An economic value of the glycated hemoglobin test in diabetes mellitus type 2 diagnosis

Glycated haemoglobin (HbA_1c) levels are widely used to monitor the efficiency of treatment for type 2 diabetes mellitus (DM2). Nevertheless, the issue of using the HbA1c test for diagnosing DM2 remains unresolved despite its obvious clinical advantages, such as no requirement for blood sampling strictly on an empty stomach or special preparation, great stability of the results even during periods of stress or intercurrent diseases and wide availability in most contemporary laboratories [1, 2]. The International Expert Committee concluded that HbA_1c of 6.5% or greater can be used as a threshold for screening DM2. The report of this committee also identified groups of patients for which HbA_1c may not be indicative; these are patients with haemoglobinopathies, children, adolescents and pregnant women [3]. Nevertheless, HbA_1c remains an efficient diagnostic method for DM2, with optimal sensitivity and specificity [4, 5].

An important factor supporting the selection of the HbA_1c test as a diagnostic method for DM2 is its relationship with the assessment of cardiovascular risks. Compared with an assessment of fasting plasma glucose (FPG), HbA_1c demonstrated not only its efficiency in the diagnosis of DM2 but also in determining the risk of cardiovascular disease and overall mortality [6]. HbA_1c assessment is a priority in clinical practice as a method for identifying a large number of patients with a high risk of DM2 and cardiovascular diseases [7]. The importance of the glucose tolerance test (GTT) is also emphasised as an ‘arbitrator’ for verifying the diagnosis of prediabetes and DM2 [8]. This test should be performed after repeated measurement of HbA_1c, which will repeat the results of abnormalities to avoid false positive reactions [9].

In 2011, WHO approved the possibility of using the HbA_1c test for the diagnosis of DM2 because of its obvious advantages. However, in routine practice, examination (including preventive medical examination and periodic screening) of the adult population is still based on FPG level. A common explanation for this situation is the high cost of the HbA_1c test, which exceeds 3–4 times in some laboratories than for glucose analysis.

AIM

This study aimed to model the treatment costs in the time perspective when the HbA_1c test is used in comparison with the FPG test for DM2 diagnosis.

METHODS

A retrospective analysis of aggregated (without patient personification) parts of the database of the Russian epidemiological study NATION [10] for 810 respondents (newly diagnosed with DM2) who have the results of HbA_1c, FPG and GTT analyses as a verifier of carbohydrate metabolism disorders. Diagnostic criteria are presented in Table 1.

Table 1. Reference values for determining the norm, prediabetes and DM2.

Notes: glucose tolerance test.

The sensitivity, specificity, PPV (positive predictive value, prognostic value of a positive result, the probability of a disease with a positive (pathological) test result; it is defined as the ratio of true-positive results, i.e. cases when the test provides a positive prognosis with an actually positive result, to the sum of true-positive and false-positive results, that is, cases when the test provides a positive prognosis with an actually negative result) [11] and NPV (negative predictive value, prognostic value of a negative result, the probability of absence of the disease with a negative (normal) test result; it is defined as the ratio of true-negative results, i.e. cases when the test provides a negative prognosis with an actually negative result, to the sum of true-negative and false-negative results, etc. cases when the test provides a negative prognosis with an actually positive result) of the tests for HbA_1c and FPG. The number of false-positive and false-negative results from the HbA_1c and FPG tests for the diagnosis of DM2 were compared.

For the clinical and economic evaluation of diagnosis of carbohydrate metabolism disorders using the tests under consideration, we used a previously developed simulation model of DM2 outcomes with a discrete course of events in the 20-year perspective [12]. The model used in the calculations for each of 810 patients in the initial NATION sample, based on the values of HbA_1c, blood pressure, body weight and lipid profile, calculates the risks of DM2 complications over time. This model enables to predict with a high probability the development of complications of the disease depending on the efficiency/inefficiency of its control. This model considers the effect of the ongoing hypoglycaemic therapy and other indicators, such as lipid profile parameters [13]. In a clinical and economic assessment, the likelihood of DM2 complications, obtained on the basis of the model prognosis, is multiplied by the cost of therapy. The model also considers indirect costs. The costs for each patient are summed up with the subsequent calculation of average indicators.

For each patient group formed depending on the levels of HbA_1c, FPG and GTT determined, target values of HbA_1c were established to control DM2, and three treatment algorithms with a modelling horizon of up to 20 years were proposed. Three post-diagnosis treatment algorithms were examined in terms of costs.

