How to use data for practice based commissioning

Commissioning is a strategic process: assessing health needs, developing new services or providers to meet those needs, contracting for services, and undertaking a range of strategic efforts to improve population health. ¹ Practice based commissioning (PBC) puts the budgetary responsibility onto GP practices, who, because of their unique knowledge about their patients, are thought to be better placed to commission, redesign, and deliver these new services. ²

Provision of data is key to this process. The Department of Health in England expects primary care trusts (PCTs) to provide this data for commissioning in a timely and helpful manner. In return, general practices will produce PBC plans that will demonstrate a good understanding of their population's health needs. ³ Further clarification by the Department of Health also suggests that if PCTs don't fulfil these information needs, the practice based commissioners will be at liberty to commission these data services from outside the PCTs. ⁴

Unfortunately the vast majority of primary care data currently available is a by-product of administrative processes. For example, the quality and outcomes framework (QOF) is a payment rather than a clinical monitoring system. ⁶ In addition, the quality of NHS data has been questioned. A potential solution is to have greater involvement of clinical staff in validation of this data. ⁷

In an ideal world validated health data would be routinely collected, stored, shared, and disseminated in a format and timescale appropriate to its use and users. In practice there are problems with all of these requirements.

Data sources

The Department of Health has specified a minimum data set for PBC. Table 1 lists this information together with some suggestions for where the data may be obtained. ¹ Commercial suppliers of data for PBC, such as Dr Foster Intelligence, have expanded the data set. The most routinely used data sets are Hospital Episode Statistics (HES), Commissioning Data Sets (CDS) and now various data sets provided by commercial suppliers.

HES, CDS and commercial suppliers of data all obtain their data from the same source: the secondary user services (SUS). This is an NHS-funded body that draws its data from hospital activity in England. It includes both inpatient and outpatient activity.

Of the three data types, CDS is presented in the most 'raw' format. Little manipulation has been performed although records can be drilled down to individual (anonymised) patient level. Lag time between activity and availability of data is about six weeks. High level expertise in manipulation and analysis are required when using this data set.

HES is a more refined version of CDS whose fundamental unit of measurement—the finished consultant episode (FCE)—links 'spells' of activity together. Each spell of activity could include one or more FCE. In order to protect confidentiality, data access is restricted to tables of information; individual patient data are not available. High level analytical skills are also required and lag time of data availability is up to nine months.

The third data source derived from SUS (mainly from HES) is provided by various commercial suppliers of data. These data are usually presented in a format which is more suited to end users.

Problems in using routine data

There are potential problems with data validity and restrictions around the smallest geographical area for which data can be obtained. Knowledge of the locality and the limitations of the particular data set are therefore essential.

This is the job of the data analyst who will spend a considerable amount of time gathering, preparing, and transforming data into a format appropriate for users. This is because information analysis is an iterative process: the original request produces initial data, which leads to refinement of the question and further data searching and analysis, and so on, until the final product is realised. Good working partnership between PCTs and practices is essential to ensure the final product is relevant and useful. Although subcontracting this work to a commercial organisation offers a range of presentation tools and flexibility, specialised interpretation is still needed. The optimal use of PBC data is a developmental process that will take years to become fully established.

What five things should your PCT be doing?

1. Provide a short briefing document that is a high level health needs review and a local basket of indicators of health needs and targets including local area agreements: this will be a starting point to stimulate a lower level analysis, but be wary that PCT priorities may not be a problem for all practices
2. For each area of interest, provide GPs with age-standardised rates to enable benchmarking: number of units (eg, number of tests or procedures) to contextualise the data; and costs
3. Provide data in a logical and useful format by combining data sets (eg, QOF, CDS, and deprivation score) into one meaningful graph: for example, by combining smoking-related admissions with smoking cessation data and costs at practice level
4. Produce maps of prevalence or mortality in very small geographical areas: this will visually demonstrate variation and inequalities in health between and within consortia, which will facilitate targeting of resources
5. Provide advice on a growing body of evidence around primary prevention: a significant PBC requirement is for practices to improve health as well as treat disease

We are grateful for the contributions from the following Western Cheshire PCT staff: Cathy Davis, Liz Keight and Jill Oakley.

Where to find more information

Contributors