Real world evidence of insulin and biosimilar insulin therapy—Opportunities to improve adherence, outcomes and cost‐effectiveness

Abstract

Insulin has been discovered for more than a century; however, its benefits to people with diabetes are yet to be fully realized due to barriers related to access, quality of care and costs. Insulin therapy remains the cornerstone of diabetes management. The multicausality of diabetes and its subtypes calls for comprehensive phenotyping and use of biomarkers to ensure timely use of insulin to achieve early glycaemic control for long‐term benefits. Biosimilar insulins are biologic products that closely resemble originator insulins without significant differences in safety or efficacy. The lower investment costs needed for research and development make biosimilar insulin more affordable to improve access. While the efficacy of insulin products is proven in controlled settings, real world evidence (RWE) from real world data (RWD) plays a crucial role in assessing the safety, efficacy, cost‐effectiveness, adherence to and impacts of different insulin products, including biosimilars, on clinical outcomes. In this narrative review, we summarized the trends of insulin use and patterns of biosimilar insulin utilization in real world practice across different regions. We reviewed RWE on clinical safety, efficacy and cost‐effectiveness of biosimilar insulin, as well as therapeutic inertia and non‐adherence with insulin therapy. We also highlighted barriers to insulin adherence and enablers for persistence, along with potential solutions to promote the use of insulin and technologies for optimizing glycaemic control in different subtypes of diabetes. During our extensive literature review, we identified data gaps in the usage of biosimilar insulin in real world practice. We advocate for implementing a diabetes register designed fit‐for‐purpose, managed by a trained doctor–nurse team with system support, to generate RWE. By setting up registers with structured data collection, we can generate high quality data for integrative analysis with electronic health records (EHR) and health claims to evaluate the impacts of insulin products and other diabetes programmes on clinical outcomes, quality of life and healthcare costs to inform practice and policies.

Plain Language Summary

Diabetes affects approximately 10.5% of the global population and insulin is a vital treatment for diabetes management. Insulin was discovered more than a century ago, although its benefits to people with diabetes are yet to be fully realized due to barriers related to access, quality of care, and costs. Real‐world evidence from real‐world data plays a crucial role in assessing the safety, efficacy, cost‐effectiveness, adherence to, and impacts of different insulin products, including biosimilars, on clinical outcomes. In this publication, the authors provided a detailed review of the patterns of use and cost‐effectiveness of biosimilar insulin, and identified major data gaps. The authors explained the methodology, utility, and limitations of generating real‐world evidence based on real‐world data from sources such as registers, electronic health records and health claims for assessing treatment effectiveness and safety. The authors proposed the implementation of a purpose‐built diabetes register with structured data collection, managed by a trained doctor–nurse team with system support. These high‐quality data can be integrated with electronic health records and health claims for evaluation of interventions, including insulin on outcomes, quality of life, and costs to inform practice and policy. Based on these premises and available data, the authors summarized trends in insulin use including biosimilar insulin, and reviewed real‐world evidence on the safety, efficacy, and cost‐effectiveness of these products. They also identified barriers like therapeutic inertia and non‐adherence, discussed enablers for persistence, and proposed solutions to evaluate the impacts of insulin products and other diabetes programs on clinical outcomes, quality of life, and healthcare costs to inform practice and policies.

1. INTRODUCTION

Effective energy utilization and storage is the essence of life characterized by growth, metabolism and reproduction. There are many stress hormones that can break down glycogen, fat and protein to raise blood glucose and fatty acid levels to meet energy demand, especially during times of stress. On the other hand, insulin secreted from the beta cells in the pancreatic islets is the key hormone which promotes the entry of glucose and free fatty acids into peripheral cells for energy generation and storage as glycogen or fat in liver, muscle and adipose tissues. ¹

Each person is endowed with a unique number of islets at birth which is predetermined to a large extent by genetics. The expression and function of these islets increase progressively throughout childhood and plateau by early adulthood. ² These islets must be used in an effective manner throughout life course to maintain blood glucose within a narrow range. Diabetes is due to a myriad of causes including, but are not limited to, genetic, epigenetic, perinatal and autoimmune factors. External causes include metabolic stresses, toxins or injuries. Other causes are long disease duration and ageing. Many patients with diabetes (PwD) will eventually require insulin for effective use of energy substrates. There are now more than 10 classes of oral glucose‐lowering drugs (OGLDs) and injectables. As such, there is a need to emphasize the importance of insulin as a life‐saving drug, ensure its access and avoid undue delayed insulin initiation for people who need insulin, irrespective of age, race and socio‐economic status for controlling hyperglycaemia, preventing complications and improving quality of life.

The discovery of insulin in 1922 was followed by many technological advancements, including different insulin formulations and delivery systems aimed at mimicking the physiological minute‐to‐minute secretion of insulin and improving ease of administration. These technologies have markedly improved the effectiveness, safety and acceptability of insulin regimens. Some examples include human insulin analogues, insulin pens, continuous subcutaneous insulin infusion (CSII), automated insulin delivery (AID) system and continuous glucose monitoring (CGM) devices. In this context, the introduction of biosimilar insulin with lower acquisition costs than originator insulin comes with considerable cost‐savings, although therapeutic inertia, patient non‐adherence and system factors remain major barriers in our pursuit of improving diabetes care in people with type 2 diabetes (PwT2D), type 1 diabetes (PwT1D) and other subtypes of diabetes.

In this narrative review, we performed an extensive literature review regarding the patterns of use of biosimilar insulin and their cost‐effectiveness and identified major data gaps. To put this review in perspective, we first explained the methodology, utility and limitations of generating real world evidence (RWE) based on real world data (RWD) from sources such as registers, electronic health records (EHR) and health claims for assessing treatment effectiveness and safety. We proposed the implementation of a purpose‐built diabetes register with structured data collection, managed by a trained doctor–nurse team with system support. These high‐quality data can be integrated with EHR and health claims for the evaluation of interventions including insulin on outcomes, quality of life and costs to inform practice and policy. Based on these premises and available data, we summarized trends in insulin use including biosimilar insulin, and reviewed RWE on their safety, efficacy and cost‐effectiveness. We identified barriers like therapeutic inertia and non‐adherence, discussed enablers for persistence and proposed solutions to enhance insulin use and improve glycaemic control in various diabetes subtypes.

2. SETTING UP WELL‐IMPLEMENTED REGISTERS TO GENERATE RWE

In an article titled ‘Registers and registries—a review’ by J. M. Weddel in 1973, ³ the author explained the principal objective of setting up registers was to collate information collected from defined groups over time, which may be used in the prevention or treatment of disease, the provision of after‐care, the monitoring of changing patterns of disease and medical care and the evaluation and planning of services. To achieve this purpose, the data structure of the register must be predefined based on existing knowledge with systematic data collection and periodic review for decision‐making with continuous improvement.

Back in 1993, the International Diabetes Federation St Vincent Declaration ⁴ proposed the setting up of diabetes centres, diabetes teams, diabetes plans and diabetes registers to improve the quality of diabetes care and reduce complications. However, the implementation of a well‐designed register requires planning, resources and commitment. Given the silent, progressive, complex nature of diabetes and the highly preventable nature of its complications against a backdrop of a growing number of technologies and the large number of providers involved in the journey of living with diabetes, there is indeed an urgent need to register all patients with diabetes for quality assurance and demonstrate impacts.

