Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Detecting undiagnosed people with type 2 diabetes accelerates the time of diagnosis and reduces morbidity and mortality.WHAT THIS STUDY ADDS Screening for type 2 diabetes using at-home self-sampling HbA1c, targeted individuals aged 50–75 who have not had their HbA1c measured in the last 2 years, is feasible and estimated to be cost-effective in Denmark.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY At-home self-sampling HbA1c tests can complement existing opportunistic screening strategies and help reduce the incidence of diabetes complications and excess mortality.Introduction The number of individuals living with type 2 diabetes is expected to rise globally, while treatment of type 2 diabetes poses a substantial economic burden worldwide. In Denmark, approximately 7% of the adult population, 333 000 individuals, lived with type 2 diabetes in the first quarter of 2025 1 and the number is expected to increase to 9.5%, by 2040.2 People with type 2 diabetes in Denmark incur an estimated healthcare cost of €595 989, including treatment, care and medication, while the indirect costs due to productivity loss, including labor market absence and premature mortality, amount to approximately €478 487.3A substantial share of the additional cost is driven by diabetes complications and the containment of these complications would reduce costs. HbA1c levels are the most significant predictor of the development and progression of type 2 diabetes complications. Compared with a HbA1c value of 48–58 mmol/mol (6.5%–7.5%), individuals with a HbA1c level of 70–80 mmol/mol (7%–9.5%), within the first year after diagnosis, have a threefold increased risk of developing microvascular complications. This risk is sixfold for those with HbA1c levels of 90–100 mmol/mol (10.4%–11.3%).4Consequently, therapeutic inertia with delayed treatment of hyperglycemia is related to significant healthcare and labor market costs.5 While therapeutic inertia is a major problem in Denmark affecting about 50% of the individuals diagnosed with type 2 diabetes, approximately 165 000 cases,5 a critical concern is the large proportion of undiagnosed and untreated cases. It is estimated that 25% of the individuals with type 2 diabetes, about 100 000 cases in Denmark, are undiagnosed.6The early-stage type 2 diabetes is often asymptomatic,7 which contributes to delayed diagnosis, thereby increasing the likelihood of developing complications. One third of people diagnosed with type 2 diabetes already have complications at the time of diagnosis.6 8 This is the main reason why screening is recommended for type 2 diabetes.9Type 2 diabetes meets many of the established criteria for screening to be effective (10). It has (1) a long asymptomatic phase during which complications may develop, (2) simple, reliable and cost-effective diagnostic tests and (3) well-established interventions for glucose, blood pressure and lipid management.Screening is generally approached within three broad screening schemes: (1) universal or population-based screening through a general direct outreach, (2) targeted screening approaching selected groups who are at high risk of having type 2 diabetes using risk assessment surveys or biomarkers such as hypertension and (3) opportunistic screening examining individuals who are presenting to health services for other reasons.The scarce randomized controlled trial (RCT) studies are unable to document that population-based screening for type 2 diabetes reduces neither morbidity nor mortality in the general population.10–12 This is, however, not surprising, as screening for type 2 diabetes identifies only a small fraction of the individuals in the general population, who will experience CVD or advanced mortality. Hence, the impact of screening on population-level CVD and mortality remains minimal, even if early detection improves outcomes for those diagnosed,13 as suggested by non-randomized controlled studies.14 15To identify an effect of population-based screening, the number needed to invite in both trial arms is substantial, while healthcare records are required to identify type 2 diabetes as the cause of comorbidities and mortality of the individuals with undiagnosed type 2 diabetes.16 17 In Denmark, the number needed to invite exceeds the number of individuals in the population, not currently screened opportunistically. In addition, despite high-quality healthcare data, type 2 diabetes remains under-reported as the underlying cause of diabetes comorbidities and mortality.6In contrast, non-randomized simulation studies as well as Systematic Review and Meta-Analysis suggest that targeted and opportunistic screening reduces morbidity and mortality,14 15 improve quality-adjusted life-year and are cost-effective at screening every 3–5 years