Title of the local study protocol
A Real-World Database Study of Canagliflozin Utilization in Type 1 Diabetes Patients Over Time among European Countries: analysis of routinely collected health data from SIDIAP database
Rationale and background
This drug utilization study (DUS) is being conducted per the request of European Medicines Agency (EMA) Pharmacovigilance Risk Assessment Committee (PRAC) to investigate the utilization of canagliflozin for treatment of patients with type 1 diabetes mellitus (T1DM) over time in a European setting. The study will focus on European countries with high cumulative exposure of canagliflozin and adequate real-world databases for drug utilization evaluation, including the United Kingdom (UK), Spain, Italy, and Belgium.
Research question and objectives:
Research question: What is the proportion of patients with T1DM among canagliflozin users during the period from January 2016 to December 2022 in a European setting and does this proportion change over time?
Primary objectives: To describe the time-trend of canagliflozin utilization in patients with T1DM using real-world databases in European countries with high cumulative exposure, including the UK, Spain, Italy, and Belgium.
Exploratory objectives: To identify and characterize diabetic ketoacidosis (DKA) events that occur in patients with T1DM and in patients with type 2 diabetes mellitus (T2DM) after exposure to canagliflozin treatment and to describe the clinical characteristics of patients with T1DM treated with canagliflozin.
Study design: An observational, retrospective cohort study using secondary data from healthcare databases that reflect routine clinical practice in UK, Spain, Italy, and Belgium.
Population: In each database, patients eligible for this study must have 1) at least one canagliflozin exposure record during the period from January 2016 to December 2022, and 2) at least 6 months continuous observation before the first exposure to canagliflozin. Patients with T1DM who are treated with canagliflozin will be identified using two definitions.
The primary definition will identify canagliflozin users who meet the following criteria as patients with T1DM:
1. Have ?1 diagnosis of T1DM on or before the first exposure to canagliflozin, and
2. Have record of insulin therapy >90% of observation time from the first T1DM diagnosis until study end, and
3. Have no diagnosis of T2DM and no diagnosis of secondary diabetes mellitus on or after the first T1DM diagnosis, and
4. Absence of non-insulin antidiabetic treatment apart from SGLTi on or after the first T1DM diagnosis
The sensitivity definition will identify patients who meet the criteria 1-3 in the primary definition and an alternative to criterion 4 as follows:
4. Absence of non-insulin antidiabetic treatment apart from metformin and SGLTi on or after the first T1DM diagnosis
Canagliflozin users who do not meet the primary definition of T1DM must meet the following criteria to be defined as patients with T2DM:
? Have ?1 diagnosis of T2DM on or before the first exposure to canagliflozin, and
? Have no diagnosis of gestational diabetes and no diagnosis of secondary diabetes on or before the first exposure to canagliflozin.
Variables:
For patients with T1DM who are treated with canagliflozin, variables of canagliflozin utilization to be investigated will include drug ingredient, total number of prescriptions/dispensing, total days of supply, average days of supply per prescription/dispensing, dosage at the initial use, and days of continuous treatment after the first exposure. Events of DKA that occur after exposure to canagliflozin treatment in patients with T1DM as well as in patients with T2DM will be investigated as an exploratory outcome. The clinical characteristics of the canagliflozin users with T1DM, including demographics, history of medical conditions, procedures, and use of other prescription medications, may also be explored.
Data sources: A total of 5 electronic medical record (EMR) databases that have been assessed for feasibility will be used for this study. The Clinical Practice Research Datalink (CPRD) provides EMR collected from UK general practitioner (GP) offices using the Vision(R) software system (CPRD GOLD) and the EMIS Web(R) software system (CPRD Aurum) for public health research. The CPRD GOLD encompasses more than 19 million patients, and the CPRD Aurum encompasses more than 39 million patients (data as of January 2021). The IQVIA longitudinal patient databases (LPD) contain separate databases that gather EMR data from a panel of GPs by using the patient management software, covering approximately 1.1 million patients in Belgium (IQVIA-Belgium) and 2.2 million patients in Italy (IQVIA-Italy) (data as of 30 June 2020), and 1 million patients in Spain (IQVIA-Spain) (data as of 7 May 2020). The Information System for the Development and Research in Primary Care (SIDIAP) database comprises clinical information registered by primary healthcare professionals or related to primary care visits that are captured in EMR, covering more than 7.9 million patients in Catalonia, Spain (data as of 31 December 2020).
Study size: This is a descriptive study that will estimate the proportion of patients with T1DM among canagliflozin users and test the trend of the proportion over time. There is no a priori hypothesis testing and no pre specified requirement of sample sizes. The three feasibility assessments carried out in 2016, 2018, and 2020 showed a limited number of patients that meet a preliminary definition of T1DM among canagliflozin users. The feasibility of analysis and precision of statistical estimates that can be achieved in this study will vary by the availability of data and patient counts in each database.
Data analyses: Descriptive analysis will be conducted within each database, using the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) and standardized analytics when possible. For patients with T1DM who are treated with canagliflozin, using the 2 definitions separately, the utilization pattern of canagliflozin will be described. Continuous variables will be summarized using mean ± standard deviation, median, and interquartile range. Categorical variables will be summarized using counts and proportions. The unadjusted prevalence and incidence of patients with T1DM in canagliflozin users will be calculated by year as well as overall. The prevalence will be calculated as the number of canagliflozin users who are patients with T1DM divided by total number of all existing canagliflozin users. The incidence will be calculated as the number of new canagliflozin users who are patients with T1DM divided by the number of all new users, with new users defined as patients not previously treated with canagliflozin. Both the absolute and relative (as proportion) patient count will be presented for the estimates of prevalence and incidence. The linear trends of annual prevalence and incidence over the study period will be tested using a general linear regression model. If potential nonlinear trends are observed, alternative models for time-trend analysis such as the Joinpoint regression model will be considered. When there is sufficient homogeneity in time-trends across data sources, individual database-specific estimates will be combined per country as well as overall through meta-analysis. The characteristics of DKA events that occur in patients with T1DM as well as in patients with T2DM after the exposure to canagliflozin treatment will also be investigated and described, including the counts and incidence proportion of events, past history of DKA, time to the event occurrence, the number of exposure records before the event, and whether or not the event occurs during a continuous treatment era with canagliflozin. The clinical characteristics of the canagliflozin users with T1DM may also be described, if the proportion is higher than expected.
Keywords: Canagliflozin, Diabetes type 1, diabetic ketoacidosis, primary health care, SIDIAP,