Diabetes is a multisystem disease with complex interactions among multiple risk factors and complications which, in turn, can be modified by treatment and self‐care. Thus, all patients with diabetes should undergo regular structured assessment including eyes, feet, blood and urine at baseline and at least every 2 years to assess control of risk factors and complications, review the appropriateness of treatment, reinforce adherence and self‐care, document occurrence of significant events and patient‐reported outcomes (e.g., depression and quality of life) in order to evaluate the effectiveness, acceptability and safety of medications.

With the popularity of EHR, there is now a tendency for EHR to replace diabetes registers. Unless the template of a well‐designed register is included in the EHR implemented by a qualified team, the collection of key factors that determine outcomes can be random and incomplete, requiring considerable data cleaning during analysis to inform decisions. Registers are not only used for analysis purposes. The use of personalized data collected during structured assessment, at the point of care, can be used to improve health literacy, empower self‐care while informing timely clinical decisions and promote shared decision‐making between patients and providers. Implementation of a well‐designed register will also contribute to the capacity building of non‐medical staff to promote interdisciplinary care.

When linked to other EHR and health claims, especially hospitalization and death data, the register can provide objective evidence on changes in disease trajectories. The process of implementing well‐designed registers also empowers the healthcare teams, notably diabetes nurses and junior doctors, informs decision‐making by policymakers, identifies unmet needs and creates new knowledge. To this end, the Hong Kong Diabetes Register ⁵ set up in 1995 as a research‐driven quality improvement programme has led to many ramifications. ⁶ , ⁷ These include the implementation of a territory‐wide risk assessment and management programme and the web‐based Joint Asia Diabetes Evaluation (JADE) Program, which have benefitted over 0.5 million PwD and contributed to the 50%–70% decline in diabetes‐related complications and death rates in Hong Kong. ⁶ , ⁷ Meanwhile, the JADE Program has served as a prototype to inspire similar programmes in the region using registers to close the gaps in diabetes care, prevention, data and professional knowledge. ⁸

3. GATHER RWE TO VERIFY TREATMENT EFFECTIVENESS AND IDENTIFY CARE GAPS

Against this background, it is noteworthy that RWE is now increasingly used in clinical, cost‐effectiveness and regulatory reviews of safety, efficacy and quality. The US Food and Drug Administration (FDA) now considers RWE as part of the evidence package for submissions seeking authorization to market new medical products, including licence applications of biologics. ⁹ Clinical evidence regarding the usage and potential benefits or risks of a medical product based on the analysis of RWD from various sources, including disease registers, EHRs, administrative claims and digital health technologies, is considered as RWE. ⁹ , ¹⁰ These RWE complement the evidence on medication efficacy obtained from randomized controlled trials (RCTs) and play a critical role in decision‐making. ¹⁰ These RWD can also be used to monitor and assess patterns of treatment and healthcare delivery in routine practice. Figure 1 illustrates the journey of gathering RWD and RWE from data sources with processing and analysis within a secure data environment (SDE) within the five Safes framework (safe data, safe projects, safe people, safe settings and safe outputs). ¹¹ This RWE can be used by multiple parties including research communities, healthcare systems and industry to promote continuous assessment and learning to guide decision‐making.

FIGURE 1.

Journey of real world data (RWD) and real world evidence (RWE) gathered from various data sources with processing and analysis within a secure data environment (SDE), used by multiple parties including healthcare systems (patients, providers, payors), research communities and industry to promote continuous assessment and learning to guide decision‐making. AI, artificial intelligence; ML, machine learning; RCT, randomized controlled trial.

4. USING ROBUST METHODOLOGIES TO CREATE RWE

Studies from RWE should follow a structured approach to generate valuable insights into treatment effectiveness, safety, cost‐effectiveness and adherence in a real world context, as well as identify gaps in diabetes treatment and management. ¹⁰ Figure 2 shows the main applications of RWD for generating RWE in diabetes.

FIGURE 2.

Main applications of real world data (RWD) for generating real world evidence (RWE) in diabetes. EHR, electronic health record; RCT, randomized controlled trials.

Most studies from RWE compare treatment effectiveness by leveraging RWD such as registers, EHR and administrative claims to emulate target RCTs. Regulatory agencies increasingly use RWE to assess medication safety, especially for rare adverse drug events (ADEs) not revealed by RCTs. These RWE are used to support drug approvals, clinical guidelines and prescribing decisions. Other utilities include the analysis of drug utilization trends and patterns, monitoring the incidence of diabetes and its complications, estimation of healthcare costs, evaluation of health policies and the development of clinical decision support (CDS) systems for personalized diabetes prevention and treatment. ¹⁰ RWE can also identify trends among subgroups and regions, facilitating the targeting of populations and the generation of new hypotheses. In this light, using RWE can significantly enhance regulatory decision‐making and clinical practice in the biosimilar market.

For example, Hong Kong, a cosmopolitan city with 7.5 million people and universal health coverage through the government‐funded Hospital Authority (HA), has built one of the world's largest integrated longitudinal EHR systems. ¹² Over 90% of people with chronic diseases such as diabetes use the HA services. Since 1999, the HA has developed the territory‐wide EHR system to provide a repository of all clinical data collected within the HA facilities. Our team curated the Hong Kong Diabetes Surveillance Database (HKDSD) ¹³ from the HA‐EHR system with a built‐in module for diabetes risk assessment and management based on the HKDR protocol. ⁵ Using this data resource, we have performed extensive analysis to track trends, identify unmet needs and verify interventions. We reported GLDs utilization trends and patterns including insulin and insulin analogues, severe hypoglycaemia incidence and identified young‐onset T2D (YOD) with poor glycaemic control for targeted intervention. ¹⁴ , ¹⁵ We also observed rising incidence of YOD (20–40 years old), ¹⁶ trends of diabetes‐related complications and hospitalization burden. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹

Leveraging the serial laboratory values in EHR, we applied propensity score‐based methods to emulate RCTs and compared treatment strategies. ²² , ²³ , ²⁴ For instance, in PwT2D treated mainly with sulphonylureas and metformin, early addition of dipeptidyl‐peptidase 4 inhibitor (DPP‐4i) within the first 2 years of diagnosis (vs. 2–5 years of diagnosis) reduced HbA1c variability and delayed insulin requirements. ²⁴ Intensified treatment with DPP4i at HbA1c < 7.5% (vs. HbA1c ≥ 7.5%) reduced the incidence of insulin initiation, severe hypoglycaemia, cardiovascular disease and end‐stage kidney disease in part due to reduced HbA1c variability. ²⁵ We also developed a machine learning model to predict the 1‐year risk of severe hypoglycaemia. ²⁶ Insulin use was identified as a top predictor of severe hypoglycaemia in older adults. This EHR model can be seamlessly integrated into EHR systems in real time, serving as decision support to prioritize and personalize interventions.

However, many RWD are products of convenience and not designed for research purposes. Apart from data quality, robust study designs and methodologies are essential to address selection bias, time‐related bias and confounders. ²⁷ Standardizing the structure of databases with registration for reporting of RWE studies will enhance transparency, reliability and reproducibility. ¹⁰ , ²⁷ Using multilevel methodologies to create a target RCT emulating a landmark study with adjustment for propensity score, time‐varying exposure, ¹⁴ , ¹⁵ inclusion of negative outcome and exposure controls and performing sensitivity analyses in stratified sub‐cohorts would improve causal inference. ²² , ²³ These factors underscore the need for well‐implemented disease registers to create robust RWE to identify care gaps and guide clinical decision‐making to improve diabetes prevention and management. In the following sections, we shall discuss our current knowledge on the RWE regarding the use of insulin including biosimilar insulin.

