Unpaid carers play a vital role in supporting vulnerable children and adults. The essential care they provide to family and friends who are ill or disabled is estimated to be worth more than £152 billion a year to the economy.
Identifying who is a carer is essential for ensuring that people who provide care have access to the right support. There are around 4.7 million unpaid carers in England, or 8.8% of the overall population, according to the national 2021 Census, so better identification of carers – including young carers – needs to be a policy priority for the government.
The government did set out a clear ambition for better identification of carers by general practices in 2019, stating that this is necessary to improve carers’ health and to reduce unwarranted variations in the support that they receive. General practices can play a critical role in helping ensure access to health checks and flu jabs, or targeted screening, for example for depression among people caring for someone with dementia. This support is important: research shows that for people providing 50 hours or more of unpaid care a week, the health impact of being a carer is equivalent to losing 18 days of full health each year. To provide better support for carers, general practices first have to know who they are.
So how many of the 4.7 million unpaid carers in England identified in the Census are captured in GP data, and what do we know about them?
What we did and why
We used data from 13 million anonymised patient records held by general practices in England as part of the Clinical Practice Research Datalink (CPRD) and compared this with data for England from the 2021 Census. (See note 1 at the end of this article for details about our methods.)
Our aim was to estimate what proportion of the 4.7 million carers identified in the 2021 Census were identified as a ‘carer’ in patient records. This helps us to understand how good general practices are at identifying and recording unpaid carers.
Just under a million carers are missing from general practice records
In a dataset of 13 million general practice records, we would expect to find around 1,153,889 unpaid carers if identification of unpaid carers was the same as in the Census – that is, about 8.8% of all patients. In fact, we found that general practices identified far fewer – just 188,038 unpaid carers, or 1.4%. This means there were just under a million ‘missing’ carers in the GP practice records we analysed.
Our results suggest that for every one carer identified by general practices, there are four more that are not being identified, according to Census data.
Who is missing?
Although the gaps between general practice and Census records exist across all patient groups, our analysis reveals that carers from certain population groups are more likely to be missed by general practices.
In the Census, 9.4% of people from the most deprived areas of England self-identified as an unpaid carer, whereas GP records have recorded just 1.4%. In comparison, 8.4% of people from the most affluent areas self-identified as unpaid carers according to the Census, whereas 1.8% were picked up in GP records.
Taking these numbers and making an adjustment for age and sex differences between these populations first, we estimate that general practices are picking up around 14% of unpaid carers in the most deprived areas, and 28% in the most affluent areas. People living in the most deprived areas are therefore around half as likely as those in the most affluent areas to be identified in GP records as an unpaid carer.
The chance of being identified as an unpaid carer by your GP practice depends on where you live
ChartEthnicity is also a factor in how likely people are to be identified as a carer in GP records. People of Black, Mixed or ‘Other’ ethnic backgrounds are less likely to be identified as unpaid carers by general practices than people from a White British background.
Using similar calculations as for deprivation, 6.8% of people from a Black background self-identified as a carer in the Census, whereas GP records pick up just 0.8% of people from a Black ethnic group as unpaid carers. In the Census, 9.2% of people from a White British background self-identified as an unpaid carer, compared to GP records where 1.6% of those from a White British background were picked up as a carer.
Again, taking these numbers and adjusting first for age and sex differences, we can estimate that 11% of carers from a Black background and 16% of carers from a White British background are being identified in GP practice data. Carers from a Black background are therefore around a third less likely to be identified by their GP compared to those from a White British background.
Where the largest gaps are
People who are young, Black and female are being poorly identified as unpaid carers in GP records compared to what might be expected based on the Census. We estimate that GP records are picking up just 7% of Black female unpaid carers aged 5 to 24.
Likewise, we estimate that just 5% of unpaid carers who are Black male adults aged 25 to 44, and 7% of unpaid carers who are Mixed ethnicity and male aged 25 to 44, are being picked up in GP records.
All of which shows that young carers, particularly those of Black and Mixed ethnicity, are among the groups of unpaid carers most likely to be missing from GP records.
The chance of being identified as an unpaid carer by your GP practice depends on your ethnicity
ChartWhy are these groups not being identified as carers in GP records?
Since the Census records people who self-identify as carers, it is unlikely that the issue lies with people themselves not recognising their caring role. There are several other possible explanations for these findings.
