This is a population-level, retrospective, observational study utilising routinely collected linked data to examine the relationship between the weight status of children aged 4-5 years and the risk of having a recorded diagnosis of asthma during the nine years following initial BMI measurement, the maximum period for which data were available.

All data used in this study were accessed via the SAIL databank, hosted at Swansea University. The SAIL databank provides a Trusted Research Environment enabling access to a number of datasets representing routinely collected administrative and health data pertaining to the population of Wales in the United Kingdom. All data within SAIL have been anonymised by removing personally identifiable information and assigning each individual a unique identifier referred to as anonymised linkage field (ALF) which can be used to link records pertaining to the same individual across different datasets and sources. Access and utilisation of data within the SAIL databank complies with the Data Protection Act of 2018 and with the General Data Protection Regulation of 2016. [12–14]

Several data sources were linked and used in this study:

The National Community Child Health Database (NCCHD) compiles data from Child Health System databases which are held by National Health Service (NHS) organisations and includes data on birth registrations, child health examinations and vaccinations. This dataset includes the anthropometric measures of children taken as part of the CMP programme described above.

The Welsh Longitudinal General Practice (WLGP) dataset includes data about patient encounters from 80% of GP practices in Wales, covering approximately 83% of the Welsh population [15]. Where diagnosis codes are available for an encounter, they are encoded using the Read coding system [16].

The Welsh Demographic Service Dataset (WDSD) contains demographic and address information for people accessing NHS services in Wales and registered to a Welsh address. WDSD was used to derive age, sex, and socio-economic deprivation quintile for the cohort.

The Outpatient Database for Wales (OPDW) contains data on all scheduled outpatient appointments in all Welsh hospitals. Diagnoses are coded using the International Classification of Disease (ICD), 10^th iteration (ICD-10) [17].

Patient Episode Dataset for Wales (PEDW) includes attendance and clinical information pertaining to all admissions to Welsh hospitals, both inpatient and day cases. Diagnoses are coded using the ICD-10 system.

The Emergency Department Dataset (EDDS) consists of administrative and clinical data for all NHS Wales Accident and Emergency department attendances. Diagnoses are coded using a dataset-specific system [18].

The Annual District Death Extract (ADDE) is a register of all deaths relating to Welsh residents, including those that died out of Wales.

Figure 1 provides a summary and outline of the steps taken to derive the study cohort. Out of 31,506 children for whom anthropometric measures were available as part of the 2012/13 CMP programme, 3,250 were excluded as their week of birth (WOB) was before 01 September 2007 or after 31 August 2008. This large number can probably be explained by the fact that in the academic year 2012/13, one Welsh county measured some children during year one rather than reception year. This fact was mentioned in that year’s programme report [11]. The report explained that these measurements were excluded from the analysis but they may however have been included in the data available in the SAIL databank. A further 31 children were excluded as the date of measurement fell outside the academic year, that is before 01 September 2012 or after 31 August 2013. Children who were underweight at the time of measurement (BMI <second centile) were also excluded since some previous studies suggest an independent relationship between underweight and wheezing disorders [25].

The previous steps resulted in 28,051 children for inclusion. To be included in the final study cohort, individuals needed to have nine years of follow-up data available. For the purposes of this study, this was interpreted as the individuals being alive and registered to a Welsh address for at least nine years following the date of anthropometric measurement. Nine individuals had died and 1,996 individuals did not continue to live in Wales for at least 9 years following the date of measurement. Longitudinal datasets were interrogated for all records pertaining to the remaining 26,046 individuals mentioning a diagnosis code denoting asthma, and those who had a mention of an asthma diagnosis before 01 September 2012 were excluded (i.e., before the anthropometric measurement), as was done in similar studies [10].

The final study cohort therefore included 22,790 individuals for whom anthropometric measurements as part of the CMP programme during the academic year 2012/13, whose age at measurement was at least 4 years and no more than 6 years, who continued to reside in Wales for at least nine years following the measurement date, and who did not have a recorded diagnosis of asthma before the start of the 2012/13 academic year on 01 September 2012.

All analyses were performed using R Statistical Software 4.2.1 [26]. Logistic regression was used to evaluate the association between weight status at time of measurement and the likelihood of having a recorded diagnosis of asthma at any time during the nine years following the date of measurement, while adjusting for the covariates of sex and WIMD quintile. Model coefficients were exponentiated, and the unadjusted and adjusted odds ratios (aOR) and 95% confidence intervals (95% CI) are reported.