5. TRENDS AND PATTERNS IN INSULIN USE IN REAL WORLD PRACTICE

5.1. Trends of insulin use

A growing body of research has examined the trends and patterns of insulin therapy in real world practice over the past decades. We searched PubMed using the terms ‘insulin’, ‘diabetes’, ‘drug’, ‘medication’, ‘trend’, ‘pattern’, ‘real‐world’ from original articles published up to 1 January 2025. In the present study, we summarized studies reporting the trends of insulin use across different regions based on registers, EHR, health claims and national health surveys. Table 1 summarizes the key characteristics and findings of 12 studies that reported changes in the proportion of insulin users over specific calendar years or time periods.

TABLE 1.

Insulin use trend in patients with diabetes reported by real world studies.

Study ID	Profile	Insulin class (group)	Proportion of insulin use (%) by year (2000–2022)
			<00	00	01	02	03	04	05	06	07	08	09	10	11	12	13	14	15	16	17	18	19	20	21	22
Yokoyama 2022 29	Region: Japan DM: T2D Age (mean): 62.7–66.8 Sample: 17 392–53 796 Scope: multicenter Data: EHR	Among all patients
		Insulin only				15.0		14.0		13.2		12.2		10.8		8.6		6.9		4.5		3.6
		Insulin + non‐insulin GLD				8.1		9.3		10.5		11.0		12.8		12.9		14.6		14.4		15.1
		Among patients with insulin treatment
		Human insulin—NPH				38.5		32.5		27.9		16.9		10.5		4.8		3.0		1.4		0.6
		Human insulin—regular				22.8		18.3		11.3		6.6		4.2		2.7		1.8		1.5		0.9
		Human insulin—mixed				55.2		41.5		25.9		17.3		10.7		5.0		3.4		3.0		1.4
		Analogue insulin—long‐acting				0.0		8.1		12.4		22.0		34.7		49.7		62.9		69.5		65.5
		Analogue insulin—prandial (fast‐acting)				14.7		28.2		36.1		38.2		40.7		42.3		46.4		41.5		40.6
		Analogue insulin—premixed				0.0		6.5		22.9		33.9		35.8		37.0		27.7		23.8		30.0
Selvin 2015 28	Region: US DM Age: ≥20 Sample: 4947 Scope: Nationwide Data: Survey		88–93					99–04								05–12
		Insulin‐only	26.8					15.0								14.1
		Insulin + OAD	3.5					9.8								14.9
Yang 2022 15	Region: Hong Kong DM Age: ≥20 years (mean) 64.6, 67.8 years Sample: 956 748 Scope: Territory‐wide Data: EHR	Insulin				11.2																	13.6
		Human insulin				11.2	10.5	9.4	9.4	9.2	9.6	9.8	9.9	9.9	10.2	10.5	10.5	10.7	11.0	11.1	11.3	11.6	11.3
		Analogue insulin				0.2	0.2	0.2	0.2	0.3	0.3	0.4	0.5	0.6	0.9	1.2	1.5	1.7	2.0	2.4	2.6	3.0	3.5
Shin 2021 35	Region: US DM: T2D Age (mean): 58.9–59.6 years (Clinformatics) 73.1–73.2 (Medicare) Sample: 264 542 (Clinformatics) 285 213 (Medicare) Scope: Nationwide; Data: Claims	Insulin a (Clinformatics database)															5.0	4.7	4.6	3.9	3.7	4.6
		Insulin a (Medicare database)															4.4	3.8	3.3	3.0	3.0
		Insulin a (MarketScan database)															4.3	4.0	3.8	3.6	3.3	3.6
Montvida 2018 31	Region: US; DM: T2D; Age (mean): 58 years (1st‐line initiation) 59 years (2nd‐line initiation) Sample: 1 023 340 (1st‐line) 357 482 (2nd‐line) Scope: Nationwide Data: EHR	Insulin (1st‐line initiation)							8	8	8	8	8	9	10	10	9	9	9	10
		Insulin (2nd‐line initiation)							7	8	9	10	11	12	14	15	16	16	16	17
Nair 2022 36	Region: US; DM: T2D; Age: 50–89 years; Sample (with ASCVD/HF): 177 254 Scope: Regional Data: Claims	Insulin (age ≥65)																	17.1	15.9	14.7	14.2	15.0
		Insulin (age 50–64)																	27.3	25.5	24.4	24.5	26.3
		Insulin (overall, estimated)																	18.3	17.1	15.8	15.2	16.2
Foresta 2023 34	Region: Italy; DM; Age: 65–90 years Sample: 251 737–308 372 Scope: Regional Data: Pharmacy	Insulin (age 65–70)												18.7					21.3						22.5
		Insulin (age 71–75)												19.7					22.4						22.1
		Insulin (age 76–80)												21.3					24.1						23.1
		Insulin (age >80)												21.6					26.4						25.1
		Insulin (overall, estimated)												20.2					23.4						23.3
Neyer 2024 42	Region: Austria DM: T2D Age (mean): 63.7–69.6 years Sample (with high cardiovascular risk): 672 (Period 99–00) 1005 (Period 05–08) 481 (Period 22–23) Scope: single centre Data: EHR			99–00								05–08														22–23
		Fast‐acting insulin		5.0								7.5														5.3
		Intermediate insulin		4.1								6.0														1.0
		Long‐acting insulin		4.1								4.5														7.8
		Intermediate‐/long‐acting + fast‐acting insulin		19.8								7.8														1.0
Christensen 2016 33	Region: Denmark General population Age: NA Sample: ~5.6 million Scope: Nationwide Data: Pharmacy	Annual prevalence of users per 1000 inhabitants
		Insulin	7.6	8.0	8.3	8.8	9.3	9.9	10.4	11.0	11.5	11.9	12.2	12.5	12.6	12.9	13.2	13.5
		Fast‐acting insulin	4.1	4.3	4.3	4.5	4.7	4.9	5.1	5.3	5.5	5.7	5.8	6.0	6.1	6.3	6.5	6.7
		Intermediate‐acting insulin	6.3	6.6	6.6	7.0	7.3	7.3	7.1	6.7	6.5	5.9	5.1	4.6	4.2	4.0	3.9	3.8
		Long‐acting insulin	0	0	‐	‐	‐	0.5	1.3	1.8	2.3	3.4	4.4	5.0	5.5	6.0	6.4	6.9
		Intermediate‐/long‐acting + fast‐acting insulin	2.3	2.5	2.6	2.8	3.1	3.5	3.7	4.0	4.3	4.2	3.9	3.7	3.4	3.2	3.1	2.9
Zhang 2024 43	Region: China DM: T2D Age (mean): 60.5–61.6 years Sample: 123 525‐213 287 Scope: Multicentre Data: EHR	Basal insulin																	11.6	11.9	12.4	12.8	12.6
		Bolus insulin																	9.6	9.8	9.9	9.8	10.1
		Premixed insulin																	22.7	21.3	21.4	18.8	17.5
		Basal + bolus insulin																	6.0	6.6	7.7	8.3	8.7
Perez‐Nieves 2022 b , 44	Region: US DM: T1D, T2D Age (mean): 35.9–38.3 (T1D) 56.8–58.6 (T2D) Sample: 77 504–115 274 (T1D) 274 872–457 591 (T2D) Scope: Nationwide Data: Claims	Patients with T1DM
		Basal only											5.4									3.7
		Short/rapid only											40.8				46.7			47.3		48.9
		Premixed											5.4									1.1
		Basal + bolus insulin											48.9				46.7			47.3		46.5
		Concentrated insulin											0.2									6.1
		Patients with T2DM
		Basal only											38.3									48.3
		Short/rapid only											6.3									7.4
		Premixed											19.2									6.6
		Basal + bolus insulin											36.1									36.1
		Concentrated insulin											0.7									18.4
Sarkar 2021 39	Region: US DM: T2D Age: ≥35 years Sample: 105 343–141 127 insulin treatment visits Scope: National representative Data: Physician‐reported data	Human insulin																		7.3	6.0	7.3	8.2	5.5
		Analogue insulin																		92.7	91.4	86.8	84.2	86.3
		Biosimilar insulin																		0.0	2.6	5.9	7.6	8.2
		Intermediate‐/short‐acting human insulin																		3.7				2.6
		Rapid‐acting insulin																		21.2				16.5
		Long‐acting insulin																		67.3				74.8
		Premixed insulin																		7.7				6.0