in adults aged 45 and above with normal blood glucose levels, but every 2 years in high-risk groups.18–20 Accordingly, the WHO and the National Institute for Clinical Excellence recommend targeted and opportunistic screening of the population at risk of having type 2 diabetes.21–23 Non-systematic opportunistic screening in clinical settings is on the rise in Denmark. In 2018, approximately 50% of 60-year olds, 60% of 70-year olds and 68% of 80-year olds had at least one HbA1c measurement recorded within the last year.24 However, opportunistic screening may not screen less health-conscious and vulnerable individuals.25 Moreover, individuals living with unknown type 2 diabetes may not regularly visit their general practitioner (GP).We suggest a fourth screening scheme using at-home self-sampling HbA1c test targeting individuals without a HbA1c measurement within the last 2 years. The quality of the national registers allows screening programs to target individuals, not tested opportunistically, while novel and cheaper test and storage methods allow sampling to take place outside the clinic settings. The aim of the study is to evaluate whether screening for type 2 diabetes using at-home self-sampling HbA1c tests, targeted individuals between 50 and 75 years of age who have not tested their HbA1c level within the last 24 months, is feasible and cost-effective in Denmark.Methods Study design The study employed a two-stage design. First, an at-home self-sampling screening pilot study was conducted to evaluate the feasibility and to obtain estimates of the participation and detection rates as well as the distribution of the HbA1c level among participants. Second, the estimates were applied to calculate the healthcare and labor market cost savings associated with advanced type 2 diabetes detection through self-sampling. This analysis leveraged existing simulations on the economic cost of therapeutic inertia at varying HbA1c levels 1-year, 3-year and 5-year post-diagnosis.At-home self-sampling Over a period of 9 months, free at-home capillary blood sampling kits were distributed to 8000 randomly selected individuals aged 50–75 years without an HbA1c measurement over the past 2 years. The population was identified using the Danish Clinical Laboratory Information Register (LABKA), which compiles laboratory test results from hospitals and GPs. The decision to target individuals aged 50–75 years was based on the current level of opportunistic screening for HbA1c by age group in Denmark, 3 while the 2-year threshold prioritized identifying individuals at elevated risk of developing type 2 diabetes,20 although at the expense of a lower overall detection rate and cost-effectiveness. The sample size was restricted due to practical reasons, that is, resources and time.Participants received an invitation letter together with a leaflet with (1) instructions on how to collect a capillary blood sample using the provided self-sampling kit, (2) information on how to return the sample for laboratory HbA1c analysis using a prepaid envelope and (3) information about how to access the test results. The test results were recorded in LABKA and made accessible within 2 weeks to the participants and their GPs through their electronic health record. Participants with a HbA1c level of 48 mmol/mol (6.5%) and above were contacted by the doctor before they were able to access their results.The at-home capillary blood sampling kit consisted of a Hemoglobin Capillary Collection System (HCCS), integrating collection, preparation and transport of blood samples for HbA1c analysis, and two 1.5 mm single-release finger-stick blood lancets. Sample stability was verified in an internal pilot verification study, in which 26 individuals from an outpatient clinic provided a total of 47 capillary samples analyzed immediately and after mailing at day 6, 11 and 13. No significant differences were observed in HbA1c levels across days, confirming that the HCCS samples remained stable for at least 13 days without refrigeration. The blood sample’s HbA1c was analyzed in the Lab with a Tosoh analyzer. The Tosoh analyzer uses ion-exchange high-performance liquid chromatography, which is regarded as the gold standard due to its high precision, rapid analysis times and the ability to separate HbA1c from other hemoglobin fractions for reliable results.26Statistical analysis Descriptive statistical analyses were conducted to examine participation and detection rates as well as HbA1c results by sex and age groups. Similarly, the sex and age distributions among invited and participants were compared with the general population, while the sex and age distribution of the screen-detected was compared with individuals diagnosed with type 2 diabetes through the Danish Health Care System (DHCS) in 2023. The sex and age distribution of