (6.5%) Asian and 50 (29.4%) Caucasian (37 of the Caucasian patients were retrieved from the same article about pseudoachondroplasia) [49]. One person of Bedouin descent was classified as Middle Eastern (0.6%). Based on the articles found in this study, an overview of the identifying features of each aetiology of windswept deformity is presented in Table 3. Figure 2 shows a flowchart that can be used to find the possible cause of windswept deformity in a child.  Table 3 for more detailed descriptions.     Evidence of (old) fractures Bar-On et al. [14] * Lab findings in vitamin D deficiency and hypocalcemia rickets depend on the phase of rickets. Phase 1: hypocalcemia causes PTH rise, leading to bone resorption and hyperphosphatemia and rise in alkaline phosphatase. This phase has a relative resistance to PTH. Phase 2: PTH rises further and overcomes the resistance. Calcium rises to normal or slightly lower than normal range and phosphate decreases further (renal excretion due to PTH). Phase 3: hypocalcemia returns worse because of depleted reserves, hypophosphatemia persists and further rise of alkaline phosphatase. In phase 3 X-ray abnormalities become visible. ↓ : lower compared to reference standard. ↑ : higher compared to reference standard.

Rickets and Other Metabolic Disorders
About half of the articles included reported rickets in patients with windswept deformity (n = 23). In five articles, accounting for a total of 68 patients, the rickets type was unspecified [1,5,6,8,13]. However, in three articles, for a total of eight patients, rickets was found to be (X-linked) hypophosphataemic, due to diminished reabsorption of phosphate in the kidneys [19,23,40]. Nutritional rickets due to vitamin D or calcium deficiency was found in another 12 articles, accounting for another 36 patients with windswept deformity [7,9,10,22,24,26,28,29,37,[41][42][43]. In four patients, from three articles, rickets was caused by a skin disorder, namely epidermolytic ichthyosis, which is present at birth [34,44,47]. This adds up to 116 patients with windswept deformity, likely due to rickets of varying types.

Trauma
In one patient, trauma seemed to be the cause of the windswept deformity [14]. This patient was a young female with a history of fractures, often untreated due to congenital insensitivity to pain, presumably being the cause of the angular deformities of the legs.

Descriptive Articles without Specific Underlying Disorder
We grouped six articles together with a total of 36 patients, in which different hypotheses for the cause of windswept deformity were brought forward: a combination of mechanical pressure and a period of illness (similar cause as Blount disease) (n = 8) [11,38]; excessive or early weight-bearing, dietic or ethnicity (n = 15) [1,13]; a combination of the previously mentioned factors (n = 3) [4]; compensation, where only one side is diseased, while the other side compensates (n = 10) [12]. All of these hypotheses were (partly) based on mechanical loading or weight-bearing. The patients in these articles were all African and typically presented with windswept deformity between the age of 2 and 3 years, being otherwise healthy. None of these patients showed signs of (healed) rickets.

