Regional differences in short stature in England between 2006 and 2019: A cross-sectional analysis from the National Child Measurement Programme.

BACKGROUND: Short stature, defined as height for age more than 2 standard deviations (SDs) below the population median, is an important indicator of child health. Short stature (often termed stunting) has been widely researched in low- and middle-income countries (LMICs), but less is known about the extent and burden in high-income settings. We aimed to map the prevalence of short stature in children aged 4-5 years in England between 2006 and 2019. METHODS AND
FINDINGS: We used data from the National Child Measurement Programme (NCMP) for the school years 2006-2007 to 2018-2019. All children attending state-maintained primary schools in England are invited to participate in the NCMP, and heights from a total of 7,062,071 children aged 4-5 years were analysed. We assessed short stature, defined as a height-for-age standard deviation score (SDS) below -2 using the United Kingdom WHO references, by sex, index of multiple deprivation (IMD), ethnicity, and region. Geographic clustering of short stature was analysed using spatial analysis in SaTScan. The prevalence of short stature in England was 1.93% (95% confidence interval (CI) 1.92-1.94). Ethnicity adjusted spatial analyses showed geographic heterogeneity of short stature, with high prevalence clusters more likely in the North and Midlands, leading to 4-fold variation between local authorities (LAs) with highest and lowest prevalence of short stature. Short stature was linearly associated with IMD, with almost 2-fold higher prevalence in the most compared with least deprived decile (2.56% (2.53-2.59) vs. 1.38% (1.35-1.41)). There was ethnic heterogeneity: Short stature prevalence was lowest in Black children (0.64% (0.61-0.67)) and highest in Indian children (2.52% (2.45-2.60)) and children in other ethnic categories (2.57% (2.51-2.64)). Girls were more likely to have short stature than boys (2.09% (2.07-2.10) vs. 1.77% (1.76-1.78), respectively). Short stature prevalence declined over time, from 2.03% (2.01-2.05) in 2006-2010 to 1.82% (1.80-1.84) in 2016-2019. Short stature declined at all levels of area deprivation, with faster declines in more deprived areas, but disparities by IMD quintile were persistent. This study was conducted cross-sectionally at an area level, and, therefore, we cannot make any inferences about the individual causes of short stature.
CONCLUSIONS: In this study, we observed a clear social gradient and striking regional variation in short stature across England, including a North-South divide. These findings provide impetus for further investigation into potential socioeconomic influences on height and the factors underlying regional variation.

Introduction

Linear growth is an important indicator of a child’s health. Poor early life growth is associated with impaired physical, neurodevelopmental, and educational outcomes, which hamper children’s abilities to survive and thrive [1], and increase later life risk of chronic disease and premature mortality [2]. Stunting, which is defined as a height for age more than 2 standard deviations (SDs) below the reference population median, is a term generally confined to low- and middle-income countries (LMICs). Short stature may therefore be a hidden problem in high-income countries, particularly in economically deprived areas, and an overlooked marker of child well-being.

Because the focus of UK public health programmes on growth is on body mass index, there are no recent data describing the prevalence of child short stature. Surveys in 2011 and 2017, and a roundtable discussion in 2014 conducted by The Patients Association, highlighted anecdotal evidence of an increasing burden of child undernutrition and identified this as a public health priority [3,4]. Moreover, the “dual burden” of stunting and overweight is increasingly being recognised globally [5].

In England, children are measured at ages 4 to 5 and 10 to 11 years through the National Child Measurement Programme (NCMP), which was introduced in 2006 to assess overweight and obesity in primary school children [6]. Data on obesity prevalence and trends are presented in user-friendly dashboards, which include aggregated data by region, ethnicity, and index of inequality [7]. Although the heights of all participating children are collected, there are no equivalent dashboards for prevalence or trends in short stature, and children with linear growth faltering are not routinely identified or directed to health services through the NCMP.

We set out to leverage these national data on height in early childhood using the 13 years of available data to map short stature prevalence across England. Our hypothesis was that there are geographical hotspots of short stature in England, which are concentrated in socioeconomically deprived areas.

Methods

This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (S1 Text). The study forms part of a wider project on child growth and development. The original analysis plan, first presented as part of the funding application for this project, is presented in S2 Text.

Study sample

We used available data from children aged 4 to 5 years in the NCMP for 13 school years between 2006 and 2019. All children attending a state-maintained primary school in the country (94% in 2010 [8]) are invited to participate in the NCMP. The programme has a very high (93%) average participation rate [6].

NCMP data collection

Detailed protocol descriptions are available for all NCMP procedures [9]. Briefly, children’s height and weight were measured by a school nurse or other trained staff member in each participating school using calibrated weight scales and stadiometers. Height and weight were measured in centimetres and kilogrammes to the first decimal place, respectively. Children who declined to participate, or whose parents withdrew them from the programme, were not measured. Children with known growth disorders, Down syndrome, and children who could not stand unaided or could not be measured accurately were measured, but their results were not uploaded to the NCMP system. NCMP data were validated at source, where records with missing mandatory fields were rejected, and records with improbable fields were flagged for the data provider to check before submission. Details on error and warning ranges for height and weight have been published elsewhere [10].

Data were available for 13 years between school years 2006 to 2007 and 2018 to 2019. NCMP data collection is conducted by local authorities (LAs), overseen by Public Health England (PHE), and data are managed by NHS Digital. NHS Digital makes a limited version of the NCMP dataset publicly available for analysis, which does not contain data that could lead to the identification of any child [11]. Access to the full NCMP dataset for the current analysis was obtained through collaboration with PHE, and analyses were conducted at PHE to comply with best data protection practices. Integrated Research Application System (IRAS) approval was not required for the current study, which was registered with the Clinical Governance Department at Barts Health NHS Trust.

Measures

For the current analysis, height-for-age standard deviation scores (SDSs) were derived for each child using UK WHO reference values [12] using the zanthro package in Stata 15 [13]. Children with SDS above 5 or below −5 were excluded. Short stature was defined a height SDS below −2.0. This cutoff was chosen for comparability with international definitions of stunting. Model covariates included ethnicity and index of multiple deprivation (IMD). Ethnicity was collected by schools for each child and was recoded for this analysis into 6 categories based on the UK government census categorisation: White (White British and White Other); Black (Black African, Black Caribbean, and Black Other); Indian; Pakistani or Bangladeshi; Mixed; and Other ethnicity. The IMD is a composite, area-level measure of relative deprivation. It comprises 7 measures of deprivation including employment deprivation, income deprivation, and crime. The IMD ranks all small areas (lower super output areas) in the country from most deprived (#1) to least deprived (#34,753 in 2019). NCMP provides an IMD decile for each child based on their postcode; we use both IMD decile and quintile in our analysis.

