In 2021, Westminster Council, London, piloted a new Community Health and Wellbeing Worker (CHWW) role in the Churchill Gardens council estate. Four CHWWs were assigned to specific buildings on the estate, to proactively visit the same 120 households that they were responsible for, each month, whether there was a clinical need or not. This approach led to many positive outcomes. In households that were visited by the CHWWs, residents eligible for immunizations, screening or health checks were much more likely to receive them, than households that had not yet been visited. Based on this positive impact, the four Primary Care Networks of Westminster (called Healthcare Central London) have recruited a further twenty CHWWs for deployment in January 2024. In order to give households equal chance of receiving the CHWW services, they have been allocated at random to neighbourhoods of 120 households in two wards, Lisson Grove and Paddington Green. Our study will draw on routinely collected data using the Whole Systems Integrated Care (WSIC) data warehouse, to look back at whether households that were allocated a CHWW were more likely, or not, to receive the immunizations, screening and health checks they were eligible for, than households not allocated a CHWW. This robust evaluation of the impact of the CHWWs will help policy makers understand whether lay people from the community can aid the uptake of and access to preventative services. Already there is national interest in the use of CHWWs, particularly ones that have a mandate to visit households even before the residents express any clinical need, because understanding and supporting families before their problems become too big, makes sense for residents and the health and social care system. This has already been seen in many countries around the world, but for England this evidence base is yet to be developed because, until now, there are no services in place such as this.

Background

Lay Community Health Workers (CHWs) are an inexpensive alternative to more costly health professionals, when delivering some interventions such as basic health promotion advice that does not require highly specialised health personnel.(1,2) Well conducted research shows that lay CHWs impact significantly on immunization schedule completion rates in the under-5s (RR 1.22, 1.10-1.37, p<0.001);(3) reduce overall morbidity in the under 5s (RR 0.86, 0.75-0.99, p<0.05);(3) treble exclusive breastfeeding rates at 6 months (RR 2.78, 1.74-4.44, p<0.001);(3) improve breast and cervical screening rates by 9.3% and 8.1% respectively (4,5) improve diabetes outcomes and blood pressure control.(5,6) Lay CHWs may be less qualified than clinicians or nurses, but because they are from the local community they are more understanding and empathetic of local issues and respond to community needs in a far more effective way than a healthcare professional would be able to do. Recent realist syntheses have shown that CHWs improve the access to public services, and increase self-esteem and a sense of self efficacy of the people they serve.(6) Recommended by the World Health Organisation (WHO) for all member States,(7) CHWs play an important role in the delivery of population-based primary care, particularly in response to COVID19 and vaccine hesitancy.(8) However, England has been slow to adopt a national, scaled CHW workforce as has been seen in other countries such as Pakistan, Ethiopia and Brazil.(9)

Since 1994, Brazil's scaled, national CHW model has led to significant improvements in population health outcomes. Rather than focussing on a priori need or sociodemographic characteristics, each CHW is responsible for 100-150 households in defined geographical areas and they visit each household on a monthly, or even more frequent, basis. This is a proactive, longitudinal approach, where CHWs build a trusting and continuous engagement with households. It has resulted in improved immunization and screening uptake because of personalised and tailored support, advice and signposting based on their intimate understanding of the household members, addressing concerns, cultural preferences, misinformation, and issues of access. This enables them to address all family members' needs and concerns much more effectively than when separate family members are targeted according to their socio-demographic and clinical characteristics, as occurs in the UK.(10) CHWs connect families to the wider health and social care ecosystem leading to reduced hospital utilization, admissions and inequalities, over time. Brazil is now deploying 270,000 CHWs to cover 100 million people across the country.(9,10)The CHWs have in part been responsible for a 4.5% reduction in infant mortality for each 10% increase in delivery of the CHW role;(11) a pro-poor increase in horizontal health equity; (12) a 13% reduction in hospitalizations for primary care sensitive conditions in municipalities with a high enrolment of the CHWs, even accounting for secular changes and other confounding variables,(13) and a 34% reduction in cardiovascular disease mortality in areas with high coverage of the CHW model.(14) In Brazil, this system has cost only $60 per person per annum.(15)

