Identify which ICBs are ‘alike’ based on various characteristics
2-stage approach was used:
1. Each variable is validated and standardised by:
a) Capping each variable value at 5 standard deviations over the mean – to avoid
b) Taking square root of all values – to reduce skew
c) Subtract mean and divide by the standard deviation (of square-rooted values)
2. A calculation of similarity (Euclidean distance) is then completed - this uses the standardised variables for two ICBs in each pair from the first stage of this approach and the weights associated with each variable. This produces a distance matrix, ranking the similarity distance between each ICB The similar ICBs are those with the lowest value in this matrix. The closest 5 to each ICB were chosen as the suggested set of peers.
Variables included in model Development:
• Adult population age groups (18-39, 65-84, 85+)
• The percentage of population with Rural/Urban residence
• The percentage of population by ethnicity (White British, Non-British, Mixed, Asian, Black, Arab or Other)
To produce the aggregated ICS level data, where the variable was a proportion, a weighted average was calculated. This averages the CCG level data while also considering the proportion that the CCG’s population makes up of the overall ICS population.
average Index of Multiple Deprivation (2019) score in the LSOAs where CCGs'
registered patients lived in April 2019
• The total population registered with CCGs' practices (April 2020)
• Adult population age groups (18-39, 65-84, 85+) in CCGs
• The percentage of people who said they are of white (non-British) ethnic origin (GP Patient Surveys 2017, 2018 and 2019)
• GP Patient Surveys 2017, 2018 and 2019)
• Percent of population who live in areas defined by the ONS Rural Urban Classification as "Rural town and fringe in a sparse setting", "Rural village and dispersed" or "Rural village and dispersed in a sparse setting" (April 2018))
Lists of 5 peers for each ICB.
Output imported onto Model Health System dashboard to support benchmarking and opportunity estimations.