Network Data Lab - Topic 6: Housing and Health — Portfolio

Project Description

This is a Health Foundation-funded project, carried out through the Cheshire & Merseyside NHS Secure Data Environment, linked routinely collected primary and secondary care records to household-level data for over 2.5 million patients and 1 million households across the Cheshire & Merseyside Integrated Care Board area.

The core aim was to understand how housing composition, energy efficiency (EPC bands), and the Liverpool City Region housing retrofit programme relate to health needs and NHS service use, and whether retrofit delivery is spatially aligned with the areas of greatest need.

Key areas of analysis:

Cohort linkage: Matched patient records to households via Unique Property Reference Numbers (UPRNs), enabling analysis at both patient and household levels
Household composition: Classified households into types (single-person, large family, multi-generational, elder-only, bereaved, multi-morbidity) and examined their relationship to healthcare demand
Seasonal variation: Compared winter and non-winter healthcare utilisation (A&E, emergency admissions, GP appointments, prescribing) across EPC bands
Need for Housing Retrofit Index: Developed a composite UPRN-level index aggregated to LSOA/MSOA, combining health vulnerability indicators (cardiovascular and respiratory admissions, mental health prescribing, frailty proxies) with energy efficiency and deprivation
Spatial alignment analysis: Used bivariate mapping and GLMs to quantify the mismatch between retrofit need, eligibility, delivery, and healthcare utilisation across Liverpool City Region

Key findings included that household composition explained approximately 45–52% of the variance in health needs and utilisation across LSOAs, and that retrofit delivery was not consistently targeted toward areas of highest need — particularly in Liverpool, Knowsley, and St Helens.

The project included a PPIE component with public advisors who had experience of housing retrofit schemes, contributing recommendations on eligibility criteria, communication, and equitable delivery.

Skills Used

SQL / R / RStudio (GLM, spatial analysis, data linkage)
NHS Secure Data Environment (SDE / NDL)
GIS and bivariate spatial mapping
UPRN-based household linkage
Composite index development
Primary and secondary care data (SUS-APCS, SUS-ECDS, GP datasets)
Public and Patient Involvement and Engagement (PPIE)