Model Design
Following the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) guidelines “Principles of Good Practice for Budget Impact Analysis,” a BIA was conducted using a model built in Microsoft Excel 2016 [16, 17]. This model evaluated the impact of introducing metformin SR, as an alternative to metformin IR, for the treatment of type 2 diabetes against a background of other antidiabetics (Fig. 1). Several studies showed comparable efficacy and safety for metformin SR when compared to metformin IR. Most of these studies, however, are considered “switch-studies,” meaning patients included in the metformin SR population have already experienced AEs from metformin IR. This leads to bias, which is why AEs were not included in this model [7, 10, 11, 13,14,15].
Figures 2 and 3 show the treatment pathway of patients who are treated following SoC, but are eligible for metformin SR, over a period of 3 years. Patients in Fig. 2 switched to metformin SR, and patients in Fig. 3 keep using metformin IR. Within these years, they could either keep using metformin or switch to other medications. Besides these metformin IR-intolerant patients, 67,249 newly diagnosed type 2 diabetes patients who initially started metformin SR treatment in year 1 were considered as well (Figs. 4, 5) [18, 19]. Guidelines from the Dutch College of General Practitioners (Nederlands Huisartsen Genootschap [NHG]) were used to outline the treatment pathway of type 2 diabetes patients [20]. Persistency was modeled based on a study that assessed the transition rates of patients with type 2 diabetes between different drugs (classes) (Figs. 2, 3, 4, 5) [21].
Following the ISPOR guidelines, we simulated the treatment pathway of each individual patient over a 3-year time horizon, based on annual persistence rates per type 2 diabetes drug category. Drug categories taken into consideration were (1) metformin, (2) other glucose-lowering drugs (OGLDs: sulphonylurea derivatives, thiazolidinediones, alpha-glucosidase inhibitors, glinides, dipeptidyl peptidase-4 [DPP-4] inhibitors, glucagon-like peptide-1 [GLP-1] agonists, sodium-glucose co-transporter-2 [SGLT-2] inhibitors) and (3) insulin. Following the distribution of patients through the treatment pathway in the model, the possibility of medication (non)adherence was assessed and corresponding probabilities of HbA1c goal achievement were included [22,23,24]. Studies in patients using metformin SR showed adherence rates ranging between 80 and 97.2% [14, 25,26,27]. In this model, we assumed adherence to metformin SR in year 1 as 90%, mostly related to the simplified dosing regimen, and adherence to metformin IR was approximately 72%. Although the effects of nonadherence on achieved HbA1c have been studied for many antidiabetic drugs, the relation is not quantified for all glucose-lowering drugs. Using pooled data on HbA1c target achievement, the drugs for which quantification was missing were graded under one common denominator: OLGD [24]. Based on the Dutch and international guidelines for type 2 diabetes treatment, the HbA1c target value was set to ≤ 7% (≤ 53 mmol/mol) [20, 28, 29]. Consequently, categories representing “goal” or “no goal” were set to HbA1c ≤ 7% (≤ 53 mmol/mol) or HbA1c > 7% (> 53 mmol/mol), respectively. Probabilities of developing type 2 diabetes related macro- and microvascular complications were linked to the obtained HbA1c levels per treatment (Table 1) [30].
Table 1 Event rates per 1000 person years for sufficient and non-sufficient HbA1c management [30] The analysis was conducted from a healthcare payers’ perspective. The influence of different parameters on the budget impact was assessed by univariate sensitivity analyses over a time horizon of 3 years, which was chosen to match the payers’ budgeting process [17].
Study Population
The baseline study population included in the model was obtained from Dutch metformin user numbers, as reported in the Dutch drug information system of the National Health Care Institute (GIPdatabank), and was estimated at 640,000 [31]. Patients on metformin IR dosages higher than 2000 mg daily were excluded, since switching from metformin IR to SR is not recommended when the daily dosage of metformin IR is above 2000 mg. Also the maximum recommended dosage for patients starting metformin SR from the start of type 2 diabetes treatment is 2000 mg [32, 33]. As metformin SR is an alternative to metformin IR, we identified which patients would be able to switch from SoC to metformin SR and who would continue on metformin IR. Factors considered in this respect were (1) the discontinuation of metformin IR amongst newly diagnosed patients (16.1%), (2) the potential market share of metformin SR as calculated from medicine issue registries from the United Kingdom (22.53%), and (3) the percentage of patients currently on metformin IR who could be actively switched to metformin SR (0.1%) [5]. As this BIA covers multiple years, type 2 diabetes patient population growth was taken into account. This was calculated by the number of newly diagnosed patients minus the number of deaths, where the number of newly diagnosed patients was estimated to be 67,000 by the Dutch healthcare research institute NIVEL (2016) [19, 30, 34]. Transition probabilities are displayed in the electronic supplementary material (ESM).
Data Collection
To represent the baseline use of metformin IR as accurately as possible, data were collected from three pharmacies representative of the Netherlands by pharmatech company Medstone [35]. These data were used to determine the average cost of metformin use in the Netherlands. The data were collected from one pharmacy situated in a city, one in a more rural area and one in a small village between August 1, 2017 and July 31, 2018 and using Anatomical Therapeutic Chemical (ATC) classification system code A10BA02. Further data handling is described in supplementary Fig. 7 (see the “Data handling metformin use” section in the ESM). Based on the summary of product characteristics (SmPC) recommendations, patients who used more than 2000 mg metformin IR were excluded from the model, leaving information on 847 prescriptions from unique patients. The obtained data included information on the total daily dosage, tablet breakdown and the number of prescriptions (Table 2).
Table 2 Metformin IR use and breakdown of the daily dosage Research has shown the daily dosages of both metformin formulations are compatible [36]. To convert the use of metformin IR to metformin SR, we assumed that the distribution of patients using certain tablet regimens was similar between both formulations. However, because metformin IR is available as 500 mg, 850 mg and 1000 mg and metformin SR as 500 mg, 750 mg and 1000 mg, some assumptions had to be made regarding the daily dosage of metformin SR (Table 3).
Table 3 Metformin IR versus metformin SR treatment regimens Costs
Costs are shown in Table 4 and the ESM (“Drug cost overview” section). Annual drug costs per patient per daily dosage (or insulin unit) were calculated using the prescription standard for type 2 diabetes from the National Health Care Institute (Zorginstituut Nederland [ZIN]) [37, 38]. The annual costs of metformin SR are calculated based on the assumption of a once-daily treatment regimen. The BIA only includes costs as paid by the healthcare payer, therefore excluding the patient’s own contribution to the total product price. Costs of insulin care and non-insulin care were based on information as provided by the Dutch Diabetes Foundation (Diabetes Vereniging Nederland) and was combined with treatment costs from health insurer declarations [39, 40]. Insulin care consisted of medication, testing strips, blood glucose measuring device, blood extraction equipment, insulin pens and syringes, remaining diabetes tools, an insulin pump and regular checkups/coaching at the ophthalmologist or with a diabetes nurse. Non-insulin care costs consisted of medication, regular checkups and coaching by a professional [39, 40]. All costs were inflated to the year 2018, using Dutch inflation rates [41]. For the BIA, discounting was not applied, as it was our goal to show financial streams at each selected budget period and not the net present value at the moment of decision-making [17].
Table 4 Costs included in the budget impact analysis, 2018 price levels