Context

Westminster is one of the worst performing local authorities for vaccine uptake (78% measles, mumps, rubella (MMR) MMR1, 85% diphtheria, tetanus, and pertussis (DPT), 60% COVID1, 53% COVID2, 21% COVID Booster, 38% Reception age flu vaccine, 61% 65+ flu vaccine),(16) due to its transient population (61% social renting),(8) but also because of the high proportion of Black, Asian, and Minority Ethnic (BAME) residents (49%) [16] and recent migrants (28% households have no English speakers).(17) In July 2021, Westminster Council piloted a new Community Health and Wellbeing Worker (CHWW) role in the Churchill Gardens Council Estate that was inspired by the Brazilian approach to primary care where CHWWs are responsible for defined micro-areas of around 120 households each, and have a mandate to visit each one, once per month, to provide health information, signpost to services, identify unmet need and promote health and wellbeing. In an evaluation of the Churchill Gardens pilot, which involved just four CHWWs visiting 120 households each in the first year of operations, there was an increased likelihood of uptake of immunizations (47%), screening (82%) and health check (190%) in visited households compared to unvisited households, for residents eligible for those services. This was due to proactive (through monthly home visits) identification of residents eligible for these services, resolving pressing social care concerns and having meaningful conversations around health and wellbeing.(18) Based on this experience, Westminster City Council (WCC), Healthcare Central London (HCL) (the four Primary Care Networks in Westminster) and One Westminster (the Voluntary sector) have collaborated to expand the CHWW pilot into the Lower Super Output Areas (LSOAs) in the bottom 20% deprivation centiles in North and South Westminster. Drawing on Additional Roles Reimbursement Scheme funds through NHS England, a further twenty CHWWs will be recruited in between September and December 2023. The overall aim of this research is to evaluate whether the scaled CHWW intervention leads to improved uptake of preventative care services (immunizations for all ages, cancer screening (breast, cervical and bowel) and NHS Health Checks) for households in high need areas (top 20% deprivation index) in London.

Intervention

CHWWs are a holistic, comprehensive, community-based support role. They are recruited from the local community (or within walking distance to it), paid full- or close-to-full time, and provide support and advice across all areas of the life course, and for any age group. This includes, but is not limited to, breast feeding advice, immunization and screening promotion and advice, healthy lifestyles and promoting wellbeing, improving community cohesion and social isolation, providing housing and social care support, signposting to wider services, identifying unmet need in households, identifying the early signs of, and supporting the management of chronic disease and any mental health issue, and referring in to the primary care team, community trusts, voluntary services and the Council services when required.

Each CHWW is deployed to cover around 120 households in defined geographies in North and South Westminster. Lower Super Output Areas (LSOAs) in the bottom 20% deprivation have been identified as priority areas for the CHWWs to be deployed and, within these LSOAs, sixty 'villages' of 120 households each have been mapped. These 'villages' are all localities that have over 80% registration of residents to the three participating GP practices where the CHWWs will be integrated into as part of their primary care teams. The intention is for all villages to eventually receive the CHWWs as part of a wider policy to create a workforce that is more place-based, proactive and personalised however, given there are not enough CHWWs being recruited in this expansion phase to cover all the 'villages', and given the need for a fair and equitable allocation of the CHWWs in this stage of the expansion, the provider consortium (HCL, WCC and One Westminster) have elected to randomly allocate 'villages' to either receive the CHWW or not. Residents will therefore have an equal chance of receiving the services of a CHWW, providing a perfect opportunity for a robust, parallel cluster study to determine the impact of the CHWWs in the intervention areas compared to the control areas, using retrospective analysis of data such as primary care service utilization, preventative service uptake and secondary care use in control and intervention households.

The intervention will involve monthly (or more frequent) household visits by the CHWW that was assigned to the village, enriched by a digitally-enabled intervention App with reminders and interactive educational materials for participants. All households in the villages allocated to receive the intervention will be approached by the CHWW allocated to that village, irrespective of demographic characteristics and previous needs. CHWWs will aim to visit all households each month, building a longitudinal relationship with each household so that problems arising from one month to the next can be captured in good time.

6. Methods 6.1 Study Hypothesis

The hypothesis to be tested is that a multifaceted, proactive, place-based, neighbourhood-level community health worker-led intervention will improve the uptake of preventative care services for households in areas of high deprivation.

