Skip to main content

Publication, Part of

National Child Measurement Programme, England 2019/20 School Year

Official statistics, National statistics

National Statistics

Data Quality

In March 2020, schools in England were closed to the majority of children in response to the covid-19 pandemic. With schools closed and school nursing teams being deployed to support the pandemic response, NCMP measurements could not take place.  This section explores the data quality work undertaken to assess the impact of school closures on the NCMP dataset.

At a national level, the outputs of the data quality analyses included in this section and the accompanying data tables, indicate that the figures in this publication are directly comparable to earlier years.

There is significant variation at Local Authority (LA) level and for some LAs the data is not directly comparable to earlier years, and it is not advised to compare LAs to each other as differences may be the result of a partial collection of data rather than true differences in the prevalence of weight categories.

Descriptions of the analyses are set out below along with links to the associated data quality tables.

 

Local Authority data thresholds

When schools closed some LAs had already completed their data collection for the year 2019/20, others were in the process of measuring children and some had yet to start measuring children; therefore, the impacts of school closures are not evenly distributed across the country.

In consultation with Public Health England, thresholds have been set to determine how complete the 2019/20 data is and how robust any analyses based on it are. The three thresholds for reliability have been set at:

1. Reliable 
Where the number of children measured in 2019/20 is >=75% of the number measured in previous years, the data is considered to be a reliable estimate and comparable to analyses reported in previous years.
2. Fit for publication but interpret with caution
Where the number of children measured in 2019/20 is >=25% and <75% of the number measured in previous years, this is fit for publication, but caution is advised when using the outputs due to the low numbers measured.
3. Unreliable, suppressed
Where the number of children measured in 2019/20 is <25% of the number measured in previous years, the data has been suppressed from publication due to small sample sizes which may present an unreliable estimate of weight prevalence in that LA.

The number of children measured in each LA varies slightly from year to year. To better estimate how complete the data is in 2019/20, the number of children measured was compared to the mean number measured in the last three years and then assigned to one of the three categories above.

LAs often measure children in Reception and Year 6 at different times of the year. LAs have been assigned to these categories for Reception and Year 6 separately, therefore, there are many LAs who will be assigned to one category for reception and a different category for Year 6.

For LAs assigned to the ‘Unreliable, suppressed’ category, no further information is included in the publication regarding their data.

For LAs assigned to the ‘Fit for publication but interpret with caution’ category, it is advised that users refer to the more detailed data quality tables to help assess whether the data can be used for their purpose.

Please refer to Tables A and B for full details of these analyses and a complete list of which category each LA has been assigned to.

 

Comparing data submitted between 2018/19 and 2019/20, Local Authority and England

Many analyses in the NCMP compare the data in the current year to the previous year. Table C presents information on the number of children measured in 2018/19 and 2019/20 and the percentage change in numbers measured for both Reception and Year 6.

Tables C and D present analyses of the changes between 2018/19 and 2019/20 in the prevalence of underweight, healthy weight, overweight, obese, severely obese and overweight and obese combined. Included in the tables is the percentage point change in prevalence in each weight category between the two years and whether this change is statistically significant or not. Where the percentage point change in prevalence is listed as ‘Sig’ this indicates that this is a statistically significant difference and indicates a high probability (>=95%) that the difference seen is genuine. Please refer to Appendix F for more information on calculating statistical significance.

These analyses are included to allow users to understand the scale of changes in the prevalence of the weight categories at LA level and whether they are statistically robust.

 

School Cohort Analyses, 2018/19 and 2019/20

 

Tables F and G include the data to accompany this section.

The school cohort analyses are designed to assess the comparability of the data submitted during 2019/20 with the data submitted in 2018/19 by creating subsets of matched school level data. These analyses assume that the children attending the same schools in the two years are likely to share similar characteristics than may be the case in the wider LA. This allows an assessment of any differences that may be present due to sampling a subset of schools in 2019/20. Data was compared for schools that were included in the NCMP in both years and those that were only measured in one year.

LAs that were abolished in 2019/20 or created in 2019/20; or did not submit any data in 2019/20 have been excluded from these analyses. Please refer to Notes tab in the data tables for full details of the constructions of these measures.

Data is presented at national and LA level and separately for underweight, healthy weight, overweight and obese weight categories. The main information presented in Tables F and G:

  • Comparing data for all children in 2018/19 and 2019/20 by weight category
  • Comparing data from children in schools that were measured in both years by weight category
  • Presenting data from children in schools that were only measured in 2018/19 or in 2019/20

There are also 3 fields that present information on the changes seen between the groups:

  • Percentage point change in prevalence from 2018/19 to 2019/20 for all children in those years (Prev_chng)
  • Percentage point change in prevalence from children in schools that were measured in both years (SchPrev_chng)
  • Difference in change in weight prevalence by calculating SchPrev_chng – Prev_chng

These outputs allow users to explore any differences in weight prevalence in 2018/19 that existed between the schools that were and were not included in the 2019/20 collection and any bias these differences may have introduced into the 2019/20 dataset; and other similar comparisons that aid interpretation.

At a national level, the data in school cohort and submitted in a single year is very consistent and does not indicate that any substantial bias has been introduced to the 2019/20 dataset due to the subset of schools that were measured up to March 2020. This is the case for each of the weight categories: underweight, healthy weight, overweight and obese.

Please refer to the data tables to assess the impacts at LA level.

Ethnicity

Overweight and obesity prevalence varies between different ethnic categories, as reported in previous NCMP publications. Therefore, data quality checks have been undertaken to ensure that the dataset remains representative by ethnic category. The proportion of children measured in each ethnic group was compared between 2018/19 and 2019/20 and the following information is included in the outputs:

  • Number of children measured in 2018/19 and 2019/20 in each ethnic group
  • Percentage change in the numbers measured in each ethnic group between years
  • Proportion of children in each ethnic group in 2018/19 and 2019/20
  • Percentage point change in proportion of children in each ethnic group between years

The proportion of children in each ethnic category was very similar in both Reception and Year 6 with a maximum change of 1.2 percentage points for children in Reception in the White ethnic group. For more information please refer to Table H.

 

Deprivation

Overweight and obesity prevalence is associated with Index of Multiple Deprivation (IMD), as reported in previous NCMP publications. Therefore, data quality analyses have been undertaken to ensure that the dataset remains representative by IMD decile. The proportion of children measured in each IMD decile was compared between 2018/19 and 2019/20 and the following information is included in the outputs:

  • Number of children measured in 2018/19 and 2019/20 in each IMD decile
  • Percentage change in the numbers measured in each IMD decile between years
  • Proportion of children in each IMD decile in 2018/19 and 2019/20
  • Percentage point change in proportion of children in each IMD decile between years

The proportion of children measured in each IMD deciles was compared between 2018/19 and 2019/20 and was very similar in both Reception and Year with maximum change of 1.0 percentage point change for children in Reception in the 4th lowest IMD decile (decile 4), see Table I for more information.

Data Quality – Missing and imprecise data

Missing data

Table J includes the data to accompany this section.

The proportion of missing data items has improved over time although it still remains quite high for NHS number and ethnicity. 24 LAs had over 25% of records with missing NHS numbers and eight LAs had more than 25% of records without an ethnicity code.

 

Imprecise data

By chance, 10% of height and weight measurements would be expected to be whole numbers. However, there is some evidence of LAs rounding heights to whole numbers as nationally 17% of heights were whole numbers in 2018/19. This was over 30% for two LAs.

For more information: See Table A to J: Data quality indicators for the NCMP collection by submitting local authority.


Last edited: 28 October 2020 2:52 pm