the general population is derived from Statistics Denmark, while the sex and age distribution of those diagnosed through DHCS is derived from the Register of Selected Chronic Diseases and Severe Mental Disorders. Comparisons were performed using the χ 2 test, with a significance threshold of p<0.05.The modeling approach of cost reductions To estimate the cost savings associated with earlier type 2 diabetes diagnosis through screening using at-home self-sampling, we applied a Danish simulation study on the healthcare and labor market cost of therapeutic inertia for individuals with type 2 diabetes across HbA1c targets as defined by the Danish Endocrinological Society, 4 including the HbA1c targets of 48 mmol/mol, 53 mmol/mol (7%) and 63 mmol/mol (7.9%). Moreover, we weighted the estimated costs of therapeutic inertia for each HbA1c target group derived from Lindvig and colleagues5 by the distribution of HbA1c observed among screen-detected participants in the screening study, assuming that the HbA1c distribution of the participants is representative for a national screening program in Denmark. The calculations were repeated across the different scenarios of earlier diagnosis.Lindvig and colleagues5 apply a type 2 diabetes model developed by the Swedish Institute for Health Economics (IHE) to calculate the increased incidence of microvascular and macrovascular complications as well as mortality over a 40-year time horizon due to therapeutic inertia for 1, 3, 5 and 7 years following diagnosis, compared with full glycemic control of 6.5% (48 mmol/mol). Subsequently, they calculate the societal costs of diabetes complications and mortality applying Danish register-based estimates provided by Kjellberg et al27 on the direct healthcare expenditures for treatment of type 2 diabetes and comorbidities as well as indirect labor market costs incurred the first 3 years following an incident of each of the 17 most common diabetes complications. The indirect costs further include absenteeism associated with complications, provided by Sørensen and Ploug.28 Income statistics were sourced from Statistics Denmark. Hence, the model inputs, including clinical and cost parameters, are user-defined. For detailed model inputs, settings and cost estimates, see Lindvig et al.5 We further deduct the treatment costs due to additional patient years following advanced diagnosis from the estimated cost savings. The treatment costs per patient year comprise of GPs chronic fee, costs to yearly tests and self-paying for foot screening.The calculations are repeated across three scenarios of early diagnosis, including 1, 3 and 5-year earlier diagnosis. Based on evidence from Denmark8 14 indicating that identifying undiagnosed individuals with type 2 diabetes advances the time of diagnosis by about 3 years, we have chosen this as our preferred scenario.Return on investment Finally, the net cost savings were compared with total screening costs per screen-detected to estimate the return on investment of type 2 diabetes screening using at-home self-sampling. The total screening costs per screen-detected are calculated as (I+S*P)/(P*D), where I is the invitation costs per invited, S is the screening costs per screened, P is the participation rate and D is the detection rate. The invitation costs per invited include the letter, packaging, postage and the at-home self-sampling equipment. The costs of identifying eligible individuals are not included, as it is possible to fully automate this process. The screening costs per screened encompass return postage and the laboratory analysis cost. We apply the participation and detection rate reported in the at-home self-sampling screening study to calculate screening costs.We assume that the screening costs per screen-detected are reduced in a screening program organized by the National Health authorities. First, evidence from Denmark suggests that a national screening program increases participation rate compared with pilot screening studies while the share of screen-detected remains constant as the participation rate increases. Second, with a larger volume of samples, it is possible to reduce purchase prices and fixed costs, for example, associated with data extraction, while workflows can be automated and thus reduce working time per sample. A plausible but cautious scenario is to consider a 10 percentage point higher participation rate and 10 percentage reduction in total costs due to scale effects.Kraka Economics was contracted to conduct the economic calculations based on the results on the HbA1c distribution among participants in the screening study and the existing simulation study.Ethics The participants received information about the pilot study, an informed consent form, with relevant information about data analysis and storage and the ethical aspects