Statistical analyses

Prevalence of short stature by sex, region, ethnicity, IMD, and time period was examined using Stata 15, and between-group differences were evaluated using chi-squared tests. We assessed whether change over time in short stature prevalence was consistent across levels of area deprivation by examining differences in rates of change by IMD quintile. We fitted a logistic regression model of the probability of short stature by IMD and NCMP year and included an IMD#year interaction term. We visualised change over time using Equiplot, a method for comparing inequality between factors such as time periods [14]. Spatial analyses were conducted using SaTScan [15] to assess geographic clustering of short stature. SaTScan analyses geographical data to identify disease or other event clusters and tests for statistical significance. The software scans a geographical area using a circular window, which is then positioned at each point in a given geographical grid and varies in size from 0 to an upper limit predetermined by the researcher. A null hypothesis of equal risk inside and outside the circular window is tested using likelihood ratios. The most probable cluster is chosen by this scanning system as that with the highest likelihood ratio. The software then systematically scans for the next most probable cluster. Geographic aggregation units for all spatial analyses were lower tier LAs. In England, lower tier LAs include LA districts, unitary authorities, metropolitan districts, and London boroughs. We scanned the data unadjusted, then adjusted for ethnicity and adjusted for both ethnicity and IMD. A Bernoulli probability model for count data was used; model specifications and details of the adjustment strategy are given in Tables A–D in S3 Text. We initially analysed data for all years and then further assessed whether the clusters identified were consistent over time by analysing 4 NCMP time periods separately (school years 2006 to 2010; 2010 to 2013; 2013 to 2016; and 2016 to 2019). We present all statistically significant clusters (threshold P < 0.05). P values were determined by log likelihoods being greater than a critical value established for each model through standard Monte Carlo and Gumbel approximation [15].

Supplemental analyses

We repeated all analyses using very short stature, defined as height <−2.67 SDS (equivalent to <0.4th percentile), which is used as a referral cutoff for investigation of short stature in the UK [16]. Due to the exclusion of some of the sample from spatial analyses because of missing ethnicity and IMD data, a sensitivity analysis was conducted running an unadjusted SaTScan model on the full dataset to check for bias. We also examined the percentages of children by sex, ethnicity, and region in our dataset against census and projections data from 2011 to -2012 to assess whether our sample matched the population.

Ethical approval

The current study is a secondary analysis of NCMP data, and, therefore, ethical approval was not required. The study was registered with the Clinical Governance Department at Barts Health NHS Trust.

Results

Prevalence of short stature

A total of 7,299,208 reception-aged children participated in the NCMP between 2006 and 2019. NCMP response rates increased from 83% in the 2006 to 2007 school year to 95% in 2018 to 2019. From this population, we included 7,062,071 children aged 4 to 5 years at measurement who had valid locality data in our analysis of country prevalence. Of these, 5,765,707 (82%) had valid ethnicity and IMD data and were included in unadjusted and adjusted spatial models of short stature clustering. Details of inclusion in the sample are given in Fig 1. Missingness by sample characteristics is given in S1 Table. Ethnicity was missing in 18% of children, while IMD was missing in very few children (0.01%). Missingness was explained in part by improvements in NCMP data validation processes over time, with a larger percentage of missing data in the first time period. Missingness was also related to region but did not appear to be related to height or short stature. We also examined 2011 Census data as well as data projections for child population numbers by region and found that the percentages of each sex, ethnic group, and region matched the population very closely (S2 Table).

Fig 1

Sample inclusion and exclusion flowchart.

IMD, index of multiple deprivation; NCMP, National Child Measurement Programme; SDS, standard deviation score.

Table 1 shows sample characteristics and prevalence of short stature. The mean (SD) age of children in the sample was 60.0 (4.0) months, and the mean (SD) height was 109.6 (5.1) cm. In the period between 2006 and 2019, 1.93% (95% confidence interval (CI): 1.92 to 1.94) of children aged 4 to 5 years were short for their age in England. Girls had a higher prevalence of short stature, with 2.09% (2.07 to 2.10) of girls compared with 1.77% (1.76 to 1.78) of boys being shorter than 2 SDs below the sex-specific mean (P < 0.001). Children in the north versus the south of England had a higher probability of short stature, with the highest prevalence in Yorkshire and the Humber (2.18% (2.15 to 2.22)) and the lowest in London (1.57% (1.55 to 1.60)). In individual LAs, short stature ranged from 0.97% (0.85 to 1.10) in Richmond upon Thames in London to 3.92% (3.69 to 4.16) in Blackburn with Darwen in the North West.

Table 1

Short stature (height <−2.0 SDS) prevalence by sample characteristics (n = 7,062,071).