Abbreviations: DM, diabetes; EHR, electronic health records; GLD, glucose‐lowering drugs; NA, not available; NPH, Human Protamine Hagedorn; OAD, oral antidiabetic drugs; T1D, type 1 diabetes; T2D, type 2 diabetes.

^a Insulin use data extracted for the second quarter (Q2) of each year as reported in the study by Shin et al. (2021).

^b Insulin use estimates derived from figures presented in the study by Perez‐Nieves et al. (2022).

One notable trend is the shift from insulin monotherapy to combination therapy with OGLDs. ²⁸ , ²⁹ A study using nationwide EHR in Japan reported a decline in insulin monotherapy from 15.0% in 2002 to 3.6% in 2018 among PwT2D, while combination therapy increased from 8.1% to 15.1% during the same period. ²⁹ In an analysis of the US National Health and Nutrition Examination Survey (NHANES) which included a nationally representative sample of adults with diabetes, insulin monotherapy use declined from 26.8% in 1988–1994 to 14.1% in 2005–2012, alongside an increase in combination therapy from 3.5% to 14.9%. ²⁸

According to practice guidelines, regardless of background glucose‐lowering therapy or disease stage, insulin should be used early in patients with severe symptoms of hyperglycaemia or when HbA1c or blood glucose levels are high (i.e., HbA1c > 10% or blood glucose ≥16.7 mmol/L). ³⁰ In a study using nationwide EHR data from the United States (2005–2016), insulin use as a first‐line therapy remained stable (8%–10%), while its use as a second‐line treatment increased from 7% to 17%. ³¹ However, there are considerable variations in insulin use across different regions. In Japan, insulin use in PwT2D declined from 23.1% in 2002 to 18.7% in 2018, ²⁹ while in Hong Kong, insulin use increased from 11.2% in 2002 to 13.6% in 2019. ¹⁵ In Austria, insulin use in PwT2D remained stable at around 34.6% between 2012 and 2018. ³² In the United States, nationwide claims data showed stable insulin use among adult PwT2D (3%–5%) from 2013 to 2018. ³³ Similarly, regional claims data reported stable insulin use in PwT2D with cardiovascular diseases aged 50–64 years (range: 24%–27%) and ≥65 years (range: 14%–17%) from 2015 to 2019. ³⁴ In Italy, an upward trend of insulin use was observed in older PwT2D aged ≥65 years with an increase from 20.2% in 2010 to 23.3% in 2021, with the highest usage observed in those aged >80 years (25.1% in 2021). ³⁵ In a nationwide pharmacy prescription survey in Denmark, insulin use in the general population increased from 7.6% in 1999 to 13.5% in 2014. ³⁶

Insulin analogues are human insulin with different amino acid sequences or fatty acid chains with altered pharmacokinetic profiles. The ultra‐short‐acting insulin analogue has a more rapid onset of action targeting postprandial blood glucose, while the long‐acting insulin analogue has attenuated peak action but a longer duration of action to improve fasting blood glucose. ³⁷ , ³⁸ In many high‐income areas, the increasing use of insulin analogues is accompanied by a declining use of human insulin. ¹⁵ , ²⁹ , ³⁹ In the aforementioned Japanese study, the use of neutral protamine Hagedorn (NPH) insulin (38.5% in 2002 to 0.6% in 2018) and regular insulin (22.8% in 2002 to 0.9% in 2018) had dropped as fast‐ and long‐acting insulin analogues became popular. In Hong Kong, in a territory‐wide EHR analysis of 0.9 million adults with diabetes, the use of insulin analogues had increased from 0.2% in 2002 to 3.5% in 2019, while human insulin remained stable at 11%. ¹⁵ During the same period, HbA1c had dropped from 7.8% to 7.4%, coinciding with the introduction of a territory‐wide diabetes risk assessment and management programme as well as other new GLDs. In Colombia, a study using pharmacy prescription data reported that among PwD prescribed insulin, the proportion of human insulin users decreased from 89.6% in 2011 to 37.4% in 2015, while that of analogue users increased from 10.4% to 62.6%. ⁴⁰

The replacement of intermediate‐acting or premixed human insulin with long‐acting insulin analogues as basal insulin aims to suppress nocturnal hepatic glucose production and achieve near‐normal fasting blood glucose levels for optimizing the effects of other OGLDs. ⁴¹ In a US study using nationally representative physician‐reported data (IQVIA National Disease and Therapeutic Index), by 2020, long‐acting insulin analogues were the most frequently prescribed formulations during all insulin‐related treatment visits (52.6% for insulin glargine and 17.4% for insulin degludec), while the use of human insulin was minimal (0.6% for NPH and 2.2% for regular insulin). In an Austrian study analysing sequential EHR data of PwT2D, the use of long‐acting analogues increased from 4.1% in 1999–2002 to 7.8% in 2022–2023, while intermediate‐acting human insulin declined from 4.1% to 1.0%. In Denmark, the use of long‐acting insulin analogue in PwD increased from 0.0% in 1999 to 6.9% in 2014, while that of intermediate‐acting human insulin dropped from 6.3% to 3.8%. ³⁶

The development of insulin analogues has provided the basis for a basal‐bolus regimen to mimic the physiological insulin profile for avoiding severe hyper‐ and hypo‐glycaemia. In China, a study using provincial‐level EHR data reported stable use of basal insulin (around 12%) among PwT2D in 2015–2019, while the use of the basal‐bolus insulin regimen increased from 6.0% to 8.7%. Longitudinal claims data from the United States revealed distinct patterns for PwT1D and PwT2D. Between 2009 and 2018, among PwT1D, the use of the basal‐bolus insulin regimen was common and stable (45%–48%), while basal‐only (5.4%–3.7%) and premixed insulin (5.4%–1.1%) had declined. In PwT2D, the use of basal‐insulin‐only increased (38.3%–48.3%), while the basal‐bolus insulin regimen remained steady at 36.1% and that of short/rapid insulin alone was uncommon (6.3%–7.4%). Likewise, the use of premixed human insulin showed a consistent decline across regions, including the United States, ³⁹ , ⁴² Austria ⁴⁴ and China. ⁴⁴