Certain groups of patients may be less likely to attend their GP practice and, therefore, there could be reduced opportunities for practice staff to identify someone who is a carer. It could also be the case that certain patients are systematically less forthcoming in disclosing their role as a carer, or may feel this is not relevant information to share with their GP. Some patients are also more likely to experience greater communication difficulties or language barriers.
We also know that GP consultations are shorter in deprived areas. This might allow less time for the GP to record who is a carer, and also, critically, less opportunity for GP staff to build the kind of rapport that encourages and supports people to disclose to their GP that they are a carer, or enables them to articulate the impact that a caring role may be having on their own health. And while our data do not allow us to be definitive about why certain patients are less likely to be identified as unpaid carers, it is possible that GP practices may be systematically less likely to ‘see’ some groups of people as carers, or to identify them as such in GP practice records.
Measuring progress against policy ambitions
More than a decade since NHS England launched its ‘Commitment to Carers’, and despite initiatives from central government and NHS England to improve identification of unpaid carers as a gateway to better support, it is clear that limited progress has been made. While a ‘quality markers’ approach has helped shine a light on good practice, implementation by general practices has been inconsistent. A core reason for this is that collecting this information at practice level remains voluntary.
A concerted effort is needed to improve identification for all carers – in order to close the gap between identification of unpaid carers in the Census and in general practice records. At the same time, general practices should consider auditing their own practice to ensure unpaid carers are being identified effectively, paying particular attention to the potential for recurring gaps in recognition. For example, internal auditing should focus particularly on inclusion of those from Black and Mixed ethnic backgrounds, and must include both children and young adults.
Better support for unpaid carers relies on identification as a prerequisite. Yet our analysis shows that the identification of carers at GP practices is very patchy, and much scope for improvement exists. There could be new opportunities here to build on – for example, afforded by the government’s energetic focus on digital and technology, enabling unpaid carers to self-identify within the NHS App, and for this information to be shared with care providers. Alongside this, we need new policy proposals that seek to address the under-identification of particular groups of carers by general practices, taking on board evidence from our analysis.
Notes
- In this analysis we used data from the latest English Census collected on 21 March 2021, and compared this with data from anonymised patient records from GP practices. The national Census is one of the most trustworthy sources of data on unpaid carers. The wording of the Census 2021 asks "Do you look after, or give any help or support to, anyone because they have long-term physical or mental health conditions or illnesses, or problems related to old age?" People were asked to exclude anything they did as part of their paid employment. Individuals aged less than 5 years were excluded in both the 2021 Census and CPRD analysis.
- In our analysis of CPRD data we used 13 million anonymised patient records held by GP practices in England: the Clinical Practice Research Datalink (Aurum) contains patient-level records covering around 13% of the English population and is broadly representative of the patient population in England in terms of deprivation, age and gender. There are >80 existing SNOMED CT codes which relate to unpaid carers, and we found GP practices vary a lot in how they use them. Some codes are generic (‘is a carer’ or ‘cares for a relative’; others are more specific such as ‘carer for person with dementia’ or ‘young carer’. While we cannot rule out entirely the possibility that some GP practices may have coded a professional who provides care as part of their job (e.g., nurse, doctor) as ‘is a carer’ in their patient record – this would be inconsistent with advice and guidance from NHS England which specifies, for example, in the context of Covid vaccinations that the ‘carers flag’ in primary care records is to be used for identifying unpaid carers).
- In the CPRD data, carers were identified as anyone recorded as an unpaid carer at any time in the five years up to the date of the Census, with no subsequent record to indicate that they had ceased being a carer in the intervening time.
- For the avoidance of doubt, we can not be certain if people identified as carers in GP records are the same individuals that self-identified as carers in the 2021 Census, as our analysis does not allow for ‘matching’ of individual people.
- Prevalence ratios are adjusted for age and gender. To improve recording of ethnicity in our analysis, data from general practice records (CPRD) was linked to hospital data (HES): this results in greater completeness of ethnicity recording than is possible using GP records alone.
- We found particularly low rates of carer identification in GP records among people coded with “Other” ethnicity. However, the numbers were relatively small and we cannot be sure that the reporting of this group is consistent between GPs and the Census.
We are grateful to the Health Foundation for their support in enabling access to CPRD data used in this analysis.