Characteristic	Mean (SD) % (n)	Short stature, % [95% CI]
Age in months
Mean (SD)	60.0 (4.0)[60.0; 60.0]	-
Height
Mean height cm (SD) [95% CI]	109.6 (5.1)[109.6; 109.6]	-
Mean height SDS (SD) [95% CI]	0.11 (1.0)[0.11; 0.11]	-
Sex % (n)
Male	51 (3,608,608)	1.77 [1.76; 1.78]
Female	49 (3,453,463)	2.09 [2.07; 2.10]
Government Office Region %
North East	5 (351,119)	2.12 [2.08; 2.17]
North West	14 (974,826)	2.14 [2.12; 2.17]
Yorkshire and the Humber	10 (719,264)	2.18 [2.15; 2.22]
East Midlands	9 (597,755)	2.08 [2.04; 2.11]
West Midlands	11 (788,566)	2.05 [2.02; 2.08]
East of England	11 (795,006)	1.89 [1.86; 1.92]
London	16 (1,110,081)	1.57 [1.55; 1.60]
South East	15 (1,080,309)	1.74 [1.72; 1.77]
South West	9 (645,145)	1.86 [1.83; 1.89]
Ethnicity % (n)
White British and White Other	62 (4,380,202)	1.95 [1.94; 1.96]
Black African, Caribbean, and Other	5 (314,293)	0.64 [0.61; 0.67]
Indian	2 (168,896)	2.52 [2.45; 2.60]
Pakistani and Bangladeshi	5 (349,060)	2.18 [2.13; 2.23]
Mixed	4 (304,769)	1.58 [1.53; 1.62]
Other	4 (248,874)	2.57 [2.51; 2.64]
Missing	18 (1,295,977)	1.96 [1.94; 1.98]
IMD (decile)a % (n)
1	14 (984,279)	2.56 [2.53; 2.59]
2	12 (860,767)	2.24 [2.21; 2.27]
3	11 (768,760)	2.09 [2.06; 2.12]
4	10 (698,317)	2.00 [1.96; 2.03]
5	9 (655,926)	1.88 [1.84; 1.91]
6	9 (625,186)	1.75 [1.72; 1.79]
7	8 (599,197)	1.70 [1.67; 1.74]
8	9 (609,677)	1.62 [1.59; 1.65]
9	9 (623,088)	1.51 [1.48; 1.54]
10	9 (636,394)	1.38 [1.35; 1.41]
Missing	0 (480)	3.75 [2.39; 5.85]
Time period % (n)
2006–2010	24 (1,723,900)	2.03 [2.01; 2.05]
2010–2013	24 (1,688,054)	1.97 [1.95; 1.99]
2013–2016	26 (1,816,273)	1.89 [1.87; 1.91]
2016–2019	26 (1,833,844)	1.82 [1.80; 1.84]

a Note: IMD deciles are ordered from most deprived (1) to least deprived (10).

CI, confidence interval; IMD, index of multiple deprivation; SD, standard deviation.

There was considerable heterogeneity by ethnicity, with White children being 3 times more likely to have short stature than Black children (1.95% (1.94 to 1.96) versus 0.64% (0.61 to 0.67); P < 0.001). Indian children and children in the other ethnic category had the highest prevalence of short stature (2.52% (2.45 to 2.60) and 2.57% (2.51 to 2.64), respectively), which were both significantly higher than White children (<0.001). Short stature was also linearly associated with IMD, with prevalence in the most deprived decile being nearly twice that of the least deprived decile (2.56% (2.53 to 2.59) versus 1.38% (1.35 to 1.41); P < 0.001). Data for 4 time periods showed that the prevalence of short stature among 4- to 5-year-old children declined between 2006 and 2019 from 2.03% (2.01 to 2.05) to 1.82% (1.80 to 1.84) (P < 0.001).

Assessment of change over time by IMD showed that short stature declined across all IMD quintiles between 2006 and 2019. Prevalence declined from 2.67% to 2.21% in the most deprived quintile (quintile 1) and from 1.68% to 1.46% in the least deprived quintile (quintile 5). Full data are presented in S3 Table. A logistic regression model (S4 Table) showed that the decline from 2006 to 2019 within each quintile was significant, although the least deprived quintiles showed slower declines in the prevalence of short stature than the most deprived quintiles, as shown in an Equiplot visualisation (Fig 2).

Fig 2

Equiplot of short stature percentage by IMD quintile and NCMP school year.

IMD Q refers to IMD quintiles, where IMD Q1 is the most deprived quintile, and IMD Q5 is the least deprived quintile. IMD, index of multiple deprivation; NCMP, National Child Measurement Programme; SDS, standard deviation score.

Spatial analyses

A total of 326 lower tier LAs existed in England in 2018, 2 of which were aggregated with neighbouring areas due to low numbers (City of London, aggregated with Hackney, and the Isles of Scilly, aggregated with Cornwall), giving a total for analysis of 324 LAs. Short stature clusters adjusted for ethnicity are shown in Fig 3, and cluster descriptions are given in Table 2. All clusters identified in this model were highly significant (P < 0.001). There was geographical heterogeneity in short stature prevalence across the country and within regions. A model adjusted for ethnicity found 8 clusters, mostly distributed around the Midlands and North of England, as well as 2 small clusters in London. The clusters with the highest adjusted relative risk (RR) for short stature were in the East Midlands (Leicester; RR: 1.50), East of England (Great Yarmouth; RR: 1.36), and London (Brent; RR: 1.34) and tended to be located in urban areas. Clusters in the North of England had lower RRs but included larger geographic locations and larger total population sizes. There was ethnic heterogeneity between clusters, with very high percentages of White children in some clusters (92% in the North East Lincolnshire cluster) and very low percentages in others (22% in the Newham cluster). Clusters showed relatively high levels of deprivation, with average IMD decile for children in short stature clusters ranging between 2.40 and 4.38 (for comparison, the average IMD for the full sample was 5.08). An unadjusted model as well as a model adjusted for both ethnicity and IMD are presented in S5 Table and S1 and S2 Figs. The general distribution of clusters was similar in the unadjusted and fully adjusted models, although the composition of clusters varied slightly. The Leicester, Great Yarmouth, Tower Hamlets, Gateshead, Rossendale, North East Lincolnshire, and South Staffordshire clusters were represented in each model.

Fig 3

Short stature clusters in England (London inset) 2006–2019, adjusted for ethnicity (n = 5,765,707).

Short stature clusters are in red. Map base layer is shapefile Local Authority Districts (December 2017) Full Clipped Boundaries in Great Britain, published by the Office for National Statistics and available at https://geoportal.statistics.gov.uk/datasets/local-authority-districts-december-2017-full-clipped-boundaries-in-great-britain/explore?location=55.450000%2C-2.950000%2C5.64&showTable=true.

Table 2

Short stature clusters in England, 2006–2007 to 2018–2019, adjusted for ethnicity (n = 5,765,707).