Algorithm 1: stage 1 Metformin → stage 2 Metformin + dipeptidyl peptidase-4 inhibitors (iDPP4) → stage 3 ­Metformin + iDPP4 + gliflozins → stage 4 Metformin + iDPP4 + gliflozins + basal insulin analogue;

Algorithm 2: stage 1 Metformin → stage 2 ­Metformin + gliflozins → stage 3 Metformin + gliflozins + agonists of ­glucagon-like peptide receptors (aGP-1) → stage 4 ­Metformin + gliflozins + aGP-1 + analogue of basal insulin;

Algorithm 3: stage 1 Metformin → stage 2 ­Metformin + aGP-1 → stage 3 Metformin + a fixed combination of basal insulin analogue (insulin glargine 100 U/ml) + aGP-1 (­Lixisenatide)

The model uses indicators of the efficiency of hypoglycaemic agents and insulin analogues for the control of DM2, taken from studies [14–28].

The calculation of the cost of the average daily dose, determined by the so-called average defined daily dose (DDD) [29], is presented in Table 2. Moreover, the average maximum selling price was calculated in case of the presence of several drugs for one international non-proprietary name (INN). For the drugs included in the List of Vital and Essential Drugs, the registered prices were considered (based on the state register of medicines [30]. For drugs not included in the state register of medicines, the wholesale prices of the Russian pharmaceutical portal PharmIndex [31] were used), for the combination of insulin glargine 100 IU/ml and Lixisenatide, the cost provided by the manufacturer was used. The daily cost was calculated considering VAT and the average wholesale premium, which was calculated based on wholesale price monitoring data [32]. For drugs of one pharmacological group, the average cost of treatment per day was calculated.

Table 2. Daily cost of therapy with hypoglycaemic drugs

Notes: * – drugs included in the List of Vital and Essential Drugs; **Included in the List of Vital and Essential Drugs since 2020, the cost is presented, that the manufacturer is going to register. INN – an international non-proprietary name; DPP-4 – dipeptidyl peptidase-4; aGP-1 – agonists of glucagon-like ­peptide 1; DDD – defined daily dose (average established daily dose); VAT – value added tax.

The average cost of GTT is 700 rubles, the cost of research of FPG is 270 rubles, the determination of HbA_1c is 650 rubles. [33–35].

The cost of treatment of complications and indirect costs associated with their development is presented in Table 3. When calculating indirect costs, we proceeded from the values of the shortfall in GDP per day, calculated based on the average salary in the Russian Federation in 2018, equal to 43724 rubles/month or 1457.47 rubles/day.

Table 3. Cost of hypoglycaemia and treatment of complications of DM2* and the indirect costs associated with their development

Notes: TD – temporary disability; CSG – clinical and statistical groups; *– tariffs are calculated in accordance with the General Tariff Agreement (GTA), taking into account the duration of treatment for CSGs specified in the GTA [36]; **– mild hypoglycaemia, does not require an ambulance or hospitalisation of the patient and is stopped by eating foods with a high glucose content. The patient is assumed to use an average of three test strips, costing 19.85 rubles/piece, for each episode of hypoglycaemia. The cost of inpatient treatment of hypoglycaemic syndrome, according to Appendix No. 4 to the GTA, is 15,180.00 rubles. (General Tariff Agreement, Compulsory Health Insurance, Complete Medical Insurance, 2019 [36])

When calculating indirect costs, the following were considered:

-shortfall in GDP due to permanent disability due to the onset of disability at working age;

-shortfall in GDP due to temporary and permanent disability.

Indirect GDP gap was calculated based on the per capita GDP of the Russian Federation (1), as well as average daily wage (2) (Table 4).

Table 4. Data for the calculation of indirect costs

Notes: * perspective parameters; GDP – gross domestic product; RF – Russian Federation.

The following are the indirect cost calculation formulas:

GDP (rubles/person/year) = (average salary in 2017 + average salary in 2018) /2 /30 days (2)

Microsoft Excel was used for calculations.

