Inequalities in English NHS talking therapy services: What can the data tell us?
Examining NHS Digital annual reports to look at deprivation
Mental health disorders such as depression or anxiety are experienced by many of us during our lives, with one in six people meeting criteria for a common mental disorder. In addition to the impact on people’s daily lives, poor mental health is associated with a higher incidence of many physical health conditions, and has far-reaching effects on society: 12.4% of all sickness absences from work in the UK have been linked to mental health disorders.
But there is variation in who is affected; for example survey data suggests that men in the lowest income households are three times more likely to have a common mental disorder than men in the highest income households. There is also variation in who accesses help; we know only 37.3% of those with common mental health disorders access mental health care treatment, but worryingly, a greater percentage of patients from low income households report unmet treatment requests.
At the Health Foundation we use data to tackle real world problems within health and health care, providing insights to help decision makers improve health care quality and outcomes and tackle health inequalities. This starts with quantifying those health inequalities, and examining what the data can tell us about what might be driving them.
This long-read will show data from NHS Digital on the England-wide Increasing Access to Psychological Therapies (IAPT) programme, highlighting variation in access and outcomes by relative level of deprivation.¹
We are focusing on mental health care given the current plans to invest in this area. As part of a commitment to address the historical under-investment in mental health services, the NHS Long Term Plan confirmed an additional £2.3bn per year investment in mental health care by 2023/24, including plans to treat 380,000 more patients per year through the IAPT programme. Mental health care has also been flagged as an NHSX priority area for digital innovation. Indeed, computerised and web-based therapy is already being implemented within IAPT services. Given this, we also highlight where digital innovation could to tackle current issues facing IAPT, but also the potential risks of doing so.
What is the IAPT programme?
IAPT is an ambitious programme to deliver psychological therapies at scale within England, which began in 2008. It is often regarded at the best at-scale roll out of these treatments worldwide. By 2018/19, there were 1,495,680 referrals to IAPT and 38.9% of these completed a course of treatment. 52.1% of treated patients recovered, matching recovery rates demonstrated in the controlled environment of clinical trials — an ambitious goal for the service to have achieved.
Policies to expand access to treatment should be good news for addressing health inequalities — but as Figure 1 shows, currently referrals from patients in deprived areas are less likely to receive a course of treatment than those in wealthier areas.
The pathway through IAPT services
IAPT offers various types of psychological therapies using a stepped-care model, where low- and high-intensity treatments are given, based on symptom severity and clinical judgement. There are also different modes of treatment available including web-based, group or 1:1 therapy. To access these, most patients self-refer into IAPT, often after sign-posting by clinicians in other services. The pathway for patients can be broken down into four key steps, shown in Figure 2. At each of these steps, patients might leave the programme; there is no clear data on why they leave but possible reasons might include the suitability of the service for their symptoms, the acceptability of the treatment type and mode to them, and waiting times for appointments.
We wanted to build an overview of how patients move through the service and highlight the patterns in the data, focusing on variation by level of deprivation². NHS Digital regularly publish on the programme, and the figures presented here use datasets from the publicly available IAPT annual report³— the code to generate each figure is available on our GitHub page. However, figures on assessment or starting treatment are not presented on a comparable basis to other stages,⁴ so we focused on referrals ended and treatment completed. We also looked at longitudinal trends where possible; data by deprivation level was first introduced into the IAPT reports in 2015/16.
IAPT is an exception among mental health care services as it closely tracks each patient’s symptoms, pooling this data to report regularly on the programme’s efficacy. This allowed us to consider variation not only access but also in treatment outcomes.
Variation in who accesses IAPT services
Figure 1 combines data on referrals and treatment completion in IAPT, but we also looked at each step separately to better understand the pattern seen. As shown in Figure 3, the number of referrals shows a clear gradient with deprivation level; double the number of referrals come from patients in the most deprived decile as compared with the least deprived decile. But the number of referrals that complete treatment does not increase with the same gradient across deprivation level; it flattens out for the most deprived areas. This mismatch results in the inequality of treatment access across deprivation levels.
Understanding variation in access
Without comparable information on the number of referrals reaching assessment or starting treatment, it is difficult to identify when patients leave the programme and this mismatch begins. However, the available data can give some insight into variation in the patient journey:
1) Thresholds for accessing treatment
Symptom severity and clinical judgement are used to guide treatment choices for each patient. We do not know if there is variation in patient need by deprivation, as symptom severity scores are not available.
But we can see the percentage of treated patients who were ‘non-cases’ — that is, their symptom scores were below the threshold of a clinically significant case (Figure 4). Non-cases may have substantial impairment not captured by the symptom questionnaires, and in these circumstances clinical discretion is important. However, the patterns in the data show that non-cases make up a smaller percentage of those treated in more deprived areas, as compared with less deprived areas. This suggest thresholds for accessing services are more stringently applied for patients from more deprived areas.