Clustera	Region	White ethnicityb % (n)	Cluster mean (SD) IMDb	Populationc	Short stature % (n)	RRd
Leicester	East Midlands	41 (20,737)	2.9 (2.0)	50,088	3.1 (1,551)	1.50
Great Yarmouth, Norwich	East of England	90 (25,098)	3.7 (2.5)	27,793	2.8 (783)	1.36
Brent	London	25 (35,997)	3.3 (1.6)	35,997	2.8 (1,002)	1.34
Tower Hamlets, Newham, Hackney	London	22 (26,020)	2.1 (1.1)	116,778	2.6 (3,083)	1.28
Gateshead, Newcastle upon Tyne	North East	81 (32,907)	3.9 (2.9)	40,757	2.6 (1,062)	1.26
Rossendale, Burnley, Bury, Rochdale, Hyndburn, Blackburn with Darwen, Bolton, Oldham, Calderdale, Pendle, Chorley, Manchester, Salford, Tameside, Wigan, Ribble Valley, Trafford, Bradford, South Ribble, Kirklees, Preston	North West, Yorkshire, and the Humber	68 (348,059)	3.7 (2.8)	515,310	2.5 (13,087)	1.25
North East Lincolnshire, North Lincolnshire, West Lindsey, Kingston upon Hull, East Lindsey, Lincoln	Yorkshire and the Humber, East Midlands	92 (95,781)	3.9 (2.7)	103,569	2.5 (2,612)	1.22
South Staffordshire, Wolverhampton, Cannock Chase, Walsall, Stafford, Sandwell, Telford and Wrekin, Dudley, Lichfield, East Staffordshire, Birmingham, Tamworth, Wyre Forest, Stoke-on-Trent	West Midlands	62 (272,336)	3.4 (2.7)	439,577	2.4 (10,618)	1.18

a Clusters are referred to in the text by the name of the first LA in the cluster description. These are determined by SaTScan and represent the centre point of the cluster. Clusters are ordered from highest to lowest RR.

b Cluster white ethnicity % and mean IMD are derived from NCMP data for children in each cluster.

c Cluster population is the total population of NCMP children included in the analysis for each cluster.

d No 95% CI is calculated for RR as the method for identifying clusters is data driven, and 95% CIs would be inappropriate.

CI, confidence interval; IMD, index of multiple deprivation; LA, local authority; NCMP, National Child Measurement Programme; RR, relative risk; SD, standard deviation.

Spatial analyses over the 4 time periods between 2006 and 2019 showed some variation over time in the precise composition of clusters, although the placement of clusters was largely consistent (Fig 4, S6 Table).

Fig 4

Spatial analysis over 4 time periods (2006–2019).

Short stature clusters are in red. Map base layer is shapefile Local Authority Districts (December 2017) Full Clipped Boundaries in Great Britain, published by the Office for National Statistics and available at https://geoportal.statistics.gov.uk/datasets/local-authority-districts-december-2017-full-clipped-boundaries-in-great-britain/explore?location=55.450000%2C-2.950000%2C5.64&showTable=true.

Very short stature (<−2.67 SDS) affected 0.36% [0.36 to 0.37] of children, with patterns of prevalence similar to those of short stature (<−2.0 SDS) (S7 Table). Spatial analyses of very short stature adjusted for ethnicity identified 6 clusters that broadly matched 6 of the 8 clusters found in the main short stature analysis (S8 Table). A sensitivity analysis of short stature (<−2.0 SDS) using the full dataset (n = 7,062,071) (S9 Table) showed high agreement with the main unadjusted model (n = 5,765,707) (S5 Table).

Discussion

Using a dataset of over 7 million children in England between 2006 and 2019, we found that 1.93% of children aged 4 and 5 had short stature at school entry. Short stature was geographically clustered with higher prevalence in the North and Midlands and lower prevalence in the South. Short stature prevalence ranged from 0.97% in Richmond upon Thames in London to 3.92% in Blackburn and Darwen in the North West, a 4-fold difference that translates into an additional 2,950 children with short stature per 100,000 children starting school. Short stature was highly associated with area-level deprivation, ethnicity, and sex. Deprivation was linearly related to short stature, with the highest prevalence found in the areas with highest deprivation. Yorkshire and the Humber had the highest regional short stature prevalence (2.18%), and London had the lowest (1.57%). However, we identified 2 high prevalence clusters in East and North London, suggesting high heterogeneity within London itself. While there were significant differences in short stature prevalence between children of different ethnicities, there was also high ethnic heterogeneity between clusters, with the proportion of White children ranging from 22% in the Tower Hamlets cluster to 90% in the Great Yarmouth cluster. Girls were significantly more likely to be short than boys (2.09% versus 1.77%, respectively), which has not previously been reported in high-income settings.

While the national short stature prevalence of 1.93% was broadly as expected, given the normal distribution of heights in a population, the regional differences are striking. Although short stature prevalence declined over time, with larger declines in more deprived IMD deciles, inequalities in the prevalence of short stature by IMD were persistent. This clear association between short stature and deprivation corresponds to a difference in absolute terms of 1,180 per 100,000 reception class children between the most and the least deprived communities (highest and lowest IMD deciles). Collectively, our findings suggest that large numbers of children—particularly those in the most deprived areas of the country—could be failing to reach their full growth potential. Many of these children may in fact be particularly disadvantaged, unhealthy, and failing to thrive.

The link between growth and social conditions is well documented, with growth deemed to be “a mirror of the condition of society” [17]. However, in England, the focus continues to be on overweight and obesity. To our knowledge, this is the first study presenting data on geographical clustering of short stature in the UK. The World Bank reports an average stunting (short stature) prevalence in high-income countries of 2.8%, but there are very few recent data for international comparison, and the database includes no data from the UK [18]. We found evidence of a clear association between short stature and area-level deprivation. The association holds even for very short stature (height SDS <−2.67), whereby prevalence falls monotonically from 0.52 to 0.23 across the IMD deciles (S7 Table). An association between height and deprivation in children of this age has previously been demonstrated [19], but the relationship between deprivation and short stature has not previously been investigated in the UK. There were 2,560 children with short stature per 100,000 in the bottom decile and 1,380 per 100,000 in the top decile (a scale factor of 1.86; Table 1) and clusters identified by spatial analyses all had an IMD below the sample average in unadjusted and ethnicity-adjusted models. Recently published data on child poverty indicators reveal an overlap between areas with high levels of child poverty and the clusters of short stature we identified [20].