of participating in a screening campaign that they needed to fill with personal information, including personal ID number and contact details. A filled and signed consent was required for the invited individuals to participate in the study.All data were anonymized and processed in accordance with Danish Law and the General Data Protection Regulation 2016/679 of The European Parliament and of The Council. The participants took part in the study voluntarily and consented on their data to be used anonymized in the study.Although there are no physical side effects in screening for type 2 diabetes with a capillary test, screening for a disease that is asymptomatic in early stages may lead to stress and/or anxiety. Participants with a HbA1c value of 48 mmol/mol (6.5%) and above were contacted by a medical doctor associated with the project to ensure that participants were informed about the implications of the result and how to handle the situation, before being able to access their result.Results Participation and detection rate The participation rate was 38%. However, due to insufficient blood samples, the successful participation rate was 36%, 2913 individuals. As central administrative register data were applied for the selection of eligible individuals, information on sex and age was complete for all invitees, participants, and screen-detected individuals.Of the successfully tested participants, 1.7% (50 individuals) were detected with HbA1c of 48 mmol/mol (6.5%) or above, while an additional 10.2% (298 individuals) had HbA1c of 42–47 mmol/mol (6%–6.4%). Half of the screen-detected had HbA1c values from 48 to 53 mmol/mol (6.5%, 7%), 24% a HbA1c level of 53 to 63 mmol/mol (7%, 7.9%), and 26% a HbA1c of 63 mmol/mol (7.9%) or above.Invited individuals were evenly distributed by sex but not across age groups. Younger age groups were overrepresented among those invited, compared with the general population, reflecting the lower likelihood of having had an HbA1c measurement within the past 2 years. Specifically, 30.1% of all invited individuals were aged 50–54 years; 18.4% aged 60–64 years and 10.1% aged 70–75 years. The age distribution of the successfully tested participants mirrored that of the invited individuals for both men and women, as participation rates were similar across all age groups, but slightly higher among women than men (38% and 35%) (table 1).Table 1Demographic characteristics of the individuals invited to screening and the participantsAge distribution (%)Invited menParticipants menInvited womenParticipants womenInvited totalParticipants total50–5430.130.027.528.228.829.155–5927.525.125.625.226.625.260–6418.419.018.819.418.619.265–6913.915.814.915.714.415.870–7510.110.013.311.511.610.7Observations4.0971.4213.9031.4928.0002.913Source: own calculations.The share of screen-detected with HbA1c of 48 mmol/mol (6.5%) or above was 2.32% among men and 1.14% among women. Individuals aged 50–59 years constituted 50% of the screen-detected compared with 40% in the same age group among those diagnosed by the DHCS. While this represents a substantial difference in the age distribution, it is not statistically significant (table 2).Table 2Demographic characteristics of the screen-detected participants and individuals diagnosed through the Danish Health Care System (DHCS)Age distribution (%)Screen-detected menDHCS detected menScreen-detected womenDHCS detected womenScreen-detected totalDHCS detected total50–5421.217.323.517.122.017.355–5927.323.529.422.728.023.259–6418.222.011.821.316.021.765–6918.219.823.520.420.020.070–7515.217.311.818.514.017.8Observations3380751754005013 475Source: RUKS register 2023 and own calculations.Economic cost savings and return on screening If screening detects individuals with type 2 diabetes 3 years earlier, the cost savings constitute €97 in direct costs and €168 in indirect cost for every screen-detected individual with a HbA1c of 48–53. For screen-detected with a HbA1c of 53–63 and above 63, the cost savings amount to €217 and €376 in direct cost as well as €377 and €652 in indirect cost, respectively, table 3, as derived from Lindvig et al.5 The total direct costs constitute €690, while the indirect costs constitute €1197, when weighted by the HbA1c distribution among screen-detected participants in the screening study resulting in total cost savings of €1887. The costs of additional patient years are deducted from the gains equivalent to €124 per patient year. Hence, the socioeconomic gain constitutes €1514 per screen-detected, table 4, if screening results in at least 3 years earlier diagnosis. If the screening detects individuals with type 2 diabetes 1 or 5 years earlier, the total socioeconomic gain constitutes €575 and €2219, respectively.Table 3Economic cost savings of screening per screen-detected with type 2 diabetes measured