Concentrated insulin formulation (e.g., Glargine‐300 with 300 units in 1 mL) enables the administration of high‐dose insulin in smaller volume. In patients with severe insulin resistance or obesity, this has allowed administration of smaller volume of insulin with a better absorption profile and lower risk of hypoglycaemia compared to the standard formulation with 100 units in 1 mL (U100). ⁴⁵ In a US claims‐based study, concentrated insulin use among PwT1D had increased from 0.2% in 2009 to 6.1% in 2018, and among PwT2D, this has increased from 0.7% to 18.4%. ⁴²

5.2. Pattern of use of biosimilar insulin

Biosimilar insulin has the same primary or highly similar secondary and tertiary structure as the originator insulin with different excipients and formulations. These insulin products leverage the available evidence on the safety and efficacy of the originator insulin. While this difference might lead to subtle variations in pharmacokinetic, pharmacodynamic and antigenicity profiles, the considerably lower cost of biosimilar insulin has made this a more affordable option with increasing popularity in some regions. Of note, although animal insulin (bovine or porcine) has largely been replaced by human insulin, ⁴⁶ they are still manufactured and prescribed in some areas including the United Kingdom. Depending on the source of originator insulin, some people are not allowed to use animal‐derived products due to religious reasons. ⁴⁷ Thus, the source of originator insulin needs to be specified in biosimilar insulin to ensure its acceptability in some communities.

Table 2 summarizes key findings from reports on the use of biosimilar insulin in real world studies. Physician‐reported data from the United States showed that biosimilar insulin use in PwT2D had increased from no recorded use in 2016 to 8.2% among all insulin‐related treatments by 2020. This rise coincided with declining use of both human insulin and originator insulin analogues. ³⁹ In another analysis of US health claims involving adults using glargine insulin, the use of biosimilar insulin increased from 8.2% in 2017 to 24.8% in 2020. ⁴⁸ This increasing trend was observed in both PwT1D and PwT2D. By 2020, 17.1% of PwT1D and 25.7% of PwT2D were prescribed biosimilar insulin glargine, reflecting its gradual acceptance since its introduction in 2016. ⁴⁸

TABLE 2.

Summary of real world biosimilar use patterns.

Study	Region	Data source	Biosimilar insulin use
Godman 2021 51	Bangladesh	Pharmacy dispensing data + hospital utilization/procurement data	In 2020, long‐acting analogues was the dominant insulin dispensed in nearly half of the community pharmacies. In 2020, biosimilars were the principal insulin glargine dispensed in over half of these surveyed pharmacies. 12.7% of the surveyed pharmacies reported dispensing rate of over 80% for biosimilars among insulin glargine preparations.
	India	Pharmacy dispensing data, + hospital utilization/procurement data	Biosimilar insulin glargine (Glaritus) is typically the only insulin glargine 100 IU/mL dispensed. The prescription of long‐acting insulin analogues (insulin glargine) has increased in 2014–2019 and decreased in 2019–2020 in five government hospitals, as measured by total defined daily doses (DDD).
	South Korea	Commercial medicine utilization data	There was generally low but increasing use of biosimilar insulin glargine, from 0.95% of total insulin glargine preparation in 2017 to 4.7% in 2019, with further growth expected.
	Malaysia	Pharmacy dispensing data	Biosimilars accounted for up to 90% of long‐acting insulin analogues dispensed in government hospitals.
	Pakistan	Pharmacy dispensing data + commercial medicine utilization data	In 2019–2020, there was limited utilization of biosimilar insulin glargine, accounting for only 3.8% of 100 IU/mL packs dispensed.
	Japan	Claims	The uptake of biosimilars is typically lower than high‐income countries. The proportion of expenditure on generic/biosimilars increased from 1.24% in 2015 to 4.81% in 2017.
Gani 2018 52	Four Asian countries	Survey (patient‐reported data)	By 2014, biosimilar insulin was used in four (India, Philippines, China, and Vietnam) of 11 countries/regions participating in the Joint Asia Diabetes Evaluation (JADE) Register (China, Hong Kong, India, Indonesia, Korea, Malaysia, Philippines, Singapore, Taiwan, Thailand and Vietnam). Among patients enrolled in 2007–2014 from four countries with records of biosimilar insulin, prescription rate of biosimilar insulin was 719 (7.1%) versus originator insulin (9130, 89.5%). Among the biosimilar insulin‐treated patients (n = 719), 58.4% were on the premixed regimen, 20.6% were on the basal‐only regimen, 13.6% on basal‐plus or basal‐bolus and 1.8% on bolus only.
	China		Prescription of insulin: 2.9% biosimilar, 97.1% originator
	India		Prescription of insulin: 9.6% biosimilar, 90.4% originator
	Philippines		Prescription of insulin: 7.0% biosimilar, 93.0% originator
	Vietnam		Prescription of insulin: 5.1% biosimilar, 94.9% originator
Tachkov 2021 53	Bulgaria	Claims	In the middle of 2020, the average market share of biosimilar versus the total insulin market was 4% in the EU market. During 2014–2020, only two trademarked biosimilar insulins were approved and available in Bulgarian market. Due to the slow entrance of biosimilar insulin, the variety of different insulin types remained unchanged with relatively stable medicine prices. Limited competition of biosimilars did not notably impact overall insulin utilization during 2014–2020.
Rai 2023 48	US	Claims	Among all adult insulin glargine users identified during 2011–2021, the proportions of biosimilar insulin users (Basaglar) were 9.1% (n = 7070) for T1D and 11.4% (n = 56 129) for T2D. In all insulin glargine, biosimilar insulin use increased from 8.2% in 2017 to 24.8% in 2020. Biosimilar insulin glargine has steadily increased in adoption since its introduction in 2016, reaching 16.0% usage in T1D and 25.0% usage in T2D by 2020. Prior to 2016, originator insulin glargine dominated the market with nearly 100% usage for both types of diabetes, but its share has gradually declined as biosimilars gained acceptance.
Fisher 2022 50	Canada	Claims, combined with pharmacy dispensing data	The mandatory switch policy from originator to biosimilar insulin glargine was launched on May 27, 2019, with PharmaCare coverage becoming exclusive to biosimilar insulin glargine after 25 November 2019. The proportion of biosimilar insulin glargine in all insulin glargine increased in both PharmaCare‐covered prescriptions (8.9% to 99.2%) and overall British Columbia prescriptions (9.1% to 79.0%) following the policy implementation. Among all patients required to switch (n = 15 344), 78.5% successfully transitioned to biosimilar within 1 year; 21.5% did not transition; 2.8% switched back to originator after trying biosimilar insulin.
McClean 2022 49	Canada	Procurement data	During May–November 2019, a mandatory switch to biosimilar insulin glargine was implemented in British Columbia to maintain provincial drug coverage, achieving approximately 90% uptake by the end of the phase‐in period. New Brunswick, Nova Scotia, and the Prince Edward Island (PEI) made insulin glargine a full benefit in fall 2017 while requiring special authorization for the originator drug, leading to a substantial increase in biosimilar purchases. Saskatchewan and Ontario also introduced biosimilar insulin glargine as a full benefit around the same time but did not restrict access to the originator drug, resulting in no significant increase in uptake of biosimilar insulin. The national adoption rate reached about 65% by the end of 2020, with notable variations across provinces due to different healthcare policies and implementation strategies.