This study assessed whether 4- to 5-year-old children with obesity as measured by BMI who had anthropometric measures taken as part of a national surveillance program in Wales were more likely to have a recorded diagnosis of asthma in several primary and secondary care linked longitudinal clinical datasets during the nine years following measurement. Seven percent of the 22,790 children who were included in this study had a recorded diagnosis of asthma in the nine years following the measurement date, 57.4% of whom were male. Of those who had a recorded diagnosis of asthma, 8.2% were classified as obese (BMI Z-score ≥ 98th Centile) as opposed to 6.1% of those who did not. Children with obesity at the time of measurement were 41% more likely to have a recorded diagnosis of asthma during follow-up, after adjusting for sex and deprivation index quintile. Females were 26% less likely to have a recorded diagnosis of asthma compared to males, after adjusting for weight status and deprivation index quintile.

Many prior studies explored the relationship between childhood obesity and the risk of bronchial asthma using different study designs [27]. Additionally, variations among studies in terms of participant age groups, exposure and outcome definitions, and study population size added to the challenges of comparing and/or aggregating study results [25]. Very few population-level studies examined the association between obesity and asthma and those we are aware of had a relatively short follow-up period [28]. To our knowledge, this is the first population-level study to examine the relationship between children’s weight status at the relatively early age of 4-5 years and the risk of being diagnosed with bronchial asthma during a 9-year follow-up period.

Egan et al. [10] performed a meta-analysis of six prospective studies looking at the relationship between childhood BMI and subsequent physician-diagnosed asthma. Their analysis found that children who were classified as obese had increased relative risk of having a subsequent diagnosis of asthma (combined risk ratio [RR] =1.50; 95% CI: 1.22–1.83). Although in their study, Egan et al. defined obesity as BMI >95^th centile, which is slightly different from our definition, their results were similar to those of our study.

Another meta-analysis by Mebrahtu et al. [25] used data standardisation across studies in terms of the exposure variable (BMI) and outcome variable (asthma or wheeze) and so avoided some of the shortcomings of other similar studies. The included studies covered over one million children and the analysis revealed a significant increase in the risk of wheezing disorders in children with obesity (OR = 1.46; 95% CI: 1.36-1.57), again very similar to our study.

A population-based longitudinal study by Black et al. [28] included over 600,000 participants, although the median follow-up time was 3 years, quite shorter than that in our study. As in our study, their results showed that children with extremely obese status had increased adjusted risk of developing asthma (aRR: 1.37; 95% CI: 1.32-1.42).

To our knowledge, this is the first study that linked population-level anthropometric data collected through the CMP in Wales with multiple primary and secondary care longitudinal datasets providing near-complete coverage of the Welsh population to study the association between obesity in children and the risk of developing asthma later in childhood. Measurements in the CMP programme are taken by trained personnel following specific standards and guidelines [11]. This fact, coupled with this being a nation-wide programme comprising a large number of participants, makes this anthropometric dataset especially valuable. The SAIL databank allowed linking the CMP dataset with several longitudinal clinical and administrative datasets providing almost-complete coverage of the Welsh population. Also, participants were followed up for nine years following measurement, a long period compared to similar studies in the literature. Nissen et al. [29] evaluated the efficacy of using asthma-specific diagnosis codes in identifying asthma patients in a large UK primary care dataset, and concluded that a recorded asthma-specific Read code had a high predictive value and was sufficient to identify patients with asthma. This study did not therefore include asthma medications.

However, we do acknowledge a number of limitations with our study. While secondary care datasets have full coverage of the population, the primary care dataset includes only 80% of Welsh GP practices who had agreed to share data with the SAIL databank, covering approximately 83% of Welsh population. Thus, some children who were included in the CMP 2012/13 dataset may be registered with non-participating GP practices. Also, only 84.3% of eligible children actually had their measurements taken as part of the 2012/13 CMP programme [11]. Parents can opt out of having their children’s measurement taken. Also, there was a measles outbreak in Wales during the spring of 2013, which may have affected whether children were measured. So it is unclear whether the missing data was missing at random.

This study relied on whether a diagnosis code for asthma was present in the child’s medical record. Asthma has a broad spectrum of symptoms and diagnostic criteria may vary across providers and settings. Finally, we were able to adjust for a limited number of confounders, namely sex and deprivation status. Other possible predictors were not corrected for such as other chronic conditions, family history of asthma and others.

This study adds to the growing body of evidence supporting the association between childhood obesity and asthma. The strengths of this research in terms of cohort size, exposure measure standardisation and long follow-up period support the significance of its contribution to this area of research. We have shown the utility of data derived from national surveillance programmes such as the CMP. A follow-up measurement later in childhood will provide valuable information regarding the trajectory of weight status during childhood and open further research avenues.

While a growing body of evidence supports the relationship between obesity and asthma in childhood, few studies have examined the effectiveness of weight loss interventions and lifestyle changes on the severity of asthma symptoms and disease control [30], and further studies are needed to explore the effect of community and individual-level weight loss interventions on asthma symptoms and control. The complications of obesity are well documented in childhood and beyond, and our study suggests that further research is warranted on weight surveillance and management in children with asthma and wheezing disorders.