as the difference in lifetime costs due to earlier diagnosis (€)Cost savings per HbA1c level1-year earlier diagnosis3-year earlier diagnosis5-year earlier diagnosisDirect costs 48–53 mmol/mol (6.5%–7%)73194303Indirect costs 48–53 mmol/mol (6.5%–7%)124336495Direct costs 53–63 mmol/mol (7%–7.9%)3399051415Indirect costs 53–63 mmol/mol (7%–7.9%)57915702311Direct costs≥63 mmol/mol (7.9% -)54114452259Indirect costs≥63 mmol/mol (7.9% -)92425063690Total direct costs2586901078Total Indirect costs44111971762Source: The Danish Diabetes Association, Lindvig et al5 and own calculations.HbA1c: hemoglobin A1c Table 4Simulation of economic gains and costs of early type 2 diabetes detection (€)Gains and costs per screen-detected (€)1-year earlier diagnosis3-year earlier diagnosis5-year earlier diagnosisDirect economic gains2586901078Indirect economic gains44111971762Direct and indirect costs per extra patient year−124−373−621Total gains57615142219Screening costs118311831183Net gains−6073311036Return ratio on screening costs0.491.281.88The Danish Diabetes Association, Lindvig et al5 and own calculations.We apply the invitation costs per invited (€5.89), the screening cost per screened (€9.67) as well as the participation rate (0.36415) and detection rate (0.0171644) obtained from the at-home self-sampling screening pilot study to estimate the screening costs per screen-detected, while assuming that a national screening program provides both a 10 percentage point increase in the participation rate and a 10% reduction in screening costs due to scale effects. Hence, the screening costs per screen-detected of a national screening program are estimated to(€5.89*0.9+€9.67*0.9 (0.36415+0.1))/((0.36415+0.1)*0.0171644) = €1,183.If screening results in a 3-year earlier diagnosis of type 2 diabetes, it provides a societal return rate of €1514/€1183=1.28, corresponding to approximately 28%. For every euro invested in the screening program, the society saves 1.28 on treatment and lost income for people with type 2 diabetes as the screening program advances the time of diagnosis and treatment and thus reduces the risk of diabetes complications that require costly treatment and lead to sick leave. This assessment excludes the substantial health benefits for quality of life and the value of this (QALY). Recent Danish cost-effectiveness studies for type 2 diabetes therapies often use a willingness-to-pay threshold of approximately DKK 224 000 per QALY (€30 000 per QALY) as a benchmark for cost-effectiveness.29Discussion Our findings, supported by existing evidence, suggest that screening for type 2 diabetes using at-home self-sampling HbA1c tests, targeted to individuals aged 50–75 years without a HbA1c measurement within the last 2 years, is feasible and cost-effective in Denmark.The share of screen-detected with HbA1c of 48 mmol/mol (6.5%) or above is consistent with the estimated share of individuals with unknown type 2 diabetes in Denmark in the same age group.6 The distribution of HbA1c among screen-detected participants is not statistically different from the distribution among patients diagnosed through the DHCS. Half of the newly diagnosed with type 2 diabetes through the DHCS in 2018 had a HbA1c level of 53 mmol/mol (7%) or above, while 31% had a HbA1c level over 62 mmol/mol (7.9%).30 However, individuals aged 50–59 are slightly over-represented of the participants compared with the individuals diagnosed by the DHCS (51% vs 40%).Finally, we estimate that a targeted screening program advancing diagnosis by 3 years across all HbA1c levels results in an average cost savings of €1.514 per screen-detected and a total social return ratio of 1.28.A key strength of the study is the novelty of the screening method. To our knowledge, this is the first study investigating screening for type 2 diabetes by applying an at-home self-sampling HbA1c test kit. In contrast, most evidence concerns screening in clinical settings using fasting plasma glucose or oral glucose tolerance tests. Another strength is the accuracy in identifying the individuals without a HbA1c measurement within the last 2 years. This is possible due to LABKA’s quality and completeness.The cost-effectiveness results require cautious interpretation due to study limitations. First, while the estimated cost savings related to undiagnosed type 2 diabetes derived from Lindvig et al5 are based on the IHE model,31 the estimated cost savings have a long time horizon. Similarly, the estimated cost savings are for simplicity provided within HbA1c intervals, reducing the accuracy of the estimated cost-effectiveness. We further assume that the screening program advances the time of diagnosis by 3 years. Previous literature suggests that screening advances type 2 diabetes diagnosis by 3–6 years,8 while two later studies report an average of 2.2 