6.2 Trial/Study Design

This is a retrospective, observational evaluation of the uptake of preventative services in households allocated to the CHWW intervention, compared to those that received usual care. Taking advantage of the random allocation of the CHWWs to twenty out of sixty 'villages' identified in the LSOAs in the bottom 20% deprivation index, the evaluation will compare the outcomes in the parallel intervention and control clusters, with a retrospective baseline observation period, involving between 120-150 households per village per fulltime CHWW.

6.3 Study Population and Setting

Although one of the wealthiest London boroughs, home to the Government in Whitehall and main tourist hotspots, Westminster is an inner-city London borough characterised by areas of high deprivation, with many council or social housing estates that have large migrant populations, populations with multiple socio-economic needs, transient populations, and people living with multiple unmet health and social care needs. The council estates of Pimlico and Lisson Grove are particularly challenging in terms of high need, and hard-to-reach population groups. Both these areas, in North and South Westminster, are priority geographies for Westminster Council and its partner organizations, to address unmet need and improve access to services. In the context of this study, the deployment of the expanded cadre of CHWWs fits well with the agenda to improve access to services, supporting residents closer in their homes, and identifying and addressing unmet need. Adhering, though, to the model of deployment for the CHWWs, the provider organizations (HCL, WCC and One Westminster) have identified ten LSOAs in the lowest 20% deprivation index in these wider geographies. Within these LSOAs, HCL has identified all residents registered to GP practices in Westminster and wider afield. It has mapped the residents to households, by matching full addresses from the primary care records, and then geo-located households into discrete 'villages' of approximately 120 households in each. There are approximately sixty 'villages' where over 80% of the residents are registered to the three partnering GP practices (Lisson Grove Medical Centre, Paddington Green Health Centre in North Westminster, and Victoria Medical Centre) in South Westminster. Each cluster will be allocated either to receive the intervention (the CHWW) or serve as control. Residents of any age that are registered to the participating GP practices serving North and South Westminster and that are living in households within the clusters are included.

6.4 Eligibility Criteria The study population includes all residents eligible to receive the CHWW intervention if they are registered to a GP practice and live in a postcode that is in an LSOA that is in the bottom, i.e., most deprived, 20% deprivation index in the Lisson Grove or Paddington Green wards of Westminster.

6.5 Identification of Target Cohort HCL will identify the target cohort based on their address, deprivation index, and the GP practice where they are registered.

6.6 Randomisation and Allocation Concealment

HCL will conduct a 'village' (cluster) level, 1:2, computer-based randomisation. HCL will randomly allocate CHWWs to the mapped 'villages' so that residences can have an equal chance of receiving the services of a CHWW. Randomisation will be stratified in north and south based on the eligible villages and on the number of the CHWWs allocated in these two regions. Randomisation won't be less than 1:1 and more of 1:3. Following baseline data collection, randomisation will be conducted at the level of the village. Allocation of villages to intervention or control will be performed by an independent statistician using a computer-generated random allocation sequence stratified by geographical area. Only cluster representatives will be informed of the allocation once randomisation has been performed. Eligible village participants will not be aware of the randomisation. Outcomes will be assessed via routinely-collected data - thus maintaining the blind. This randomisation offers the fairest way to distribute the CHWWs at this stage in the expansion in Westminster, and is an ideal opportunity to conduct an evaluation that controls for selection and confounding biases, that will enable a robust assessment of the impact of CHWWs within the intervention villages in contrast to the control villages.

6.7 Control Arm

Control group is usual primary care services without access to the features described above. All participating villages will receive 'usual' care from the GP practice to which they are enrolled.

