Assessment, management, and outcomes of diabetes in severe mental illness
: investigations using electronic health records data resources

Student thesis: Doctoral ThesisDoctor of Philosophy



It is well established that people with severe mental illness (SMI; schizophreniform and bipolar disorders) have a significantly shorter life expectancy, and their risks of type 2 diabetes mellitus (T2DM) are also higher than expected. Relatively little research has evaluated outcomes of T2DM or its management in people with SMI, still less in a UK context. However, novel data resources, particularly linkages across healthcare sectors present opportunities to increase the evidence base.


A series of analyses were carried out based around SMI cases defined through the Clinical Record Interactive Search (CRIS) platform at the South London and Maudsley NHS Foundation Trust (SLAM) representing mental healthcare use in four boroughs of southeast London.

For the first study, a natural language processing (NLP) application was developed to improve ascertainment of recorded HbA1c levels, and the frequency of recorded levels was investigated in patients following newly diagnosed SMI, comparing this according to demographic and clinical factors.

The second study developed this further by investigating HbA1c recording in a series of cross-sectional samples of patients with SMI, additionally considering those with or without a recorded diabetes diagnosis.

The third study used a linkage between CRIS and Lambeth DataNet (LDN) data from all primary care services in one of SLAM’s catchment boroughs. Drawing information from all LDN patients with T2DM, those with new SMI diagnoses were compared to matched controls in relation to measures of diabetes care, glycaemic control, and outcomes. Associations with adverse outcomes were investigated further in relation to ethnicity and diagnostic subtype within the SMI cohort.

The fourth study used a linkage between CRIS and national Hospital Episode Statistics (HES) data. In patients with pre-existing SMI, those with newly recorded diabetes were ascertained and compared with matched controls in relation to hospitalisations for cardiovascular disorders and mortality. These associations and additional hospitalisations for diabetes-related adverse outcomes were investigated further in relation to ethnicity and diagnostic subtype within those with comorbid SMI and diabetes.


Study 1: Of 21,462 patients with SMI, HbA1c level was only recorded in 32.2%. Recording was higher in non-white ethnic groups, younger age groups, men, and those with schizophreniform (compared to bipolar) diagnoses. When recorded, mean HbA1c was higher in older age groups, men, non-white ethnic groups, and those with schizophreniform disorders.

Study 2: In the primary sample of 5046 patients with SMI in mid-2015, HbA1c levels were recorded in 22.8%, again associated with younger age, non-white ethnicity, and schizophreniform diagnoses. A relatively high proportion of the sample (8.8%) had recorded diabetes.

Study 3: In a cohort of 693 cases with SMI and T2DM compared to 3366 controls, age-adjusted all-cause mortality was increased 4.5-fold; however, of a range of outcomes investigated, only reduced likelihood of a foot examination and reduced likelihood of retinal screening were statistically significant.

Study 4: In 6107 cases with SMI and recorded diabetes compared to 12,407 controls, diabetes comorbidity was associated with indicators of more severe SMI and a higher risk of hospitalisations for cardiovascular disease or amputations, but not with higher age-adjusted mortality.


Comorbidity of SMI and diabetes was associated with adverse outcomes, although longer follow-up periods may be required to capture some inequalities. Suboptimal monitoring of dysglycaemia in mental health services and suboptimal healthcare receipt in primary care would be obvious targets for future interventions.

Date of Award1 Apr 2023
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorFiona Gaughran (Supervisor), Robert Stewart (Supervisor) & Brendon Stubbs (Supervisor)

Cite this