Vitamin D deficiency causes rickets in an urban informal settlement in Kenya and is associated with malnutrition

Abstract The commonest cause of rickets worldwide is vitamin D deficiency, but studies from sub‐Saharan Africa describe an endemic vitamin D‐independent form that responds to dietary calcium enrichment. The extent to which calcium‐deficiency rickets is the dominant form across sub‐Saharan Africa and in other low‐latitude areas is unknown. We aimed to characterise the clinical and biochemical features of young children with rickets in a densely populated urban informal settlement in Kenya. Because malnutrition may mask the clinical features of rickets, we also looked for biochemical indices of risk in children with varying degrees of acute malnutrition. Twenty one children with rickets, aged 3 to 24 months, were identified on the basis of clinical and radiologic features, along with 22 community controls, and 41 children with either severe or moderate acute malnutrition. Most children with rickets had wrist widening (100%) and rachitic rosary (90%), as opposed to lower limb features (19%). Developmental delay (52%), acute malnutrition (71%), and stunting (62%) were common. Compared to controls, there were no differences in calcium intake, but most (71%) had serum 25‐hydroxyvitamin D levels below 30 nmol/L. These results suggest that rickets in young children in urban Kenya is usually driven by vitamin D deficiency, and vitamin D supplementation is likely to be required for full recovery. Wasting was associated with lower calcium (p = .001), phosphate (p < .001), 25‐hydroxyvitamin D (p = .049), and 1,25‐dihydroxyvitamin D (p = 0.022) levels, the clinical significance of which remain unclear.

high-latitude. In tropical or subtropical climes, there is generally abundant exposure to sunlight, but vitamin D deficiency may arise in association with risk factors such as darker skin pigmentation, atmospheric pollution, and covering skin for religious or cultural reasons (Baroncelli et al., 2008;Elder & Bishop, 2014;Trilok Kumar et al., 2015). However, in such settings, vitamin D deficiency may not be the only cause of rickets. Studies from several sub-Saharan African countries have implicated chronic dietary calcium deficiency in the pathogenesis of rickets and demonstrated nutritional cure with calcium supplementation alone (Thacher et al., 1999;Pettifor, 2004;Prentice et al., 2008;Braithwaite et al., 2016). The aetiology may be multifactorial, because calcium deficiency tends to exacerbate the impact of marginal vitamin D status, and may further interact with other nutritional deficiencies, such as of iron (Braithwaite et al., 2012;Pettifor, 2014).
A recent report highlighted the existence of a significant burden of rickets amongst children living in an informal settlement in Nairobi, Kenya (Edwards et al., 2014). Most children with clinically-defined rickets were younger than 2 years, and rickets was associated with acute malnutrition and developmental delay (Edwards et al., 2014). In this paper, we describe a similar group of children presenting to a primary care facility in Mathare, another of Nairobi's informal settlements. Via detailed dietary and biochemical evaluation, we provide evidence that deficiencies in both calcium and vitamin D are likely to play a role in the aetiology of rickets in urban Kenyan settings.
Because malnutrition may mask the clinical features of rickets, we also looked for biochemical features indicating rickets risk in children with varying degrees of acute malnutrition. To this end, we demonstrate an association between subclinical vitamin D deficiency and severe wasting.

| Study design
The study was designed primarily as a case-control study comparing children with rickets (cases) to children without rickets or acute malnutrition (controls). Additionally, children with severe and moderate acute malnutrition (SAM and MAM) were compared to controls.