6.8 Data Collection

The broad remit of the CHWWs means outcomes could feasibly be assessed across the entire life course, but this study will focus only on the impact, if any, on uptake of preventative services (immunizations of any type, cancer screening programmes, and NHS Health Checks). Outcomes will be assessed via routinely collected data derived from the Northwest (NW) London Whole Systems Integrated Care (WSIC) dataset. An application to WSIC Data Access Group (SDRAG) has been approved, granting approval for the use of the routinely collected data for this research purpose. We will draw on routinely collected data sources at 3 timepoints for each 'village' (intervention and control). The WSIC data is obtained from linked primary and secondary care data warehouse, covering 95% of the NW London population (2.3m residents). WSIC has identifiable care organisations and postcodes, allowing mapping and linkage to healthcare provider variables such as staffing, and includes contacts with social, community and mental health care. As of 2023, WSIC is including a Unique Property Reference Number (UPRN) to link individual records to households permitting household level analysis of impact of CHWWs. The flagging of the (identifiable) UPRNs in the intervention and control groups will be done by the WSIC team, and then all study analyses conducted using de-identified data.

At baseline, a complete data extract will be done in each village for all eligible people i.e. residents of households that have been mapped to each of the sixty 'villages' and registered to a GP practice. At each subsequent time point data extracts will be repeated for the target population.

Our primary outcome measure is a composite score of all immunizations, screening and health check uptake in both groups at the household level as previously described and proposed.(24) As WSIC data is not grouped by household, HCL will provide the addresses of all households that are entered into the study, and whether these are in the intervention of control groups, to be linked within the WSIC dataset. We will establish the reliability of either the address matching or UPRN matching approach. Individual eligibility for services during the intervention period will be determined by national guidelines and previous uptake. Secondary outcomes will include unscheduled GP visits, total emergency department visits, hospitalizations, re-admission rates and all-cause mortality.

6.9 Quantitative Outcomes Primary A composite outcome at household level for uptake of preventive services defined as: vaccinations, cancer screening and NHS health checks (based on current recommendations/UK guidelines), that residents were eligible for and that were received during the intervention period - after the first and second full year of intervention.

Secondary

• Individual components of the primary household composite outcome
• unscheduled GP visits
• total emergency department visits
• total hospitalisations
• 30-day re-admission rates and all-cause mortality

6.10 Sample Size

Sixty villages randomised 1:2 (intervention vs control) to a cluster parallel design with an observational retrospective baseline period will provide > 90% power (2α=0.05) to detect an absolute improvement of 10% in the uptake of preventive services assuming a within-period intraclass correlation coefficient (ICC) of 0.15 and baseline uptake of 15% under usual care conditions (demonstrated from pilot evaluation). This calculation assumes a "two-period decay" thus allowing for the within-cluster correlation within the same measurement period to differ to that from a different measurement period. We have assumed that the within-cluster correlation from different periods would be 80% of the correlation from the same period. For repeated within-household measurements, we have assumed a correlation of 0.5. Finally, this calculation assumes that at least 120 households will be included in each village, thus resulting in a total of at least 7,200 households (2,400 randomised to the intervention, and 4,800 randomised to usual care). Calculations were performed using the online shiny calculator available at https://clusterrcts.shinyapps.io/rshinyapp/.

6.11 Statistical Analyses Data will be analysed at the household level, adjusting for clustering by village following the intention-to-treat principle i.e., by analysing each household according to the randomised group, regardless of actual exposure to the intervention. The main model will be a repeated-measure logistic regression with within-cluster correlations modelled using random effects. Within-household correlations due to multiple measurements (on measurement during the baseline period and one during the intervention period) will be modelled using random or repeated effects. The intervention effect will be estimated as the adjusted odds ratio and 95% confidence interval and converted to an adjusted absolute difference. A detailed statistical analysis plan will be developed and made publicly available prior to unblinding.

Sub-group analyses will be conducted according to pre-specified analysis plan including sensitivity analyses, potential covariate adjustments, analyses for secondary endpoints and detailed assumptions (e.g., missing data handling) will be developed prior to unblinding.

7. Study Timelines Steering group formation, study set up, IRAS/HRA application, DAG application, statistical analysis plan - Q3/4 2023 Data extract WSIC - Q2, 2024, Q2 2025, Q2 2026 Statistical analysis, unblinding - Q4, 2026 - Q3 2027 Publication preparation, dissemination - Q3 2007