In Canada, policy interventions have played an important role in driving uptake of biosimilar medications including insulin. ⁴⁹ The implementation of a mandatory switch policy in British Columbia in 2019 led to a near‐complete transition to biosimilar insulin glargine within 6 months, with PharmaCare‐covered prescriptions rising from 8.9% to 99.2%. ⁵⁰ While this shows the potential impact of strong policy measures, adoption rates vary considerably across provinces. In regions implementing less stringent policies, the uptake of biosimilar insulin glargine was only modest although it had been anticipated that the national adoption rate of biosimilar insulin glargine would approach 65% by 2020. ⁴⁹

In a study based on published data from six Asian countries, while there was high usage of long‐acting biosimilar insulin in India and Malaysia, it only accounted for 4.9% of the insulin market in South Korea in 2019 and 3.8% in Pakistan in 2020. ⁵¹ In the JADE Program, ⁵ which recruited PwT2D from 300 clinics in 2007–2014, among the four countries where biosimilar insulin was available (India, Philippines, China and Vietnam), the prescription proportion of biosimilar insulin among all insulin users was 7.1%, ranging from 2.9% in China to 9.6% in India. ⁵²

Compared to the United States, Europe had a more conservative uptake of biosimilar insulins. In Bulgaria, only two trademarked biosimilar insulins were approved in 2014–2020 with minimal market share of insulin. ⁵³ Among countries in the European Union (EU), the average use of biosimilar insulins accounted for only 4% of the total insulin market by 2020 due to slow adoption in some regions. ⁵⁴ An analysis of health authority databases across Europe reported the continued dominant use of long‐acting insulin analogues despite the availability of biosimilar insulin. ⁵⁵ This preference might reflect the perception by doctors that despite their higher costs, originator insulin analogues might be superior to biosimilar insulins in terms of efficacy and safety. ⁵⁵

Taken together, RWE revealed several key transitions in insulin therapy, including the shift from monotherapy to combination therapy, the increasing adoption of long‐acting insulin analogues and the emergence of biosimilar insulins as a cost‐saving alternative. These trends were mainly observed in PwT2D, although they were also evident in PwT1D. Comparisons of utilization patterns across studies are challenging due to differences in data sources, healthcare systems and prevalence of diabetes. However, the overall evidence points to marked regional variations in insulin use, particularly in the adoption of biosimilars. This variation is highly context‐dependent, influenced by regulatory measures, clinical perceptions and market competition. Most importantly, these RWD came from regions where insulin trends were well documented, such as North America, Europe and Asia. Considering that 80% of PwD live in low‐ and middle‐income countries, ⁵⁶ such as those from Africa and South America, the patterns of insulin use and their implications, including disparities, effectiveness and safety, are largely unknown. These huge data gaps call for the implementation of large‐scale registers and databases to provide RWE in these under‐represented regions.

6. CLINICAL SAFETY, EFFICACY AND COST‐EFFECTIVENESS OF BIOSIMILAR INSULIN

In RCTs, the clinical safety and efficacy of biosimilar insulin are comparable to originator insulin products. In a meta‐analysis of 14 RCTs ⁵⁷ involving 6188 PwT1D or PwT2D, long‐ or short‐acting biosimilar insulin had similar efficacy to originator products in terms of lowering HbA1c or fasting glucose at 26 or 52 weeks with no increase in incidences of hypoglycaemia, severe hypoglycaemia or anti‐insulin antibodies. However, compared to originator insulin, RWE on efficacy and safety of biosimilar insulin is limited, possibly due to low uptake and lack of novelty from the researchers' perspectives.

There are also data gaps regarding the administration of biosimilar insulin, such as the performance of biosimilar insulin pens. It is also unclear whether biosimilar insulin cartridges are compatible with existing originator insulin pens. ⁵⁸ Nevertheless, given their comparable clinical efficacy and safety profiles to originator insulin, albeit with lower cost, biosimilar insulin is an important tool for improving access and overcoming the limitations of human insulin with more variable glycaemic control due to its less physiological profiles.

Studies based on RWD from various healthcare systems indicate substantial cost savings associated with switching from originator to biosimilar insulins. In the United States, Basaglar and Semglee cost 22% and 64% less than originator insulin glargine. ⁵⁹ According to price data on active pharmaceutical ingredients exported from India, biosimilars of human insulin and insulin analogues cost less than US $72 and US $133 per patient‐year, respectively. ⁶⁰ This is compared to US $6000 for originator insulin analogues in the United States, providing a strong case to improve access to biosimilar insulin of high quality in all markets.

However, the low uptake of biosimilar insulin in current practice has limited the impacts of this potential cost savings. In England, the use of glargine biosimilars had generated savings of £900 000 in 2015–2018 in primary care although this represented only 3.4% of a potential saving of £25.6 million. ⁶¹ In Spain, the utilization of biosimilar insulin glargine yielded a saving of €110.5 million in 2009–2019 although there was no overall increase in insulin utilization during this period. ⁶¹ Given the scarcity of RWE of biosimilar insulin, ⁶² the cost‐effectiveness of biosimilars remains uncertain, calling for more RWD and systematic economic analysis across high‐, middle‐ and low‐income regions.

7. ADHERENCE AND ADOPTION OF INSULIN AND BIOSIMILAR INSULIN THERAPY

7.1. Therapeutic inertia and non‐adherence to insulin therapy in real world practice

Therapeutic inertia has long been recognized in diabetes management, especially for insulin initiation, intensification and persistence. ⁶³ The INSTIGATE study reported a mean HbA1c level of 9.2% at the time of insulin initiation. ⁶⁴ In a UK cohort study, only 25% of insulin‐naïve patients were initiated on basal insulin therapy within 2 years, and 50% started it within 5 years, following documented treatment failure with OGLDs. ⁶⁵ Analysis of representative EHR data indicated increasing prescription of newer agents, such as glucagon‐like peptide 1 receptor agonists (GLP1‐RAs) and sodium glucose transporter 2 inhibitors (SGLT2is) in Hong Kong, ¹⁴ , ¹⁵ United States ⁶⁶ and United Kingdom. ⁶⁷ However, the proportions of patients achieving glycaemic targets had declined in both the United States and United Kingdom. ⁶⁶ To this end, insulin remains a preferred option in PwD who had poor glycaemic control despite adhering to multiple OGLDs to ensure optimal use of glucose as an energy substrate. ⁶⁸

Initiating insulin aside, there is also a challenge in improving adherence to insulin therapy. The International Diabetes Management Practices Study Wave‐7 (2016–2017) revealed that one in seven insulin‐treated patients outside North America and Western Europe reported non‐persistence with insulin. Specifically, 14.0% of PwT1D and 13.7% of PwT2D discontinued insulin for 1–2 months on average. ⁶⁹ Retrospective analyses indicated, on average, 63% of adult insulin users were adherent compared to 73%–86% among patients using OGLDs. ⁷⁰

7.2. Barriers to insulin adherence and persistence

Barriers to insulin adherence are multifaceted, including factors at patient, physician and system levels (Figure 3). ⁷¹ Common patient‐reported barriers include fear of hypoglycaemia, needles and weight gain, along with the social stigma of injecting in public. ⁶⁹ , ⁷² Additional challenges include complicated regimens (e.g., basal‐bolus injections), medication costs, inadequate supports and high expectations for innovative medications. ⁶⁹ , ⁷² , ⁷³ On the other hand, one in three PwT1D experienced impaired awareness of hypoglycaemia (IAH), the latter being associated with a sixfold increased risk of severe hypoglycaemia. Structured education on flexible insulin therapy, including psychotherapeutic and behavioural techniques, might reduce the rates of severe hypoglycaemia. ⁷⁴ , ⁷⁵ Younger age, recent diabetes diagnosis, lower education levels and insufficient access or support for self‐monitoring of blood glucose (SMBG) tools correlated with poor persistence with insulin. ⁶⁹ These barriers underscore the need to invest in patient education and empowerment programmes in order to bring out the best of these technologies and medical knowledge.

