Making sense of the community services data set

Sarah Scobie and Jonathan Spencer reflect on the new community services data and argue there is much more we still need to know. If community services matter then commissioners, providers and regulators need to take improvements in community data quality seriously.

Blog post

Published: 29/08/2018

The amount that NHS trusts got for community services fell 4% last year, which seems at odds with the Five Year Forward View’s aspiration to shift services out of hospitals, into the community and closer to patients.

But the community services landscape is complicated. There is a diverse range of providers, both within and between areas – with community services split between acute, mental health, and combined NHS trusts, community interest companies, private providers, local government and the third sector. It’s not clear whether the fall in NHS trust community services spend reflects a drop in commissioning services from NHS providers, or a fall in community provision overall. 

What is happening in the community?

A lack of data about services in the community makes it difficult to unpick where, what and how services are being provided. 

However, seven years after a standard for community data was first published, there is now a community services data set (CSDS). This is a mandatory flow of patient level data about NHS-funded community services to NHS Digital, and since October 2017 has covered both adults and children. The data from October to April is now available, and starts to give a sense of the scale of service provision. 

Over the seven months there were over 4.7 million referrals and 26 million care contacts recorded for people over 18, reported by over 100 providers. By comparison, over a similar period we would expect approximately 9.5 million A&E attendances for adults, and 50 million outpatient appointments.

Just under half of all referrals to community services were for people 65 and over, and almost a fifth of referrals were for people 85 and older. Musculoskeletal services account for the largest number of referrals for adults, and the most common 20% of services cover 80% of referrals.

Data quality is a challenge

While these numbers might indicate the scale of service provision, there are significant challenges in making sense of them.

One giant caveat is that the data is only for the providers who have submitted data, and not for England as a whole. And even counting the number of organisations that provide community services is not simple. When it comes to NHS providers, while all but one of 18 community trusts submitted data to CSDS, only 77 of 172 NHS providers have done so (see chart).

Number of trusts providing community services who have submitted data to CSDS 29/08/2018



A trust is deemed to have submitted if they recorded any successful submission between October 2017 and April 2018. Trusts providing community services identified from trusts providing reference costs.


Nuffield Trust analysis of Community Services Data Set, NHS Digital data.

Read more

The number of missing non-NHS providers is more difficult to determine, which is worrying given that non-NHS providers make up nearly half the value of contracts for community services. While five local authorities and 21 independent sector or community interest company providers submitted community data to CSDS, a review of community services providers registered with CQC indicates we would expect 20 local authorities and at least 33 community interest companies to have done so, as well as other independent providers.

There are also large gaps in providers’ submissions to the CSDS. Over half of referrals have a missing or invalid reason for referral, and one provider in eight has not submitted any information on care activities.  

What do we need to know?

There are some baseline questions we need to answer about services provided in the community. For example:

  • What services are delivered by community service providers? How do these vary geographically, such as by CCG, GP registered practice, and organisation type? 
  • Who are the patients, and what activities are undertaken, for which cohorts of patient? Are there disparities in service provision associated with deprivation or other community characteristics? How does case-mix vary?
  • How do quality and outcomes of care vary? For example, how long do patients wait for treatment, and what happens when they are discharged from care?

A critical obstacle to using the current data is uncertainty about what proportion of the population is covered. This makes it difficult for providers, commissioners or health systems to use the data for benchmarking or to monitor activity trends. Community services are by definition population based, and analysis on a population basis would enable the quality of data to be understood, and indicate to integrated care leaders where the data gaps are.

We recommend, as a start, publishing referral rates by broad age group at local authority or CCG level. This would identify outliers where data has not been submitted from one or more providers.

A standard set of activity measures is also urgently needed, to understand the variation in delivery of community services. Metrics could cover waiting times, discharges, caseload, and time spent on care, as well as the current data on referrals, contacts and activities. Given the diversity of services provided in the community, for benchmarks to be meaningful they will need to be segmented by service, professional group, and patient demographics. For providers, addressing the gaps in data is going to be critical for implementing the productivity measures proposed for community services. Only one of these measures – care contacts/patient – can currently be calculated from the published data, and is only available at organisation level, and not by service. 

If community services matter, then commissioners, providers and regulators need to take improvements in community data quality seriously. Improving community data will benefit everyone, and enable systems to have a fuller understanding of the care needs of their population. All providers need to submit data, and be better at completing existing submissions. Commissioners can also help with the completeness of data submitted for their populations – and ensure that independent, as well as NHS, providers are submitting data.

At the Nuffield Trust, we will be investigating what can be learnt from the existing published data for a subset of providers who are submitting high-quality data on a regular basis.

Suggested citation