| Setting
The study took place between July 2013 and April 2014 at the Baraka Health Centre (BHC), Mathare, Nairobi. BHC is supported and administered by the nongovernmental organisation "German Doctors Nairobi" and provides free healthcare to children aged less than 5 years (http://www.germandoctorsnairobi.co.ke/). Mathare is the second largest urban informal settlement in Kenya. It is home to at least 200,000 people, who mainly live in overcrowded ironsheet housing with limited access to safe water and sanitation facili-

| Participants
Children with rickets were recruited from among those presenting to BHC. Inclusion criteria for cases were age 3 to 24 months, with clinically diagnosed rickets, based on the presence of one or more of wrist widening, rachitic rosary, swollen knees, bow legs, or bone pain on walking (Thacher et al., 2002). Where wrist widening was the only feature, clinical evaluation was supplemented with radiographic assessment (right wrist X-ray), and children were only included in the Children were excluded (from all of the groups) if they required emergency medical care, if they were being treated for tuberculosis, if they had a fracture within the preceding 3 months, and in the absence of informed parental consent for participation. Children were also excluded if they were known or found to have HIV infection, or if they were younger than 12 months old and known to be HIVexposed.

| Study procedures
Participants were clinically assessed by a senior clinician (KDJJ or CUH).
Demographic characteristics were recorded including anthropometric

Key messages
• In urban Kenya, rickets in young children is usually related to vitamin D deficiency.
• Children with acute malnutrition commonly have low vitamin D status without clinically evident rickets. measurements. A contextually-relevant, recall-based food frequency questionnaire for the main carer was developed based on the most frequently consumed sources of calcium and phosphate in this setting, in order to estimate intake of calcium and phosphate beyond that derived from breastfeeding over a 1 week period. The questionnaire was administered by a trained field worker in Kiswahili or English as preferred.
Nutrient intakes were calculated with reference to the United States Department of Agriculture's National Nutrient Database for Standard Reference (Release 28) (United States Department of Agriculture, 2015). Right wrist X-ray changes were graded according to the method of Thacher et al., whereby Grade 1 rachitic changes were widened growth plate and irregularity of metaphyseal margins, and Grade 2 were metaphyseal concavity and fraying of margins (Thacher et al., 2000).

| Statistical analysis
Analysis was performed in STATA Version 12.0. Weight-for-length (WLZ), length-for-age (LAZ), weight-for-age and occipitofrontal circumference-for-age z-scores were calculated using WHO Anthro

| Clinical and biochemical phenotype of rickets
Most of the children with rickets had wrist widening and rachitic rosary as opposed to lower limb features (Table 1). Developmental delay, acute malnutrition, and stunting were common (Table 1).
Children with rickets were relatively hypocalcaemic, hypophosphataemic, and had low 25(OH)D, compared to controls (Table 2). ALP and PTH were elevated (Table 2). Rickets was not associated with elevated 1,25(OH) 2 D (as has been reported in calcium deficiency rickets), and there were no detectable differences in inflammatory markers (CRP and AGP) or in dietary intake of either calcium or phosphate between children with rickets and controls (Table 2). (Pettifor & Prentice, 2011). Although some children with rickets had very low 25 (OH)D, six (29%) of them had levels above 30 nmol/L (Figure 1).
Twenty-five(OH)D was not associated with age (supplementary Figure   1). To explore the cause of rickets in these children, we plotted

| Relationship between anthropometric and biochemical variables among children without rickets
Amongst children with SAM, the nine with kwashiorkor had a different biochemical profile from those with severe wasting (MUAC <115 mm): kwashiorkor was associated with lower albumin and ALP, and with higher 25(OH)D and CRP (supplementary Figure 2). In order to assess relationships between biochemical variables and degree of wasting (WLZ and MUAC) and stunting (LAZ), we performed linear regression excluding those children with rickets. Degree of wasting (WLZ) and absolute MUAC were associated with lower serum adjusted calcium,  Figure 3).
Stunting (adjusted for age) was also associated with lower phosphate (Table 3)