Our data demonstrate a striking North–South divide in short stature prevalence in England (Fig 3). A total of 6 of the 8 clusters identified in the ethnicity-adjusted analysis were in the North and Midlands, and there were no clusters outside of London in the South. Spatial–temporal analyses were more complex, but they validated the main results by finding similar patterns of distribution of short stature, where clusters were mainly concentrated in the North and Midlands of the country. This is consistent with national samples of children born between 1920 and 1990 in the UK, in which children born in the North were shorter than children born in the South [21]. A 2015 analysis of around 10,000 White children participating in the Millennium Cohort Study described a “Midlands effect,” where children in the Midlands were more likely to be shorter than −1 SDS, compared with children in other regions. The clusters also correspond with the findings of the Marmot Review 10 Years On, which documented regional inequalities in health that have disproportionately affected the North and Midlands. The 2020 Marmot Review also described increased child poverty, greater infant mortality in the most deprived decile, and persisting socioeconomic and regional inequalities in child development and school readiness [22]. While the NCMP data clearly show a relationship between deprivation and short stature, the proximal mechanisms for poor linear growth in high-income settings are not well established. These may include greater infection burden, ambient pollution, poor diet quality, and vitamin D deficiency. Additional mechanisms including adverse childhood experiences, pregnancy outcomes, and epigenetics should be explored in further research.

We found that Indian, Pakistani, and Bangladeshi children had the highest prevalence of short stature, and Black and Mixed children had very low prevalence of short stature. The clusters we identified have high ethnic heterogeneity, with the proportion of White children in high prevalence clusters ranging from over 90% in the North East Lincolnshire cluster to around a quarter in the London clusters. The relationship between ethnicity and linear growth in childhood is complex. The 2006 WHO Multicentre Growth Reference Study indicated that globally, the growth of economically advantaged, breastfed infants, and children of non-smoking mothers is similar [23]. The INTERGROWTH-21st study demonstrated similar findings for fetal and neonatal growth [24]. However, absolute heights and growth patterns between populations are starkly different. The relationship between short stature and ethnicity in the UK is likely to be shaped by complex pathways, including socioeconomic status, immigration patterns, discrimination, and genetic factors.

We found a statistically significantly higher prevalence of short stature in girls compared with boys. Population data on sex differences in short stature are useful, as existing data come from analyses of referral patterns to growth disorder clinics and are therefore subject to selection bias. Evidence from the United States of America suggests that boys are referred more readily and investigated more thoroughly than girls [25,26]. Short stature is more common in boys than girls in LMICs, and it has been postulated that this is due to higher rates of adverse birth outcomes and increased vulnerability to infection and other morbidity during infancy in boys [27,28]. These data might suggest that this is not a reasonable explanation for sex differences in referral patterns in the UK setting, as the incidence of adverse birth outcomes and childhood infections are higher for boys in the UK [29,30]. It should be noted that the use of UK90 growth references means that rates of short stature in girls and boys will be slightly different to those calculated using WHO standards, which are used in most of the international literature. This may explain some of the discrepancy observed between our results and international findings.

We found a decline in national short stature prevalence between 2006 and 2019. Analyses of British birth cohorts between 1946 and 2001 have previously demonstrated a narrowing of socioeconomic inequalities in child height [31-33]. However, in the context of increasing socioeconomic inequality in England over the last decade [22], the reasons for such a striking decline in prevalence during this period are not clear, especially since the steepest fall was in the most deprived decile (Fig 2). One potential explanation for these findings is that children from the most deprived areas may be benefiting from targeted interventions. This would also explain findings from the 2020 Marmot Review, which found that children from low-income families have demonstrated better development and educational outcomes in low-income areas than children from low-income families in high-income areas [22]. Another hypothesis relates to evidence from diverse populations that childhood obesity is associated with taller stature in childhood (although a shorter adult height). As such, higher levels of overweight and obesity in childhood in deprived children over the period 2006 to 2019 could have driven accelerated linear growth in many children who otherwise might have had short stature. This is supported by published NCMP data, which show static or increased prevalence of overweight/obesity in 5-year-old children in the most deprived IMD quintile, while there is a downward trend in the least deprived quintile [34].

This study had several strengths including the very large national dataset, with coverage of 93% of reception-age children in England over a 13-year period, leading to high precision in our estimates of short stature. Additionally, where previous analyses of associations between height and deprivation in the UK have used data from historical birth cohorts, this analysis of contemporary data allows for more relevant policy inferences for 21st century children. The study also had several limitations. To investigate associations between short stature and deprivation, we used available data on IMD, which is not a measure of individual or household deprivation nor does it completely capture socioeconomic or environmental variables. As such, residual confounding is very likely. The population reference used (UK WHO) is constructed using data from the UK 1990 for the age group analysed, which was, in turn, developed using data from White British children. This may limit its generalisability in children of other ethnicities. Additionally, the NCMP data are cross-sectional, and there is only a single data point for each child, making causal inferences inappropriate.

There are 3 major public health implications of our findings. Firstly, the geographical “hotspots” and the clear socioeconomic gradient demonstrated by these data show that where a child is born, and the environment in which they grow up, both are associated with their height at the age of 4 to 5 years. This should prompt immediate action by local and national government to address the upstream factors that underlie short stature, especially in the areas where we have identified clusters. Moreover, studies of child growth trajectories in LMIC have identified substantial catchup growth by 5 years of age among children who had short stature in infancy [35]. It is therefore possible that the prevalence of short stature is even higher at younger ages in children experiencing poverty in England.

Secondly, the concept of “stunting,” or short stature, which broadly reflects socioeconomic determinants of growth faltering, is typically used to describe health inequalities in LMIC but rarely used in high-income settings. In the UK, stunting or short stature is not considered to be a major public health problem because the overall prevalence is much lower than in LMIC. We contest that this difference in approach leads to children with short stature being overlooked in the UK. Our analyses of a large national dataset over the past 13 years, however, identify a clear social gradient and striking regional variation in short stature that should not be ignored.

