Background
Both type 2 diabetes and latent-type autoimmune diabetes are increasingly being diagnosed in children and young adults, obesity is becoming more prevalent in this group and in people with type 1 diabetes. Consequently, newly presenting hyperglycaemia in younger people is increasingly a diagnostic challenge. To date, there are no published estimates of the pre-test probability of each diagnosis in Emirati people who undergo screening with anti-GAD and anti-IA2 antibody titre.
Aims
We aimed to establish the prevalences of latent–type autoimmune diabetes in Emirati individuals presenting with hyperglycaemia below and above the age of thirty. A secondary aim was to determine whether body mass index (BMI) at presentation could provide useful diagnostic information in each age group.
Method
Imperial College London Diabetes Centre (ICLDC) is a specialist outpatient diabetes and general endocrine clinic with branches in Abu Dhabi and Al Ain.
We accessed all records where anti-GAD or anti-IA2 antibodies were reported at the initial consultation in Emirati people who presented to ICLDC between 01/2006 and 03/2017, were followed up for a minimum of two years and who were reviewed at ICLDC within the last year (n=12,449).
The diagnosis of latent-type autoimmune diabetes was made on the basis of anti-GAD titre of ≥10 IU/ml and recorded time to first insulin treatment of more than 6 months.
Results
167 individuals were identified as possible latent-type autoimmune diabetes according to our criteria.  In individuals presenting with hyperglycaemia under the age of 30 years, 35.9% were subsequently diagnosed with type 1 diabetes, 55.3% with type 2 diabetes, 3.2% with latent-type diabetes and 5.7% with other diabetes aetiology or prediabetes.  In individuals aged 30 years and over, 0.5% were diagnosed with type 1 diabetes, 93.6% with type 2 diabetes, 1% with latent-type diabetes, and 5.0% with other diabetes aetiology or prediabetes.
In Emirati individuals aged between 16 and 30 years, a BMI cut-off of 28.4 kg/m2 provided positive predictive value (PPV) of 0.77 and negative predictive value (NPV) of 0.77, with area under the ROC curve (AUC ROC) of 0.84 (0.823-0.858) for discrimination of type 2 diabetes from all other diagnoses (n=1,960). In individuals aged 30 years or above, a BMI cut-off of 27.5 kg/m2 provided PPV of 0.97 and NPV of 0.22 for type 2 diabetes versus other diabetes diagnosis (AUC ROC 0.809 (0.792-0.826), n=10622). 
In individuals diagnosed with either type 1 or latent-type autoimmune diabetes, a cut-off of age 25 years at diagnosis provided PPV of 0.58 and NPV of 0.96 (AUC ROC 0.927 (0.91-0.945)) for distinguishing between the two disorders.
Discussion
We present pre-test probabilities for each diabetes diagnosis in Emirati individuals undergoing anti-GAD and anti-IA2 screening as part of the work-up for hyperglycaemia. BMI remains a useful tool for distinguishing type 2 diabetes from other diabetes diagnoses in Emirati individuals over the age of 16 years. Although a cut-off of age ≥ 30 years is commonly used to distinguish latent-type autoimmune diabetes from type 1 diabetes, our data indicate that this may underestimate the prevalence of this type of diabetes in our population.

Presentation of latent-type autoimmune diabetes in Emirati people under the age of 30 years