Networked Data Lab: Mental Health in Clinically Extremely Vulnerable People

The satellite analysis explores the mental health needs of all shielded patients in North West London. The satellite analysis is built on the central analysis with a special focus on mental health, and follows the same shielded patients definition period as the central analysis (detailed in Methods). Additionally, we explore the coding for suicide risk before and after the shielding period for the shielded patients’ cohort.

The North West London partner of the NDL held a PPIE workshop with 50 local community members to inform the first satellite analysis priorities. Of those who filled out a survey on demographic information, 40% of attendees were from ethnic minority backgrounds. The workshop focused on identifying key health and care research priorities for North West London citizens since the beginning of the COVID-19 pandemic that could be answered by analysing the Discover dataset, with a particular focus on aiming to reduce health inequalities. For example, areas of concern raised at the workshop included mental health, digital exclusion to health care and delays to treatment and diagnosis of other conditions.

The research topics from the workshop were reviewed by an analyst to exclude any that the Discover dataset could not be used to answer.

The seven topics of interest from the workshop included in the prioritisation exercise were:
1. Virtual consultations 2. Seldom-reached communities
3. Local availability of services
4. Diagnosis of other conditions
5. Worse COVID-19 outcomes for people of colour
6. Mental health
7. Social and community care

You can view the full report here.

In this study we use the longitudinal Discover dataset. This dataset provides linked coded
primary care, acute, mental health, community health and social care record for over 2.5
million patients who live and are registered with a GP in NWL. This dataset extracts data
from over 400 provider organisations including 360 GP practices, 2 mental health and 2
community trusts and all acute providers attended by NWL patients (in the form of
Secondary Uses Service (SUS) data). This dataset contains linked data from primary care,
secondary care, community, mental health, social care and high cost drugs.
We utilise the shielded patients list from the central analysis, along with primary care data on
diagnosis, through Read Codes and Long-Term Conditions (LTC) table in the dataset and
secondary care data on admissions.

755 patients who were shielding (0.76%) had a new recording of a suicide risk code following the introduction of shielding, compared to 323 patients who had evidence of risk before shielding started (0.32%). The majority of shielding patients (98.92%) had no evidence of suicide risk assessment before, during or after shielding.

The number of shielded patients with a recording of intentional self-harm during any of the shielding time periods is very low (0.24%). As it has not been compared to a control cohort we cannot report on its relative frequency in the population. Both the total number of patients and the monthly frequency of patients with recording of admission for intentional self-harm decreased during and after the shielding period. Although this data provides insight into intentional self-harm admissions, it only covers patients admitted to hospital and therefore does not cover every case of intentional self-harm. Full analysis of linked ONS data would provide more insight into intentional self-harm before, during and after shielding.

Age appears to play a large role in the shielding population as to whether patients have a LTC of anxiety, depression or mental health (serious mental illness), with the 50-59 year age category being most affected (OR = 4.18). Deprivation (IMD Decile) also has a stepwise impact on the mental health of the shielding population, with those in the most deprived decile having 1.66 times greater odds of having a LTC related to mental health recorded than the odds of those in least deprived areas. Frailty also has a significant impact on mental health, with those in the severely frail category having 2.06 times greater odds of having a LTC related to mental health recorded than those in the fit category. BMI results are mixed, and it would be valuable to explore impact of BMI as a binary variable (Healthy/Not healthy), as opposed to four BMI categories. Data are included in Figure 5 and Table 7.

Similarly to the overall mental health findings, the odds of shielding patients who were 50-59 years of age to have a recording of anxiety were 3 times greater than the odds of the reference category (<30 years of age). In contrast, an IMD Decile of 5 was used in this case as the reference category and a smaller step-wise change in odds from most- to least deprived was calculated for patients with anxiety. The odds of white patients having a record of anxiety was 1.8 greater than the odds of Asian or Asian British patients. Frailty was also an important factor in the recording of anxiety, with the odds of severely frail patients suffering from anxiety being 1.9 times greater than that of fit patients. BMI results are mixed, and it would be valuable to explore impact of BMI as a binary variable (Healthy/Not healthy), as opposed to four BMI categories. Data are included in Figure 6 and Table 8.

Similarly to the overall mental health findings, the odds of shielding patients who were 50-59 years of age to have a recording of depression were 5.5 times greater than the odds of the reference category (<30 years of age). Additionally, a step-wise decrease in odds from most- to least deprived was calculated for patients with depression, where in the most deprived areas (IMD Decile = 1) the odds of suffering from depression was 1.8 times that in the least deprived areas (IMD Decile = 10). The odds of white patients having a record of depression was 1.69 greater than the odds of Asian or Asian British patients. Frailty was also an important factor in the recording of depression, with the odds of severely frail patients suffering from depression being two times greater than that of fit patients. Suprisingly, the odds of a diagnosis of depression was not altered by differences in BMI, with overweight patients having equal odds as patients with a healthy BMI, contradicting previous reports (2). BMI results are mixed, and it would be valuable to explore impact of BMI as a binary variable (Healthy/Not healthy), as opposed to four BMI categories. Data are included in Figure 7 and Table 9.

Mirroring our findings on all mental health LTCs, the odds of shielding patients who were 50- 59 years old having serious mental illness were significantly greater, 4.5 times that of patients <30 years old. In terms of deprivation, there was a much steeper increase in the odds of developing serious mental illness in shielding patients from the most deprived group (OR = 3.9), compared to the least deprived group. In contrast to findings from anxiety and depression, the most affected ethnicity group for serious mental illness were the Black or Black British and mixed-race groups, whose odds were 1.64 times and 1.69 times greater than of Asian or Asian British patients. Data are included in Figure 8 and Table 10. It is important to note that the sample size of patients with a record of serious mental health issues was small (n = 3,202), therefore findings should be considered with caution as the resulting confidence intervals are wide.

The numbers of shielding patients with a LTC of anxiety and depression prior to the introduction of shielding were relatively high, at 17.5% and 21. 78% respectively. Records of long-term conditions related to mental health (serious mental illness) are low at 3.5%. Following the introduction of shielding, there was a <1% increase in the number of shielding patients having anxiety, depression or mental health (serious mental illness) diagnosis.

We have shared our key findings with with the NWL community through
community engagement workshops, reports to key stakeholders, and the Data Access
Committee meetings.