Finally, while the weights and heights of most 4- to 5-year-old children in England are being systematically measured in the school setting, children with poor linear growth are not being highlighted to their families or general practitioners. This misses a valuable opportunity to identify children whose poor early life growth may be associated with poor health and delayed neurodevelopment in childhood and chronic disease and all-cause mortality in adulthood [36]. UK guidance recommends referral for children with height below the 0.4th centile (<−2.67 SDS). This guideline is stricter than other European countries and has low sensitivity for detecting growth disorders (around 30%) [37]. The combination of height SDS, parental heights, and decreased growth rate can be used effectively for growth monitoring with optimal cutoff levels. However, this is currently not possible in the UK as parental heights are not routinely assessed, and repeat measurements are not undertaken. Around 60% to 80% of short children (<−2.0 SDS) are estimated to have no identifiable aetiology following review [38]. We argue that screening should start earlier in life than the current NCMP programme, since school-age measurements may be too late to mitigate many of the factors underlying short stature. It is now well established that the “first 1,000 days” (conception to age 2 years) is the period during which linear growth and neurodevelopment are most sensitive to environmental modification. We therefore propose that height should be systematically screened at younger ages in the UK, in line with other European countries. For example, children in the Netherlands and Finland have heights and weights routinely measured at least 10 times in the first 1,000 days [39]. The National Screening Committee and the Health and Social Care Committee have both recommended growth screening in early childhood, yet no such programme currently exists [40, 41]. Earlier, systematic, nationwide screening and identification of linear growth faltering could trigger timely referral for investigations to identify those with underlying medical disorders and an opportunity for psychosocial and educational intervention prior to school entry in those without underlying medical problems.

STROBE Statement.

Checklist of items that should be included in reports of observational studies. STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

Proposal.

Barts Charity grant proposal research analysis plan (2018).

(DOCX)

Click here for additional data file.

SaTScan model specification.

Table A: Height (cm) by ethnicity regression results, boys (n = 2,946,560). Table B: Height (cm) by ethnicity regression results, girls (n = 2,819,147). Table C: Height (cm) by ethnicity and IMD regression results, boys (n = 2,946,560). Table D: Height (cm) by ethnicity and IMD regression results, girls (n = 2,819,147). IMD, index of multiple deprivation.

(DOCX)

Click here for additional data file.

Ethnicity and IMD missingness by sample characteristics.

IMD, index of multiple deprivation.

(DOCX)

Click here for additional data file.

Population sex, Government Office Region, and ethnicity in children, 2011–2012 (Census 2011 and ONS population projections).

(DOCX)

Click here for additional data file.

Percentage of children with short stature (<−2.00 SDS) by NCMP school year and IMD decile (n = 7,061,591).

(DOCX)

Click here for additional data file.

Logistic regression of short stature (<−2.00 SDS) by IMD and year, including IMD#year interaction (n = 7,061,591).

IMD, index of multiple deprivation; SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Short stature (<−2.00 SDS) clusters, unadjusted and adjusted for both ethnicity and IMD (n = 5,765,707).

IMD, index of multiple deprivation; SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Time period analysis of short stature (<−2.00 SDS) (n = 7,062,071). SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Very short stature (<−2.67 SDS) prevalence by sample characteristics (n = 7,062,071). SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Very short stature (<−2.67 SDS) clusters in England, 2006–2007 to 2018–2019, unadjusted, adjusted for ethnicity, and adjusted for ethnicity and IMD (n = 5,765,707).

IMD, index of multiple deprivation; SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Short stature (<−2.00 SDS) clusters, unadjusted, full sample (n = 7,062,071). SDS, standard deviation score.

(DOCX)

Click here for additional data file.

Short stature clusters in England (London inset) 2006–2007 to 2018–2019, unadjusted.

(DOCX)

Click here for additional data file.

Short stature clusters in England (London inset) 2006–2007 to 2018–2019, adjusted for ethnicity and IMD. IMD, index of multiple deprivation.

(DOCX)

Click here for additional data file.

6 Apr 2021

Dear Dr Orr,

Thank you for submitting your manuscript entitled "A spatial analysis of child stunting in England between 2006-2019" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by April 8, 2021.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Beryne Odeny

Associate Editor

PLOS Medicine

10 Jun 2021

Dear Dr. Orr,

Thank you very much for submitting your manuscript "A spatial analysis of child stunting in England between 2006-2019" (PMEDICINE-D-21-01543R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Jul 01 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Beryne Odeny,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1) Please revise your title according to PLOS Medicine's style. Please place the study design (e.g., "A retrospective cohort study”) in the subtitle (ie, after a colon).

2) In the abstract Methods and Findings:

a) Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions).

b) Please combine the Methods and results sections into one section, “Methods and findings”.

c) Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text.

d) Please quantify the main results (with 95% CIs and p values).

e) Please include the important dependent variables that are adjusted for in the analyses.

f) In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

g) Please address the study implications without overreaching what can be concluded from the data; the phrase "In this study, we observed ..." may be useful. Mention only specific implications substantiated by the results.

3) Author summary - At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary.

4) Did your study have a prospective protocol or analysis plan? Please state this early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

5) Your study is observational and therefore causality cannot be inferred. Please remove language that implies causality, such as impact. Refer to associations instead. For example, in the discussion section, replace the statement, “… both have a clear impact on height at the age of 4-5” with “… are associated with height at the age…”

6) Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)." The STROBE guideline can be found here: http://www.equator-network.org/reporting-guidelines/strobe/ . When completing the checklist, please use section and paragraph numbers, rather than page numbers.

7) In the Methods and Results section:

a) Please provide 95% CIs and p values for all estimates.

b) When a p value is given, please specify the statistical test used to determine it.

c) Please present numerators and denominators for percentages, at least in the Tables

8) For your Tables and figures, please define the abbreviations such as SD, IMD, NCMP, SE, CI

9) Please use the "Vancouver" style for reference formatting and see our website for other reference guidelines. For example, six names should appear before et al. Please ensure that journal name abbreviations match those found in the National Center for Biotechnology Information (NCBI) databases, and are appropriately formatted and capitalized. https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references. Please ensure that weblinks are current and accessible to date.

a) Six names should appear before et al, in the reference section. Please correct refs # 20, 29, 31

b) Please include access dates for all weblinks and ensure that all weblinks are current and accessible.

10) Please include line numbers in you next revision

Comments from the reviewers:

Reviewer #1:

This is a very important study and the first to my knowledge that height of children from the NCMP program has been analysed separately from obesity alone and that has powerful socio-economic implications.

The study is well conducted appropriately statistically analysed and very well presented including the extra supplemental material, although the statistical process is one of spatial analysis, 'regional analysis' might be a more accessible term for to use for the title.

I am uncomfortable with the use of the term 'stunting' here. Although the WHO does use this for height of children below -2 standard deviations it also includes in the definition of not achieving their full adult high potential. That is very difficult to infer from this study given the information presented here, hence in my view the term 'short stature' is a more preferable term to stunting as there is no evidence of a long term growth decrement and a reduction of adult height compared with genetic or familial expectations. The simple explanation could be different growth patterns. We know already that Afro-Caribbean childen are more physiologically advanced than Caucasians.