Over the last 4 years, we can see that the percentage of those receiving treatment that are non-cases has dropped across all levels of deprivation, and the differences between more and less deprived areas have been narrowing. Nevertheless, substantial differences persist - showing continued inequalities in the application of criteria for treatment access. It is important to gather more information to understand the reasons for this, to ensure both consistency and the ability of clinicians to exercise clinical judgement when appropriate.
2) Waiting times
Waiting times are also an important barrier to accessing treatment, but details by level of deprivation are not publicly available. We know that nationally, standards⁵ for waiting times for first treatment appointments were achieved in 2018/19, but there is substantial variation between providers. It is also worth noting that while patients wait on average 20 days for their first appointment, the waiting time between their first and second appointment is 49.1 days. This suggests issues in providing continuity of treatment, which may reflect service capacity — and is likely to impact on treatment completion rates too.
Outcomes: are there also variations in recovery rates?
The target for outcomes in IAPT is that has a 50% of treated patients should recover, with recovery based on patient symptom scores. While nationally this has been achieved, patients from more deprived areas have lower recovery rates and fall below this target (Figure 5). Nevertheless, recovery rates are improving, with outcomes for the most deprived patients showing year-on-year progress.
Recovery rates might vary because of the severity and complexity of the condition, with a patient’s social context an important factor to consider. But quality of care also plays a critical role, with factors such as waiting times, percentage of referrals treated and the number of treatment sessions identified as significant predictors of clinical outcome. Choice over treatment options may also be relevant — we know that treatments considered to be credible by patients are more likely to be effective.
What conclusions can we draw?
In this blog, we’ve highlighted inequalities in access and outcomes within IAPT. This is well-established within the field and discussed within the IAPT manual, which outlines evidence-based guidance for effective and efficient delivery of IAPT services. The longitudinal evidence suggests improvements are being made. However, further progress is needed and while we focus on deprivation, the data also shows that people from minority ethnic groups and those identifying as gay, lesbian or bisexual⁶ also have inequalities in access and outcomes with IAPT services.
But the public data we rely on here gives a limited picture on exactly where barriers in the pathway occur and do not allow us to assess possible reasons for variation. Key data missing include:
- Severity of symptoms
- Waiting times
- Types of treatment offered
It is also important to consider what happens to the 61% of referrals who do not complete treatment, and whether they receive appropriate support through other avenues. Anecdotal evidence suggests clinicians in other services are often unaware of a patient’s engagement with the IAPT programme due to a lack of information-sharing. This might undermine efforts to provide support through other routes.
NHSX has highlighted mental health care as a priority area for digital innovation, and there is significant potential for technology to be used thoughtfully to address some of this variation within IAPT. For example, digital tools could help services move patients along the pathway from referral to treatment completion, or could be used to enable communication between IAPT and other services, allowing clinicians to follow up patients sign-posted to IAPT and ensure they have appropriate support.
In terms of the delivery of treatment, there are already evaluations underway looking at the cost-effectiveness of internet-enabled therapies. But the suitability of these therapies across different patient groups and whether they can address the needs of those currently under-served by mental health care services, also needs to be considered.
Next steps for the Health Foundation
We plan to use pseudo-anonymised individual-level patient data to get a more detailed understanding how patients move through the IAPT programme. Given the health inequalities evident in the publicly available data, we want to build a more granular picture of how waiting times, symptom severity and treatment delivery mode might affect how patients progress through the pathway, and how this varies between different patient groups.
We also have another project underway with other sources of pseudo-anonymised health care data; we’re using GP-records linked to hospital data, via Clinical Practice Research Datalink (CPRD) to assess inequalities in access and outcomes for patients with mental health conditions using other NHS services. We are focusing on the impact of additional long-term conditions but will also consider the impact of deprivation as well.
We think this work to build the evidence base of current inequalities is key to ensuring that the push to expand access to mental health care services serves those who most need support.
Did you find this helpful? Any other topics would you like to see in the future? Share in the comments below or get in touch on Twitter.
Karen Hodgson is a Senior Data Analyst in the Data Analytics team at the Health Foundation. All code for generating the plots shown here is available on GitHub.
Footnotes
[1] Deprivation is measured using Indices of Multiple Deprivation (IMD, 2015). This is a measure which ranks each area in England in terms of relative deprivation, across 7 difference domains.
[2] In the NHS Digital IAPT annual data, 0.72% of all ended referrals have no IMD data available
[3] For full descriptions and variable definitions, see NHS Digital Annual Reports.
[4] For consistency, we report only on variables which are calculated based upon all referrals ending within the relevant year.
[5] The national standard for starting treatment is that 75% of new referrals should be seen within 6 weeks and 95% of patients should be seen within 18 weeks.
[6] Missing data is higher for these demographic groupings, which may result in biases. 13.72% of ethnicity data and 32.96% of sexual orientation data is not available.