8. Ethical Considerations

The intervention and the randomization component are happening whether this evaluation takes place or not, and so on the basis of this we have applied for Health Research Authority (HRA) Approval rather than Ethical Consideration as advised by the Imperial College London Research Ethics Committee (REC). The ethics committee sits under HRA, and when HRA provides approval it is inclusive of ethics. We have also applied to, and received approval for, access to the WSIC data for the purposes of this retrospective evaluation, which falls within the WSIC global ethics approval provided for use of data for research purposes in NW London (current reference is REC18/WM/0323/IRAS252449). As the study only involves anonymised, unidentifiable routinely collected data analysed retrospectively, it does not require further ethical opinion. We have not carried out a Data Protection Impact Assessment (DPIA) because a DPIA is necessary whenever controllers intend to conduct processing activities that are likely to result in a high risk to the rights and freedoms of data subjects. The WP29 developed a list of criteria for assessing the high risk involved in certain types of processing, including, among others:

• the existence of assessment or scoring operations,
• the presence of automated decision-making,
• systematic monitoring,
• the use of special categories of data,
• the existence of large-scale data processing,
• matching operations between different databases,
• personal data relating to vulnerable individuals,
• the use of new technologies,
• the fact that the processing may inhibit the data subject from either exercising their rights or using a particular service.

Our study is not high risk according to any of these criteria.

9. Data Safety Monitoring Committee Data Safety Monitoring Committee (DSMC) not required as all falls within the acceptable use and governance of WSIC and has been approved by SDRAG based on retrospective, anonymous analysis of routinely collected data. No primary data will be gathered.

10. Publication Policies The research team will be responsible for ensuring standard practices for dissemination of the results either via peer reviewed publications, policy documents, or media releases. Wherever necessary, expert opinion would be sought to guide that group on appropriateness of the content. A publication protocol will be developed and follow.

11. Significance CHWWs are effective at identifying unmet need, coordinating care, bridging health and social care, addressing loneliness and social isolation, identify mental health problems, encouraging residents to access health care services, resolving intractable issues with housing or domestic violence, and many more things besides.(19) Since 2022, building on the community of practice established through the National Association of Primary Care,(20) the CHWW role is being piloted also in Bridgewater, (ten CHWWs) in Calderdale (twelve CHWWs) and in Cornwall (eleven CHWWs). Other places such as Frimley, Milton Keynes, Thanet, and Norfolk are all looking to do the same. The National Association of Primary Care is promoting CHWWs as a go-to strategy to integrate health and social care and strengthen primary care.(20) The pilot in Westminster was cited as a best practice Case Study in the 2022 Fuller Primary Care Stocktake Report(21) and was nominated for a Quality Improvement Award by the Royal College of General Practitioners, as well as an MJ Award for Local Authority Innovation in 2023. The work will feature in the House of Lords Enquiry into the Future of Primary Care and has featured in multiple national and international media reports as a rare example of the Global North learning from the Global South.(22,23) This research will support an emerging evidence base around the benefits of a proactive, hyperlocal, community-led support service that is integrated into primary care. The findings are likely to advance locally relevant knowledge on scaling up a strategy to overcome entrenched inequities in access to effective health care for under-served populations. By placing this research at the intersection of policy, health care providers and consumers, the evidence generated has substantial potential to inform decision-making for system planners. Such approaches, if found to be effective and cost-effective, have the potential to positively impact on the healthcare of millions of people in the UK on a daily basis.