| DISCUSSION
Vitamin D deficiency appeared to play a key role in the pathogenesis of rickets in young children in an informal settlement in Nairobi. Children with rickets tended to present with wrist widening or rachitic rosary, and acute malnutrition and developmental delay were common comorbidities. The phenotype of rickets was clinically and biochemically distinct from that caused by calcium deficiency, which is the more prevalent form reported in several other sub-Saharan African settings (Thacher et al., 1999;Pettifor, 2004;Prentice et al., 2008;Braithwaite et al., 2016). This has important implications for treatment, for calcium supplementation alone has proven to be adequate for treatment of rickets caused by calcium deficiency. In Nairobi, it is likely that provision of supplemental vitamin D would be required (Thacher et al., 1999;Oginni et al., 2003). Furthermore, in this setting, acute malnutrition was associated with relatively low vitamin D levels, and 33% of children with SAM had absolute vitamin D levels in the deficient range.
The current study has not addressed the underlying causes of vitamin D deficiency in this setting. However, the fact that Nairobi's climate should support sufficient vitamin D biosynthesis raises important behavioural questions around sunlight exposure. Edwards et al. reported that 71% of children with rickets in another Nairobi informal settlement had less than 3 hours' sunlight exposure per week (Edwards et al., 2014). Poverty and urban living may act synergistically to limit infants' exposure to sunlight. In Mathare, the imperative for mothers to seek employment very early in the postnatal period means that infants are often looked after in indoor informal daycare facilities  Note. Outputs of linear regression analyses. Output for LAZ is adjusted for age. LAZ = length-for-age z-score; MUAC = mid-upper arm circumference; WLZ = weight-for-length z-score.
*Calcium is adjusted for albumin as described in the methods section.
breastmilk (Salameh et al., 2016). Understanding the health and developmental consequences of growing up in sub-Saharan African urban informal settlements is important because urbanisation is proceeding at a high rate, does not appear to be associated with reduction in poverty, and is leading to dramatic increases in the number of people living in such circumstances (Henderson et al., 2013;Awumbila, 2014).  Table 1).
Whether these children had historic, healed, or "mild" rickets is difficult to assess, but they exemplify the fact that clinical diagnosis alone fails to reflect heterogeneity in the underlying pathophysiology and intensity of disease activity in a way that may be meaningful for treatment.
At present, there are no internationally accepted diagnostic criteria for rickets. Thacher and colleagues' radiographic scoring method was designed for, and validated in, children older than 12 months who were independently mobile, but in Nairobi, rickets affects very young children and is associated with a high frequency of motor developmental delay (Thacher et al., 2002;Edwards et al., 2014;Munns et al., 2016).  (1,25(OH) 2 D) with weight-for-length z-score. p values are for linear regression analyses corrected as described SAM is associated with significant long-term health consequences, and the development of strategies targeted at healthy growth and development in the recovery phase is a research priority (Briend & Berkley, 2016). Vitamin D deficiency is a plausible mechanism explaining at least some of the increased vulnerability to severe infectious disease in SAM, because a number of studies have indicated, it increases the risk of respiratory tract infections (Esposito & Lelii, 2015). Public health strategies to improve vitamin D levels may provide benefit beyond prevention and treatment of rickets, but should be tested in rigorous controlled trials. Solar bottle bulbs (which involve the incorporation of water-filled plastic drink containers into the roof of a settlement) represent an innovative approach to increasing indoor sunlight exposure in urban informal settlements (www.literoflight.org), however the use of common polyethylene terephthalate plastic bottles means that the UVB wavelengths important for vitamin D production are filtered out (Sackey et al., 2015). Adaptation of this approach may be feasible and could provide a cheap and durable solution.
In summary, we have identified a group of young children living in an urban informal settlement in Nairobi who have rickets that appears to be predominantly caused by vitamin D deficiency and who would be likely to require vitamin D supplementation for full recovery. Development of standardised tools for diagnosis of rickets in this age group that could help establish the burden of disease should be a clinical and research priority. The apparent preponderance of vitamin D deficiency over calcium deficiency as cause of rickets in this setting prompts larger and more detailed epidemiologic studies in this and other settings.