RESULTS

In 810 DM2 patients from the NATION database, the diagnostic significance of HbA_1c and FPG tests was compared, and the GTT served as an ‘arbiter’ in verifying the diagnosis. With the same glucose challenge during GTT, the difference between the minimum and maximum values of the level of FPG was 37%, but it was only 19% in the HbA_1c test (Fig. 1). When standardising the data, the total number of negative responses was significantly larger in the FPG test than in the HbA_1c test (77.4% and 36.5%, respectively). In addition, the number of false negative results was three times higher in the FPG test than in the HbA_1c test (Table 5). The HbA_1c test was three times more accurate in determining a carbohydrate metabolism disorder than the FPG test. In addition, the frequency of false-positive responses was slightly higher in the HbA_1c test than in the FPG test. Therefore, we can conclude that the FPG test determines the norm better, but it diagnoses a carbohydrate metabolism disorder much worse than the HbA_1c test, which is also confirmed by different sensitivity values of these tests. Therefore, we can assume that the HbA_1c test meets the requirement of timely diagnosis of carbohydrate metabolism disorders despite its higher cost than the FPG test. As for the greater frequency of false-positive results in the HbA_1c test than the FPG test, the costs for them are more than compensated with the prevention of various macrovascular complications in patients diagnosed timely with the HbA_1c test and with the early treatment of DM2. A false-negative result of the FPG test in many cases can lead to the fact that the disease is not diagnosed during its onset, the patient is allowed to go home and he seeks medical help again, possibly a few years later. By this time, the patient has already had complications that result not only in deterioration in the quality of life but also entail costs for health and society as a whole. A NATION study revealed that half of the patients identified were unaware of their DM2, and another Russian study showed that only 25% of patients in the group of DM2 patients aged 40–59 years had no complications, which is exactly one of the consequences of untimely detection of the disease [10].

Fig. 1. Spread of fasting glucose (A) and glycated haemoglobin (B) at equal glucose challenge levels.

Таблица 5. Результаты сравнения анализов глюкозы крови натощак и гликированного гемоглобина для диагностики нарушений углеводного обмена (n=810)

Notes: PD – prediabetes; DM – diabetes mellitus; PPV – positive predictive value (share of all true positive predictions relative to all positive predictions made); NPV – negative predictive value (share of all true negative predictions in relation to all negative predictions made).

If only the utilitarian aspect of the case (the cost of performing the tests) is considered, we would conclude that 27,000 thousand rubles would be spent per 100 patients in FPG test, whereas the investment efficiency expressed in terms of the coefficient (the ratio of costs of disorders identified to false results) would be 18,225 rubles./8775 rub. = 2.08, and for the HbA_1c test, the same indicator would be equal to 33,800/36,200 = 0.93. With this approach, the FPG test is the more economic choice for diagnosis instead of the HbA_1c test. However, such a conclusion should be confirmed by the clinical and economic results of the diagnosis consequences, which is much more accurate in one case (in the case of diagnosis using HbA_1c) than in the other (FPG) case.

In clinical and economic modelling, the probability of DM2 complications in the time horizon of 20 years was predicted, and the cost of their therapy for a conditional patient with characteristics from the NATION database was calculated (Table 6).

Таблица 6. Усредненные моделированные затраты на одного пациента на период 20 лет (руб.)

Notes: SIF – Social Insurance Fund, FPG – fasting plasma glucose, GTT – glucose tolerance test, GDP – gross domestic product, HbA_1c – glycated haemoglobin.

In simulation, the probability of developing DM complications was calculated, some of which are life threatening. The following complications, most often resulting in disability, were selected for inclusion in the model:

- coronary heart disease, angina of effort;

- acute myocardial infarction;

- acute cerebrovascular accident;

- heart failure;

- diabetic foot syndrome.

Figure 2 demonstrates the changes in time of the probability of disability, which is the sum of probabilities of the above DM complications (in other words, the graph shows the probability of any or several complications leading to disability), represented by the cumulative sum separately for the two tests compared. As shown in Fig. 2, the increase in probability of disability is higher with the FPG test than with the HbA_1c test, which is explained by the lower sensitivity of this test for the early diagnosis of DM2 and by the large number of patients with undiagnosed diabetes.

Fig. 2. Estimated changes in the number of patients with disabilities over a 20-year period (modelling based on the NATION base).

Calculations reveal that with all the proposed patient management algorithms with the disease diagnosed timely using the HbA_1c test, the costs are lower than in the case of a delay in diagnosis when using the FPG test. Thus, the average cost per patient for diagnosis using the FPG test was predicted to be higher by 22.5%–24.9% than that using the HbA_1c test, depending on the algorithms considered. The greatest difference is determined by the costs for treatment of complications (3,317,908 rubles/patient for fasting glucose test and 2,206,003 rubles/patient, respectively). The costs of arresting complications with any algorithm in the HbA_1c group are lower than those in the FPG group.

Indirect costs in the group of DM2 diagnosing using a glucose test accounted for 67.2%–71.0% of the total cost, whereas those in the group with HbA_1c diagnosed ranged from 61.1% to 66.1%. At the same time, the largest percentage of total costs was noted according to algorithm 1, which may indicate a more effective contribution of treatment with ‘aGP-1 + basal insulin’ to prevent complications.