12. References

1. Walt G. Community health workers in national programs. Just another pair of hands? Milton Keynes, Open University Press, 1992.
2. Kahssay H, Taylor M, Berman PA. Community Health Workers: The Way Forward. Geneva, World Health Organization, 1998.
3. Lewin S et al Lay health workers in primary and community health care for maternal and child health and the management of infectious diseases Cochrane Database of Systematic Reviews 2010, Issue 3.
4. Baron RC, Rimer BK, Breslow RA, Coates RJ, Kerner J, Melillo S, et al. Client-directed interventions to increase community demand for breast, cervical, and colorectal cancer screening a systematic review. Am J Prev Med. 2008 Jul;35(1 Suppl):S34-55.
5. Viswanathan M, Kraschnewski J, Nishikawa B, et al. Outcomes of Community Health Worker Interventions Agency for Healthcare Research and Quality. Evidence Report/Technology Assessment Number 18. AHRQ Publication No. 09 E014 June 2009.
6. Kane SS, Gerretsen B, Scherpbier R, Dal Poz M, Dieleman M. A realist synthesis of randomised control trials involving use of community health workers for delivering child health interventions in low and middle income countries. BMC Health Services Research. 2010 Oct 13;10(1):286.
7. WHO (2012) - Optimizing health worker roles to improve access to key maternal and newborn health interventions through task shifting. Available at: https://www.who.int/publications-detail-redirect/9789241504843.
8. Bhaumik S, Moola S, Tyagi J, Nambiar D, Kakoti M. Community health workers for pandemic response: a rapid evidence synthesis. BMJ Global Health. 2020 Jun 1;5(6):e002769.
9. Haines A, de Barros EF, Berlin A, Heymann DL, Harris MJ. National UK programme of community health workers for COVID-19 response. Lancet. 2020 Apr 11;395(10231):1173-5.
10. Harris M. Integrating primary care and public health: learning from the Brazilian way. London J Prim Care (Abingdon). 2012;4(2):126-32.
11. Macinko J, Marinho de Souza M de F, Guanais FC, da Silva Simões CC. Going to scale with community-based primary care: an analysis of the family health program and infant mortality in Brazil, 1999-2004. Soc Sci Med. 2007 Nov;65(10):2070-80.
12. Macinko J, Lima-Costa MF. Horizontal equity in health care utilization in Brazil, 1998-2008. International Journal for Equity in Health. 2012 Jun 21;11(1):33.
13. Macinko J, Dourado I, Aquino R, Bonolo P de F, Lima-Costa MF, Medina MG, et al. Major expansion of primary care in Brazil linked to decline in unnecessary hospitalization. Health Aff (Millwood). 2010 Dec;29(12):2149-60.
14. Rasella D, Harhay MO, Pamponet ML, Aquino R, Barreto ML. Impact of primary health care on mortality from heart and cerebrovascular diseases in Brazil: a nationwide analysis of longitudinal data. BMJ. 2014 Jul 3;349:g4014.
15. Rocha R, Soares RR. Evaluating the impact of community-based health interventions: evidence from Brazil's Family Health Program. Health Econ. 2010 Sep;19 Suppl:126-58.
16. Office for Health Improvement and Disparities. Public Health profiles. Available at: https://fingertips.phe.org.uk.
17. Office for national statistics. Available at: https://geoportal.statistics.gov.uk/datasets/ons::lower-layer-super-output-area-2001-to-ward-2011-lookup-in-england-andwales/About.
18. Junghans C, Antonacci G, Williams A, Harris M. Learning from the universal, proactive outreach of the Brazilian Community Health Worker model: Impact of a Community Health and Wellbeing Worker initiative. on vaccination, cancer screening and NHS health check uptake in a deprived community in the UK. BMC Health Services Research (forthcoming).
19. Year One Evaluation of the CHWW programme in Westminster, Imperial College London, June 2022.
20. National Association of Primary Care. Community Health and Wellbeing Worker programme. Available at: https://napc.co.uk/community-health-and-wellbeing-worker-programme/.
21. NHS England. Next steps for integrating primary care: Fuller stocktake report. Available at: https://www.england.nhs.uk/publication/next-steps-for-integrating-primary-care-fuller-stocktake-report/.
22. BBC News. What the NHS is learning from Brazil. Available at: https://www.bbc.co.uk/news/health-65530272.
23. GP Online. Podcast: How an initiative from Brazil could help general practice and improve outcomes 2023. Available at: https://www.gponline.com/podcast-initiative-brazil-help-general-practice-improve-outcomes/article/1824651.
24. Watt H, Harris M, Noyes J, Whitaker R, Hoare Z, Edwards RT, et al. Development of a composite outcome score for a complex intervention - measuring the impact of Community Health Workers. Trials. 2015 Mar 21;16(1):107.

Inclusion Criteria:

• Individuals resident in the Lisson Grove or Paddington Green wards of Westminster

Exclusion Criteria:

• Individuals not registered to a GP practice

The study population includes all residents eligible to receive the CHWW intervention if they are registered to a GP practice and live in a postcode that is in an LSOA that is in the bottom, i.e., most deprived, 20% deprivation index in the Lisson Grove or Paddington Green wards of Westminster.

All data is stored and analysed in the secure WSIC data warehouse, and the terms of our ethics agreement is that data sharing is not possible.