FIGURE 3.

Barriers to insulin adherence at system, patient and provider levels.

Healthcare providers, notably non‐diabetes specialists, might hesitate to initiate or titrate insulin (especially with increasing choices of OGLDs) due to misconceptions about insulin therapy, concerns over side effects, insufficient experience and confidence with insulin management, affordability issues from the payor's perspective either out of pocket or co‐payment or restricted use in formulary in a public‐funded setting. Patient‐related factors include reluctance to take regular injections and pay out of pocket or co‐pay for insulin therapy. Insufficient patient–provider communication and engagement might exacerbate patient fears, especially if insulin initiation was presented as a punitive rather than a biological need in patients with suboptimal glycaemic control. ⁶⁹ , ⁷⁶ , ⁷⁷

In low‐ and middle‐income countries, additional challenges include insufficient structured education and support programmes, limited access to trained doctors and educators, lack of subsidized medications, poor general education and low health literacy of lay people. ⁷⁸ In these countries, the cost of diabetes care, including insulin and SMBG tools, consumes a substantial portion of household income and could be as high as 29% in Nigeria and 70% in Mali. ⁷⁹ In this light, reliable access to affordable insulin syringes, SMBG devices and test strips essential for optimizing glycaemic control remains a significant challenge for many PwD. ⁸⁰

7.3. Potential solutions to promote use of insulin

Using trained nurses to address the aforementioned barriers is a potential solution. In a cluster‐based RCT conducted in Australia, ⁸¹ nurse‐led insulin initiation and mentoring increased insulin initiation rates (70% vs. 22%) and reduced HbA1c by 0.6%, without worsening emotional well‐being, compared to primary care practice without nurse support. Similar findings were reported in Europe and the United States. ⁸² These consistent data indicated that nurses were more effective in addressing patients' concerns during insulin initiation and titration. In contrast, general practitioners, despite having the knowledge, were often challenged with time constraints in addressing the multiple concerns of patients requiring insulin treatment. ⁸³ , ⁸⁴

Information technology‐based interventions may be another solution. Computerized reminders, ⁸⁵ feedback ⁸⁶ and education may enhance timely insulin intensification by primary care physicians. ⁸⁷ In a meta‐analysis, information technology‐based interventions improved HbA1c level (mean treatment difference: −0.33%) in PwT2D compared to the control group. ⁸⁸

Biosimilar insulins offer a cost‐effective alternative to originator insulin analogues with comparable efficacy and safety. However, there is considerable variability in healthcare providers' knowledge and confidence in using biosimilars across different regions, ⁵⁷ including concerns about safety, efficacy and immunogenicity. ⁸⁹ , ⁹⁰ Closing these professional knowledge gaps through targeted education programmes is crucial to optimize initiation, titration and persistence with insulin, including biosimilar insulin, in real world settings.

7.4. Use of technologies to optimize insulin therapy

The insulin pump or CSII allows more precise minute‐by‐minute adjustments to the delivery of insulin doses. An important consideration is the compatibility of biosimilar insulins with the infusion (pump) systems (e.g., infusion sets, tubing and reservoirs) and their possibility of causing occlusions of infusion sets. The biosimilar insulin aspart, SAR341402 (U100, SAR‐Asp), has been evaluated in Medtronic MiniMed insulin pumps ⁹¹ and showed similar in vitro physiochemical stability compared to the originator rapid‐acting insulin aspart (replacement of proline with aspartic acid at position B28 on the B‐chain) by NovoNordisk (NN) (U100, NovoLog® or NovoRapid® as brand names). In a 4‐week crossover study of PwT1D, the number of participants reporting ≥1 infusion set occlusion was similar between the two treatment groups: 14/43 in the biosimilar SAR‐Asp group (33 events) and 12/43 in the originator NN‐Asp group (24 events). The number of participants with ≥1 episode of unexplained hyperglycaemia was similar between the two groups (31/43 on SAR‐Asp [154 events]; 32/43 on NN‐Asp [175 events]). Likewise, SAR342434 (U100; SAR‐Lis; insulin lispro) is a biosimilar/follow‐on to the originator insulin lispro (U100; characterized by an inversion of the prolin‐lysin amino acid sequence at positions 28 and 29 on the B‐chain) which has also been shown to be safe for use in CSII. ⁹²

Smart or connected insulin pens have memory functions which can be connected with CGM and phone apps. ⁹³ However, many of these connected or smart pens only work with insulins from a single manufacturer, and it remains to be seen if they could operate within a range of biosimilar insulins. Technology aside, some of the fundamental challenges of insulin delivery remain, such as the pain of injections, lipohypertrophy and the inconvenience of multiple injections. Thus far, non‐parenteral insulin delivery has been met with mixed success. Despite the approval of inhaled insulins for PwT1D and PwT2D, uptake has been low, in part due to challenges in ease of administration and concerns over long‐term pulmonary safety. ⁹⁴

8. INSULIN REQUIREMENT IN DIFFERENT SUBTYPES OF DIABETES

In PwT1D with absolute insulin deficiency, insulin prevents life‐threatening hyperglycaemic crisis and ketoacidosis. Despite the promise of adjunctive therapies, immunotherapy and stem cell transplantation, achieving insulin‐free remission in PwT1D remains challenging. ⁹⁵ In a global survey, the incidence of T1D among adolescents and young adults increased from 7.78 per 100 000 population in 1990 to 11.07 per 100 000 population in 2019, accompanied by a 7.4% increase in mortality. ⁹⁶ Even in low prevalence areas for T1D such as Hong Kong, the incidence of T1D in people aged ≤20 years had risen from 3.5 to 5.3 per 100 000 person‐years in boys and from 4.3 to 6.4 per 100 000 person‐years in girls in 2002–2015. ¹⁶ Globally, there are 1.8 million PwT1D with the largest increase in incidence in low‐ and middle‐income countries. ⁹⁷ It has been estimated that one in five PwT1D live in low‐income and lower‐middle‐income countries with poor access to insulin. This has led to a widening gap in life expectancy between PwT1D living in low‐income and their counterparts in high‐income areas. ⁹⁷

Latent autoimmune diabetes in adults (LADA) or more appropriately referred to as slowly progressing autoimmune diabetes is characterized by the presence of immunogenetic markers similar to classical T1D. These patients exhibit a hybrid metabolic state between that of T1D and T2D, with marked phenotypic heterogeneity. The initial presentation resembles T2D, followed by rapid failure of OGLDs and insulin dependency. The diagnosis of slowly progressing autoimmune diabetes in adults is often missed or misclassified resulting in delayed insulin therapy. Autoimmune islet antibodies can be detected in 2%–12% of PwT2D ⁹⁸ with considerable heterogeneity in the distribution of age, age of diagnosis and speed of deterioration in beta cell function.

In Chinese patients with T2D presentation, a high glutamic acid decarboxylase (GAD) autoantibody titre predicted early insulin requirement with robust glucose‐lowering responses. ⁹⁹ In an expert consensus recommendation for the treatment of slowly progressing T1D, insulin was recommended in patients with C peptide <0.3 pmol/L. For patients with intermediate C peptide levels (0.3–0.7 pmol/L), DPP4i, GLP1‐RA, thiazolidinediones (TZDs) or SGLT2is can be used depending on the individual's cardiovascular risk profile. In these patients, supplementary insulin should be considered if HbA1c targets are not met.