Discussion
This systematic review gives an overview of all previously published aetiology hypotheses for windswept deformity. Windswept deformity is generally limited to an individual with genu valgum on one side, and genu varum on the contralateral side. Further specific phenotypical descriptions vary depending on the aetiology.
Our results display that windswept deformity can be a manifestation of a broad variety of pathologies. However, in patients in whom no underlying illness was found, the deformity was deemed idiopathic, and the following hypotheses are brought forward: weight-bearing (due to excessive weight or early walking) [1,11,13,38], epiphyseal instability [4], stress factors (illness) [4] and geographical genetic factors [1,4,13]. Though these hypotheses are comparable between articles, they lack supporting evidence.
Rickets of different types was found to be the most common pathology manifesting windswept deformity. Although it appears to be the most frequent cause of windswept deformity, windswept deformity is far from the most common presentation of rickets. Thacher [7], presents different clinical features and their utility in predicting radiologically active rickets; only 14% of the children with radiologically active rickets have windswept deformity and the probability that a child with windswept deformity has radiologically active rickets is 41%. Unfortunately, no further research has been conducted on factors influencing the development of windswept deformity, genu valgum or genu varum in children with rickets. Bhimma et al. [22] concluded that mainly vitamin D deficiency is responsible for rickets, however, this may be aggravated by calcium deficiency. These two deficiencies may explain the geographical and ethnic distribution of windswept deformity. As seen in Table 2, most of the patients with windswept deformity were African, and about half of the included studies were performed in sub-Saharan African countries. The diet of rural African children is often low in milk and other dairy products, hence leading to a reduced calcium intake from the diet, despite the fortification of products with calcium [51]. The vitamin D deficiency may be explained by the increased sunlight exposure required for black children [22], or by cultural and/or religious factors that limit the exposure to sunlight, such as covering garments or veils. Hypophosphataemic rickets is most commonly found in its X-linked inheritance form and causes rickets due to its defects in renal handling of phosphorus [52]. Furthermore, epidermolytic ichthyosis is described in windswept patients caused by rickets due to marked hyperkeratosis of the skin [53]. Consequently, there is a decreased synthesis of vitamin D in the epidermis stimulating parathyroid hormone secretion, and a higher risk of rickets [54].
The cases of windswept deformity occurring in patients with other metabolic disorders are often comparable to the pathophysiology behind the different forms of rickets. In calcium deficiency rickets, patients present with secondary hyperparathyroidism as the low calcium levels stimulate the increased production of PTH. Similarly, primary hyperparathyroidism, distal renal tubular acidosis (dRTA) and renal osteodystrophy cause abnormalities in calcium and phosphorus levels. Exposure to high levels of fluoride in drinking water may lead to a decrease in strength by altering the structural integrity of the bone microarchitecture, which possibly leads to skeletal deformities, such as windswept deformity. Teotia et al. describe the difficulty in differentiating calcium-deficiency rickets from fluoride toxicity, and believe that every child presenting with bone disease in areas endemic to fluorosis is a case of skeletal fluorosis until proven otherwise [46].
A variety of skeletal dysplasias have been described in cases of windswept deformity. Often, these dysplasias are caused by a specific mutation, and a positive family history may therefore be present. Additionally, these patients often present additional symptoms, for example, brachydactyly and craniosynostosis for SEMDFA [45]. PSACH is caused by a mutation in the COMP gene and is usually inherited in an autosomal dominant manner. Although PSACH is not usually discovered until the age of 2-3 years, when disproportionate short stature, waddling gait and evidence of increased joint laxity starts to develop, these features should be used to distinguish PSACH as the cause of windswept deformity [49]. Most patients with a syndrome or dysplasia presenting with windswept deformity have other clinical features, such as malformations of the eyes and face, which can be used to identify the underlying syndrome.
Only a single article, by Bar-On et al. [14], describes trauma leading to windswept deformity. The precise location of the fractures is not described, and therefore, it is impossible to conclude whether this case of windswept deformity was due to compensation for one malformed leg, or if the trauma occurred in both legs, leading to the deformity. The child in this single case suffered from congenital insensitivity to pain, and therefore had a higher risk of multi-trauma leading to lower limb deformities.
The remaining articles, in which other hypotheses for windswept deformity were described, explained its development in patients where there is no underlying cause found. We found no explicit evidence of early walking or excessive weight. Although the beforementioned causes were poorly reported, all patients in these studies had comparable stories. All were of African descent, otherwise healthy, with no signs of healed rickets and the age of onset was between 2 and 3 years of age. There might be geographical or genetic factors that could explain the distribution of windswept deformity in the unspecified group.
When a child presents with windswept deformity, the number of possible underlying causes is extensive, and a complete overview is helpful to make the correct diagnosis. Table 3 shows an overview of the identifying features of each aetiology found in this review. To find a possible cause we advise extensive history taking, including family history (to exclude genetic dysplasias or syndromes) and developmental history; detailed physical examination, looking for clinical features suggestive of rickets (thickened wrists and ankles, and/or rickety rosary), or other identifying features associated with dysplasias or syndromes (facial dysmorphisms, ligamentous laxity, deformities at multiple joints, etc.); and additional testing, such as X-rays of the lower limb and wrist, looking for evidence of rickets or other abnormalities which might fit specific skeletal dysplasias, blood panel (alkaline phosphatase, calcium, phosphate, magnesium, PTH, 25-OH-vitamin D, albumin, creatinine and fluoride) and spot urine test (calcium, phosphate, creatinine and Ph) to exclude or identify rickets and/or other metabolic causes. If required, additional blood tests (e.g., chloride, potassium, 1.25-di-OH-vitamin D, arterial blood gas (ABG) and FGF23) can be performed. Genetic analysis may be indicated when a specific dysplasia or syndrome (COMP, MATN3, TRP4, RSPRY-1 and HSPG2 mutation), or hypophosphataemic rickets (PHEX mutation) is suspected. Additionally, a renal ultrasound can be performed in the case of distal renal tubular acidosis or a skin biopsy to confirm hyperkeratosis. However, when the cause of windswept deformity cannot be found it may be multifactorial, including mechanical loading or weight-bearing.
Despite the large number of articles included in this review, only very few focus on the aetiology of windswept deformity, and most articles have a different aim, describing the deformity as an ancillary finding. Additionally, most articles have old publishing dates. On the other hand, most articles occur in low and middle-income countries, which may result in an underestimation of the problem as less research is conducted in low and middle-income countries. Hence, the specific data available on patients with windswept deformity is often limited. Despite not being found in the literature as causes of windswept deformity, logically there are more underlying disorders that can cause windswept deformity, for instance, other skeletal dysplasias or metabolic disorders (e.g., hypomagnesemia or hypo-albuminemia).

Conclusions
Currently, the existing evidence on the aetiology of windswept deformity of the knee shows a broad spectrum of underlying causes that may lead to its development. However, when none of these specific causes can be identified, it appears that the aetiology is multifactorial, resting on the hypotheses of weight-bearing, epiphyseal instability, stress factors and geographical genetic factors. Presently, not enough evidence is available to confirm these hypotheses, and more research is necessary. Nevertheless, we have presented an overview, which helps guide clinicians presented with a case of windswept deformity. A thorough (family and developmental) history, followed by physical examination, and additional tests, such as X-rays, blood panels, urine tests, renal ultrasound and genetic analysis, may help identify the underlying disorder.