I have no additional comments to make about the ethical or analytical processes. These appear to be well done and I am not an expert in spatial analysis.

Data of the proportions of the sample each of the region are presented in table 1 together with the ethnicity breakdown. It would be helpful for comparative purposes if the figures from the 2011 population survey were included to validate these values.

The changes in the IMD quintile over time as presented in figure 2 is very important information and perhaps reasons that this has occurred could be explored in more detail in the discussion section other than putting this down to obesity. Presumably with the NCMP data you can affirm or deny that conclusion.

Height determinations are made according to the UK 1990 British references. Reference 11 is correct but WHO data are used until age 4 then UK 1990 thereafter. These are built exclusively on the white British population therefore further adjustment needs to be made to represent all the ethnic minorities present. It is little surprising that those of Asian background are significantly shorter and those of Afro-Caribbean taller. That discussion needs to be expanded. The presentation of the proportion below the second centile by ethnic background is important. However, there are no references to known heights of adults from relative ethnic backgrounds presented and perhaps a further sub-analysis based on this information might be helpful for further interpreting the findings. Reference to the country of origin might need to be made.

Furthermore, the data used to produce the 1990 charts was collected from a number of sources and at that stage there already was a north south divide evident in heights of children within the UK. The 1995 Freeman and Cole paper in Archives of Disease in Childhood 1995; 73: 17-24 discusses differences in heights across the country with Newcastle infants being lighter and Leeds school children being shorter than those in the south. The male-female disparity was different however.

On page 17 there is a comment that those without an identifiable cause of stunting would be labelled as idiopathic short stature. However, a simple height measurement below the second centile cannot really define either idiopathic short stature or stunting. The value of height screening has been much debated but there is significant evidence in particular for Dutch and Finnish studies as discussed in the paper that a stepwise approach can be taken to identify short children with significant pathology but often requires repeat measurements and knowledge of the parents' heights. Hence, if an argument was constructed to use height measurement to detect socioeconomic deprivation and to detect pathology more often than the NCMP programme could allow, then the discussion needs to reflect that. I don't think that the final line of the abstract conclusion, 'Many children in the most deprived areas of the country may be failing to reach their full growth potential' can be fully justified based on the information presented here without additional expansion. Overall, though this is an important paper and needs to be presented in the public domain.

Reviewer #2: The authors describe a spatial analysis of stunting in English children aged 5, based on the NCMP. The paper is excellent, with an extremely large dataset imaginatively analysed and carefully interpreted. I have some mainly minor comments on the study analysis, presentation and interpretation.

1. I should declare my conflict of interest, having constructed (with Mike Preece and Jenny Freeman) the British 1990 height reference, which at age 5 equates to the UK-WHO reference. The stunting rate should be close to 2.28%, the proportion below -2 SDS, so the paper provides a large-scale validation of our chart construction skills.

2. I am pleased to see that the mean stunting rate of 1.93% is close to the expected value, though the apparent sex difference in rates probably reflects a bias in the reference - see later.

3. Abstract. Note that the UK-WHO is a growth reference, not a growth standard.

4. When describing the national state school coverage, it would be useful to say what percentage of children aged 5 are in state schools.

5. What is a "suppressed" version of the NCMP dataset?

6. Why was the cut-off of ±5 SDS used for data cleaning, as opposed to say ±4?

7. The definition of severe stunting used in the paper is wrong (page 5). The cut-off is -2.67 not -2.65 SDS, following the convention proposed by me (Cole TJ. Do growth chart centiles need a face lift? BMJ 1994;308:641-642). Thus the nominal severe stunting rate is 0.38% rather than 0.40%. The whole sentence needs redrafting, as it is not true to say it is "often" used as a referral cut-off, it is _the_ cut-off for defining short stature in the UK. See for example https://www.rcpch.ac.uk/sites/default/files/Boys_2-18_years_growth_chart.pdf. This relates to comments later about idiopathic short stature - they should relate to the -2.67 not the -2 cut-off.

8. The Statistical Analyses section does not include a sentence specifying the level of significance being used. Despite this p-values appear throughout the text, mainly p < 0.001. I suggest omitting most if not all of the p-values, as the sample size is enormous and the standard errors tiny, ensuring significance even for very small differences. The confidence intervals provide the same information more effectively.

9. In Figure 2 it would be useful to reverse the IMD legend so it's in the same direction as the graphic. What does the title "Equiplot (13)" mean?

10. The cluster results in Figure 3 are very interesting, particularly the position of Leicester, which tops the list but appears as a tiny blob in the figure. I suggest labelling the clusters, primarily for non-British readers. It would also be worth discussing in a bit more detail exactly why Leicester should achieve this dubious distinction. Clearly it has a high proportion of Asian children, but ethnicity is adjusted for so it ought not to be the explanation. Has Leicester featured prominently in other indicators of deprivation?

11. The title for Table 2 looks odd, saying it's adjusted both for ethnicity and for ethnicity and IMD. I suspect just the former is true. The table gives the stunting rate and RR to two decimal places, which is appropriate, but giving the white ethnicity percentage to two decimal places is not. It aims to distinguish between clusters like Brent, where 25% of children are white, and Great Yarmouth, where 90% are white. These percentages can given as whole numbers, and greater precision is unnecessary. This should apply throughout the text as well as the table. See my guidelines on the presentation of numerical information at http://adc.bmj.com/cgi/content/full/archdischild-2014-307149. Also in Table 2 I suggest omitting the log-likelihood ratio and p-value columns, as they are respectively uninterpretable and uninformative. Note too that log-likelihood ratios to two decimal places are grossly over-precise. Table 1 would also be better with fewer decimal places.

12. Coming back to the sex difference, the stunting rates by sex are 2.09% for girls and 1.77% for boys. The nominal rate is 2.28%, so it's not so much that girls are high as that boys are low. But in fact the differences are small - the corresponding z-scores are -2.10 and -2.04, which equate to 2-5 mm deviation from the expected cut-offs. The Discussion states that this sex difference has not previously been reported in a high income setting, but it is almost certainly not generalisable. It is much more likely a bias in the reference as originally constructed. It may be worth mentioning the BMJ letter that we published soon after publishing the reference: Preece MA, Freeman JV, Cole TJ. Sex differences in weight in infancy - Published centile charts for weight have been updated. BMJ 1996;313:1486. It referred to infancy not age 5, but it indicates the care needed to avoid building a sex bias into the reference.