years compared with clinically detected cases.14 32Furthermore, some of this study’s assumptions likely result in an underestimation of the true economic burden of undiagnosed type 2 diabetes and the potential cost savings from earlier diagnosis. First, in line with Lindvig et al,5 we assume constant HbA1c levels in the prediagnosis period. However, HbA1c levels increase over time due to a declining β-cell function decreasing patients’ ability to maintain glycemic control. Second, other biomarkers associated with diabetes progression and risk-enhancing behaviors, including systolic blood pressure, dyslipidemia, body weight, smoking and alcohol intake, are assumed to remain constant before and after diagnosis. However, type 2 diabetes diagnosis and treatment may lead to improvements in other biomarkers and risk-related behaviors.11 Third, the healthcare cost does not include municipal expenses such as rehabilitation, nursing care and diabetes aid costs.16 Prior research indicates that municipal nursing care constitutes a significant cost associated with diabetes complications.33 Moreover, the study does not include non-pecuniary effects of early diagnosis and treatment, such as an increased quality of life for the screen-detected and their relatives. Finally, the study does not estimate the optimal frequency of screening. The 2-year threshold prioritizes identifying individuals at elevated risk of developing type 2 diabetes,20 although at the expense of a lower overall cost-effectiveness.Conversely, other assumptions may result in an upward bias in the estimated cost savings. First, screen-detected participants are assumed to lower their HbA1c to 48 mmol/mol (6.5%), following diagnosis. This is a simplification of real-world clinical practice and may overestimate the cost savings from early detection. The Danish type 2 diabetes treatment guidelines4 34 define four therapeutic HbA1c targets allowing for individualized treatment considering disease progression and comorbidities. Furthermore, HbA1c levels may increase over time despite of treatment, while treatment may intensify, as suggested by evidence from real-life clinical practice.34 35 Second, costs associated with disease management during additional patient years are deducted from the economic cost savings of early detection. It includes GPs chronic fees, costs of yearly tests and self-paid foot screening. Type 2 diabetes drug costs and adverse drug-specific health outcomes are not included as disease management cost. Sortsø et al36 find, however, that most direct treatment costs are due to diabetes complications while diabetes treatment accounts for only 8% of total direct costs.Finally, alternative screening approaches and voluntary self-measurement should be acknowledged for further research. Lower cost methods, such as urine dipstick testing, may increase participation and reduce screening costs per screen-detected. However, inherent limitations such as lower sensitivity and diagnostic specificity may lead to missed cases or increased need for confirmatory testing. Hence, empirical evidence directly comparing the cost-effectiveness of at-home HbA1c self-sampling with other low-cost screening strategies are needed. In parallel, the increasing availability of personal glucose measurement technologies necessitates consideration of voluntary self-measurement. While such technologies may expand testing among health-conscious individuals, voluntary use is subjected to self-selection and is less likely to reach populations at highest risk of undiagnosed type 2 diabetes. Hence, future evaluations of screening programs should consider integrating self-reported measures with appropriate clinical follow-up, while maintaining a focus on under-screened populations.Conclusion This study contributes to the existing literature by being the first to evaluate the feasibility and cost-effectiveness of screening for type 2 diabetes using at-home self-sampling HbA1c tests. The study targets individuals without a HbA1c measurement, filling a gap in our understanding of diabetes screening strategies.The study suggests that the at-home self-sampling screening setup is cost-effective. Importantly, this method should be viewed as complementary rather than competing with other screening strategies.By implementing a diverse array of screening strategies tailored to different population segments, healthcare providers can identify a broader spectrum of individuals with type 2 diabetes at earlier stages. Hence, reducing the share of diabetes complications and excess mortality and consequently lowering the direct and indirect economic burden of type 2 diabetes.While these results are promising, future research is needed to evaluate the feasibility and the long-term cost-efficiency of different screening initiatives across national settings.