Type 3c diabetes refers to chronic hyperglycaemia due to pancreatic diseases, the most common being chronic pancreatitis. ¹⁰⁰ In a large single‐centre review, 79% of patients with type 3c diabetes had chronic pancreatitis, followed by pancreatic ductal adenocarcinoma (8%), haemochromatosis (7%), cystic fibrosis (4%) and previous pancreatic surgery (2%). Most of the patients with type 3c diabetes have insulin deficiency, often complicated by deficiency of exocrine function such as nutritional malabsorption. ¹⁰⁰ The concomitant impaired glucagon response and low glycogen stores predispose patients with type 3c diabetes to hyperglycaemia and hypoglycaemia. ¹⁰¹ Many of these patients have ‘brittle diabetes’ requiring treatment with multiple daily injections of insulin. ¹⁰²

Patients with monogenic diabetes, often due to rare variants implicated in beta cell biology, may also require insulin at diagnosis or during the subsequent clinical course. For neonatal diabetes mellitus (NDM), apart from carriers of mutations in the Potassium Inwardly Rectifying Channel Subfamily J Member 11 (KCNJ11) or ATP Binding Cassette Subfamily C Member 8 (ABCC8) (40% of patients with NDM) who are sulphonylurea‐responsive, patients with other forms of NDM often require lifelong insulin treatment. ¹⁰³ Patients with transcription factor (hepatocyte nuclear factor HNF1‐alpha or HNF4‐alpha) monogenic diabetes may respond to sulphonylureas initially but eventually often need insulin replacement with progressive beta cell exhaustion. ¹⁰⁴

Type 2 diabetes is characterized by progressive loss in beta cell structure and function with absolute or relative insulin deficiency in the presence of insulin resistance, the latter often due to obesity. ¹⁰⁵ Approximately 36% of PwT2D require insulin within 8 years of diagnosis. ¹⁰⁶ In a territory‐wide EHR analysis in Hong Kong, despite the availability of newer GLDs such as GLP1‐RA and SGLT2is, the proportion of PwT2D treated with insulin remained static (11.2%–13.6% in 2002–2019). ⁵⁷ When projected against the growth of PwT2D worldwide, insulin use was estimated to increase from 516.1 million 1000 international units (IU) vials (10 mL vial of U100 insulin) (95% confidence interval [CI]: 409.0–658.6 million) per year in 2018 to 633.7 million (95% CI: 500.5–806.7 million) per year in 2030. ¹⁰⁷

In the JADE Program ⁵ including over 90 000 PwT2D from 11 countries in Asia, 25% of them were treated with insulin and 29% had chronic kidney disease (CKD). Within this programme, premixed insulin was popular albeit with a high risk of hypoglycaemia in patients with CKD. ¹⁰⁸ Patients with YOD diagnosed before the age of 40 years had a more rapid decline in beta cell function than those with late‐onset diabetes. ¹⁰⁹ In Chinese PwT2D, fasting plasma C peptide levels declined more rapidly with disease duration in YOD, especially those with lean body mass, than their counterparts with late‐onset diabetes (LOD) diagnosed after the age of 40 years. ¹¹⁰ Patients with YOD were more likely to be treated with premixed or intensive insulin regimens and had worse glycaemic control than patients with LOD. ¹⁴ , ¹⁰⁸

At the end of the spectrum, older adults with diabetes are also commonly treated with insulin due to long disease duration and renal impairment. A US survey estimated nearly one in five PwT2D aged over 75 years were treated with insulin. ¹¹¹ In a large retrospective cohort study of older adults aged >65 years with T2D, initiation of glargine and detemir use was associated with reduced risk of hypoglycaemia compared with NPH insulin use (glargine versus NPH insulin: hazard ratio [HR] = 0.71, 95% CI: 0.63–0.80; detemir vs. NPH insulin: HR = 0.72; 95% CI: 0.63–0.82). ¹¹² There is a consensus to use longer‐acting basal insulin analogues to reduce the risk of hypoglycaemia in older adults. However, in real world practice, these high‐risk patients were more likely to be treated with human insulin than their younger counterparts. ¹⁴

9. FUTURE DIRECTIONS AND OPPORTUNITIES TO IMPROVE ADHERENCE, OUTCOMES AND COST‐EFFECTIVENESS

Insulin has been discovered for more than a century, although its benefits to PwD are yet to be fully realized due to barriers related to access, quality of care and costs. The multicausality of diabetes and its subtypes calls for comprehensive phenotyping and use of biomarkers to ensure the timely use of insulin to achieve early glycaemic control for long‐term benefits. With increasing longevity, many PwD will eventually require insulin to avoid hyperglycaemic crisis, reduce complications and improve quality of life. The minute‐to‐minute variations in blood glucose, subject to changes by many lifestyle factors, call for the use of insulin analogues that mimic physiological patterns to avoid widely fluctuating blood glucose.

Insulin is more than a prescription but requires a service package for initiation, titration and persistence along with monitoring tools for feedback. Self‐management is a cornerstone in diabetes management where education with ongoing support is an integral part of insulin management. Increasing competition from manufacturers, especially from Asia, will eventually drive down the costs of biosimilar insulin, CGM and CSII, to benefit more PwD. Scientific evaluation with regulatory approval is critical in ensuring the quality of these technologies; it is here where diabetes registers implemented by a trained doctor–nurse team with system support can become an invaluable tool to assess the impacts of these technologies on clinical outcomes, quality of life and healthcare costs to inform practice and policies.

AUTHOR CONTRIBUTIONS

AY, JCNC and EC contributed to conception, design and writing manuscript. JY and JTKC contributed to design and writing manuscript.

FUNDING INFORMATION

This research received no specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors.

CONFLICT OF INTEREST STATEMENT

EC has received speaker honoraria/institutional research support from Astra Zeneca, Boehringer Ingelheim, Bayer, Merck KGaA and Sanofi. All proceeds have been donated to CUHK to support diabetes research. JCNC reported receiving grants (through institutions) and/or honoraria for consultancy or giving lectures from Abbott, Applied Therapeutics, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Hua Medicine, Merck, Novo Nordisk, Powder Pharmaceuticals, Roche, Sanofi, Servier and ZP Therapeutics. She is the Chief Executive Officer (pro bono) of the Asia Diabetes Foundation that developed the web‐based JADE platform for implementation of data‐driven diabetes care. She is a co‐inventor of patents owned by the Chinese University of Hong Kong with claims of using biomarkers and models to predict diabetes and its complications. She is a co‐founder of GemVCare, a technology company, with partial support from the Hong Kong Government, which uses biogenetic markers and information technology to implement precision diabetes care and prevention through partnerships. She is a member of the Global Council of the European Association Study of Diabetes and Honorary Fellow of the International Diabetes Federation.

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/dom.16386.

Yang A, Yu J, Cheung JTK, Chan JCN, Chow E. Real world evidence of insulin and biosimilar insulin therapy—Opportunities to improve adherence, outcomes and cost‐effectiveness. Diabetes Obes Metab. 2025;27(Suppl. 5):45‐62. doi: 10.1111/dom.16386

DATA AVAILABILITY STATEMENT

Data sharing not applicable to this article as no datasets were generated or analysed.