13. The second paragraph of the Discussion refers to idiopathic short stature, but as mentioned above it needs to be made clear that this diagnosis formally requires evidence of severe stunting, i.e. height SDS < -2.67. The same idea is picked up at the foot of page 17, where it is suggested that stunting is not recognised as an indicator of deprivation.

14. The scale of the dataset leads to some very powerful trends. The first paragraph of the Discussion states—correctly—that "Deprivation was linearly related to stunting", yet this statement holds even when applied to severe stunting, where in supplemental table C the severe stunting rate by IMD falls, perfectly monotonically, from 0.55 to 0.24 across the ten categories. I find this astonishing, and it calls to mind the famous phrase of James Tanner: "Growth as a mirror of the condition of society" (published as Tanner JM. Growth as a mirror of the condition of society: secular trends and class distinctions. Acta Paediatr Jpn 1987;29:96-103). This paper would be worth citing.

15. The reference to the dual burden and the suggestion that increasing obesity is fuelling a reduction in stunting in the more deprived is well made.

16. The paper lists its limitations, but I disagree that the last two mentioned are valid limitations. "As such, we will miss children with faltering growth whose height is crossing centiles and children whose height is above -2 SDS but who are below their target height as predicted by parental heights." Cross-sectional data cannot be expected to provide evidence of centile crossing, and the chosen definition of stunting does not involve parental height. I recommend removing the sentence.

17. A few minor comments on the Supplemental materials. The regression results should specify the units as cm. I suggest omitting the SE columns as they are so tiny (and the SE for the constant in Table A3 needs some attention). I found the table titles odd, a mix of letters and numbers with no obvious structure.

Tim Cole

Reviewer #3: This is a well-conducted spatial analysis of child stunting in England between 2006-2019. The study design, dataset, statistical methods and analyses, and presentation (tables and figures) and interpretation of the results are mostly adequate and of a good standard. The stats analyses in the supplementary information are very thorough and comprehensive with missing data issue particularly well addressed. Only a couple of minor points needing attention.

1) Table 2. Not clear what exactly was adjusted for. In the title, it says 'adjusted for ethnicity and adjusted for ethnicity and IMD' but in the table, there is only a line saying 'Adjusted for ethnicity'. Can authors please clarify this?

2) Time trends of stunting hot spots are striking as shown in Supplementary Figure F. Maybe authors can move Figure F into the main paper as it's of public health interest and give it a bit more discussion on the changes over time and the reasons behind these changes.

Any attachments provided with reviews can be seen via the following link:

[LINK]

5 Jul 2021

Submitted filename: 210701 Responses to reviewers.docx

Click here for additional data file.

26 Jul 2021

Dear Dr. Orr,

Thank you very much for re-submitting your manuscript "Regional differences in short stature in England between 2006-2019: A cross-sectional analysis from the National Child Measurement Programme" (PMEDICINE-D-21-01543R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by three reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Aug 02 2021 11:59PM.

Sincerely,

Beryne Odeny,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1) Thank you for providing your STROBE checklist. Please replace the page numbers with paragraph numbers per section (e.g. "Methods, paragraph 1"), since the page numbers of the final published paper may be different from the page numbers in the current manuscript.

Comments from Reviewers:

Reviewer #1: This is now a fantastic paper. Congratulations to the authors and for all the revisions. I recommend publication.

Only 2 minor changes suggested:

Abstract P2 line 11 : should be ....heights from a total of 7xxxxxx children were analysed.

Discussion p16 line 3:A 2015 analysis ...... children in the midlands were most likely to be to be shorter than SDS-1. I'm not completely clear what this means

Good luck!

Reviewer #2: The authors have responded very positively to my suggestions. I have a few more minor comments for them to consider.

1. The Discussion refers to the stunting rate in an international context, but it should be remembered that the British 1990 reference is not widely used internationally. This needs to be emphasised, and it would also be worth giving the rate as based on the WHO standard/reference. For girls the -2 SDS cut-off at 60 months is exactly the same (99.91 cm) for UK90 and WHO, but for boys it is 100.7 cm for WHO as against 100.5 for UK90. With a higher boys WHO cut-off there ought to be more WHO stunting, and I estimate the boys rate with WHO as 1.93% as against 1.77% with UK90. You can probably calculate it directly.

2. This comparison between UK90 and WHO is also useful for interpreting the sex difference, in that the UK90 and WHO cut-offs are broadly the same so that perhaps the girls' excess is more genuine that I previously thought. Note that reference 25 (mentioned on page 16) refers explicitly to weight in infancy, and hence it is not directly relevant to height at age 5 - I referred to it just to highlight how tricky it can be to estimate growth centiles.

3. I note that the acronym SDS is used with the -2.67 cut-off but not with -2.0. I suggest harmonising its usage.

4. In Table 2 just one decimal place would be better for the cluster means.

5. The phrase 'short-statured children' appears on page 3, which might be better as 'children with short stature'. The later bullet point starting 'Short stature' should either omit 'also' or else reverse the sentence.

6. The maps of the clusters look very good, but I wondered if they could use colour more effectively, e.g. by adding a colour scale to reflect the % of white children in each cluster - just a thought, but it would emphasise how heterogeneous the ethnicity mix across the clusters.

Tim Cole

Reviewer #3: Many thanks authors for their effort to improve the manuscript. I am satisfied with the response and revision. No further issues needing attention.

Any attachments provided with reviews can be seen via the following link:

[LINK]

2 Aug 2021

Submitted filename: 210728 Responses to reviewers RR2.docx

Click here for additional data file.

5 Aug 2021

Dear Dr Orr,

On behalf of my colleagues and the Academic Editor, Dr. Zulfiqar A. Bhutta, I am pleased to inform you that we have agreed to publish your manuscript "Regional differences in short stature in England between 2006-2019: A cross-sectional analysis from the National Child Measurement Programme" (PMEDICINE-D-21-01543R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Beryne Odeny

Associate Editor

PLOS Medicine