DISCUSSION

The WHO recommendations adopted in 1999 are currently used in most countries of the world, including Russia, when screening for the detection of carbohydrate metabolism disorders [37]. In the Russian Federation, fasting glycaemia (FG) determination for preventive medical examination and periodic screening is a common screening test. However, studies on the prevalence of DM2 and other disorders of carbohydrate metabolism, depending on the diagnostic criteria and the diagnostic method (fasting glycaemia or oral GTT (OGTT)), showed that failure to conduct OGTT decreases the detection rate of early disorders of carbohydrate metabolism. This finding was demonstrated during population studies conducted in the Moscow region [38]. Given the international experience in DM diagnosis, which proceeds from the position on a high degree of correlation of HbA_1c values and FG values and 2 h after the load, as well as the fact that HbA_1c indicates the condition of chronic hyperglycaemia and, accordingly, predicts better the risk of vascular complications of diabetes, the inclusion of a study on HbA_1c level for screening DM and other carbohydrate metabolism disorders could improve significantly the situation with timely detection of patients in the Russian Federation, especially in high-risk groups (hereditary factor, age, overweight, etc.).

Our data obtained using Russian epidemiological material (the results of a large-scale population study NATION) coincide with data obtained in other countries, which also reveal greater diagnostic significance of the HbA_1c test in comparison with the FPG test. Thus, the prevalence rates of prediabetes and DM2 in Vietnam when using the HbA_1c test (which was conducted with the use of qualified experts in a sample of 3523 patients) were 34.6% and 9.7%, respectively, and those with the use of the FPG test were lower, only 12.1% and 6.3%, respectively [39].

Economic modelling of changes in time of DM2 development and its complications with timely (using the HbA_1c test) diagnosis of the disease and its comparison with the FPG test with a large number of false-negative results, and, therefore, the delayed start of treatment, was performed for the first time. References to the high cost of the HbA_1c test, as evidenced by the study, in Russian healthcare conditions, may become inconsistent because of the compensation of diagnostic costs with lower costs for preventing the disease complications timely in the future. The reduction in indirect costs due to the effective control of DM2 is also especially important to consider. Nevertheless, DM2 is a steadily progressing disease; therefore, even highly effective methods of controlling it with hypoglycaemic agents cannot prevent its complications. For effective prevention, lipid-lowering therapy and nephroprotective, anti-hypertensive and other anti-pathogenetic agents are necessary.

Study limitations

The analysis was performed based on modelling using the average cost of drugs by pharmacological groups and the proposed algorithms which number is not limited to those presented in the work. When predicting complications and their costs, the effects of hypolipidemic, anti-hypertensive and nephroprotective drugs that should be prescribed within the control of DM2 were not considered. Economic calculations were made subject to the tariff agreement.

CONCLUSION

1. The HbA_1cdetermination in comparison with the FPG test for the initial diagnosis of carbohydrate metabolism disorders (prediabetes and DM2) is characterised by high sensitivity and few false-negative results in the Russian patient population. As a result, the HbA_1c test has diagnostic advantages over the FPG test during the initial examination (medical examination). Further study of this issue is necessary to increase the diagnostic significance of the HbA_1c test while narrowing the circle of potential patients, for example, in high-risk group(s) for impaired carbohydrate metabolism.
2. The average cost per one patient, including the cost of the tests considered and the cost of DM2 complications, is lower when used for the initial diagnostics of the HbA_1ctest compared with the determination of FPG when considering various strategies (algorithms) for timely therapy, up to 27% of the costs for the prevention of DM2 complications with treatment initiated timely.
3. Modelling with a high degree of probability reveals that the increase in the onset of disability when using the FPG test for diagnostics (from 27% in year 10 to 73% in year 20) is higher because of untimely diagnostics and the presence of complications at the time of diagnosis than when using HbA_1cfor the same purpose (from 23% in year 10 to 52% in year 20).
4. The indirect costs for all the modern treatment algorithms considered in the work are predicted to be lower in patients in whom the diagnosis was performed using the HbA_1ctest, while these costs dominate in the overall cost structure.

ADDITIONAL INFORMATION

Source of financing. The work was not funded.

Conflict of interest. Professor O.I. Karpov is an employee of Sanofi Russia. Authors declare no obvious or potential conflicts of interest related to the publication of this article.

Contribution of authors. M.V. Shestakova developed the concept and edited the manuscript; A.S. Kolbin developed the concept, formed the analysis plan and wrote the article; O.I. Karpov developed the concept, performed the data analysis and wrote the article; G.R. Galstyan edited the article; A.Yu. Mayorov edited the article and provided scientific advice; M.A. Arepieva performed mathematical modelling. A.A. Kurylev performed analysis of the data and wrote the article; M.A. Proskurin performed data processing and mathematical modelling.