Blood sugar self-monitoring in type 2 diabetes - costs and benefits for the German healthcare system

The benefit of blood glucose self-monitoring in patients with type 2 diabetes is still controversially discussed in Germany, although no other diagnostic method has been examined so intensively in randomized, controlled studies, as these are more typical for drug evaluations. The economic evaluation of this diagnostic procedure has also largely followed the rules of drug evaluation up to now. For methodological reasons alone, however, this approach can lead to erroneous results.

>> For years, almost all countries have been struggling with the phenomenon of rising healthcare costs, regardless of the healthcare system they choose. In Germany, total spending rose from EUR 196 billion in 1997 to almost EUR 245 billion in 2006 and now accounts for almost 11% of the gross national product. The reasons for the increase in spending are varied; Not only medical progress and demographic factors such as the increasing aging of society contribute to this phenomenon, but also more general developments such as an increasingly unhealthy lifestyle and higher expectations of the healthcare system, among other factors, play an important role here. This is offset by society's limited willingness to finance high increases in the healthcare system, which ultimately amounts to budgeting the available resources. This in turn leads to the need for an economic evaluation of the costs and benefits of therapeutic procedures in order to be able to help with prioritization. However, such evaluations can only provide a decision-making aid. Health economics, like almost all other economic disciplines, is part of applied economics, which only explains why or how something is (or should or could be), but not whether something is “good” or “bad” for society. These statements are reserved for normative economics (welfare economics), which then approaches such questions with completely different approaches and concepts.
Diabetes mellitus in Germany
Diabetes mellitus has now become a widespread disease in Germany. Even if the GDR's diabetes register was not continued with German reunification, the number of diabetics in Germany can be estimated fairly reliably based on samples of insured people. It must currently be assumed that there are 7 million people in Germany with diagnosed diabetes mellitus, of which around 6 million have type 2 diabetes. Only 25 % of type 2 diabetics are treated purely with diet, around 50 % with oral antidiabetic therapy and 25 % with insulin or combination therapy. The full extent of the problem becomes clear when one takes into account that, according to estimates by the International Diabetes Federation (IDF), the total number of all diabetics in Germany (i.e. including previously undiagnosed diabetics) is closer to 11.8 % of the population and this number is expected to increase by 2025 13.3 % will increase. Younger age groups will increasingly be affected and the number of people with “pre-diabetes” (Impaired Glucose Tolerance, IGT) is not even taken into account in these figures - then another 7 % would probably be added to the population (International Diabetes Federation 2006).
Costs of diabetes mellitus
The costs associated with this development pose enormous challenges for the German healthcare system. According to a recent estimate, the costs for the treatment of an “average” diabetic in Germany are almost EUR 3,200 per year and patient, with the costs increasing continuously with the duration of the disease (Fig. 1a).
However, a large part of the costs are not attributable to the primary therapy (e.g. doctor visits, medication costs), but to the treatment of serious diabetes-related complications such as heart attacks, strokes or amputations, whereby a trend over time can also be observed here: the number increases with the duration of the diabetes disease in particular the follow-up costs for complications that have occurred (Fig. 1b).
Already today, the treatment costs for diabetes mellitus (depending on the prevalence taken as a basis) correspond to approx.
Blood sugar self-monitoring
Blood glucose self-monitoring (BZSM) has been an undisputed standard in the treatment of insulin-dependent diabetics for years. Recently, there has been increasing recognition that SMSM under certain conditions can also be a very useful therapeutic element in the management of non-insulin-treated type 2 diabetes (McAndrew et al. 2007; McGeoch and Moore 2007).
Against the background of the figures mentioned above, the question naturally also arises as to whether BZSM is just a useful therapy element, or whether it is also a cost-effective and affordable one.
However, the economic evaluation of the benefits of self-monitoring of blood glucose is associated with some problems. Unlike a pharmaceutical, BZSM cannot have a direct, intrinsic effect; it is a diagnostic procedure that requires the ability to correctly interpret the measurement result displayed and to derive an appropriate and sustainable action from it. In the case of patients with orally treated type 2 diabetes, this action is not necessarily limited to switching oral therapy or adding short-acting insulins to the regimen. A change in lifestyle (in the sense of altered eating and exercise behavior) can even have more far-reaching effects than increased drug therapy, since a change in lifestyle not only has a favorable effect on blood sugar levels, but also positively changes other cardiac risk factors such as blood pressure and blood lipid levels. Accordingly, it is also very important in which frequency and form the blood glucose self-measurement is carried out. Of course, erratic measurement at undefined times cannot provide the same information as structured measurement (for example, a 7-point daily profile on three consecutive days, once a month), which provides the patient with direct feedback on his or her behavior. This exact point is hardly documented at all in the randomized clinical trials (RCT) published to date. Frequency and measurement scheme are only specified precisely in a few cases and almost all RCTs on the use of SMSM are of short observation duration and are limited to the investigation of a surrogate parameter, namely the HbA1c change. However, this HbA1c value only represents an integrated measure of the circadian blood glucose values of the last six to eight weeks and monocausally influences only two of the typical diabetes-related complications (diabetic retinopathy and nephropathy). The much more frequent complications (myocardial infarction and stroke) also depend on the HbA1c value, but also on other factors such as blood pressure, blood lipid levels and short-term blood glucose peaks (Ceriello 2003). Furthermore, the complications caused by diabetes occur with a long time lag, sometimes on the order of years.
Except for one retrospective cohort study (Martin et al. 2006), there is no study examining the relationship between SMSM and mortality or morbidity.
Decision security and “value of
Information"
The existing study situation therefore leaves decision-makers largely uncertain about the benefits of BZSM with regard to patient-relevant endpoints. Despite the existing, insufficient information, there is still a need to decide whether this therapeutic element should be prescribed at the expense of the health insurance companies or not. Economically, this information deficit is referred to as the "Value of Information" (VOI). Weinstein and colleagues describe the value of information as “the difference between the expected consequences (utilities) of a decision based on a particular piece of information and the expected consequences (utilities) of making the decision without that information.” The VOI is also applicable to information that reduces uncertainty regarding a parameter in a decision, such as the effectiveness of a treatment, the duration of a therapy effect or the applicability to different patient groups. Postponing a decision until more information is available is one course of action, but at the expense of resources, delays and possible errors made or omitted in the meantime. “Resources can be wasted or patients harmed by choosing the more expensive or less effective treatment option while waiting for definitive evidence” (Weinstein et al. 2001).
This value of information can also be calculated in monetary units (Claxton 1999) and thus also represents an important element of decision-making. Unfortunately, this has not yet been reflected in the current methods of IQWiG. In my opinion, decision-makers such as the G-BA are thus deprived of an important decision-making element.
Importance of assumptions in economic analysis
In this situation of uncertainty, disease models that simulate the natural course of a disease are an important decision-making aid. Although there are very different modeling techniques, the results of the published models for diabetes mellitus show a high degree of agreement, provided that the same initial assumptions are used (2007; Brown et al. 2000). However, the results published to date on the cost-effectiveness of BZSM come to very different conclusions (Table 1).
The different results of the previous cost-benefit analyzes suggest that different assumptions were used in the cost studies. It is particularly problematic if the assumptions are based on one of the numerous meta-analyses on the clinical effectiveness of BZSM. Meta-analyses are rightly a highly valued element of evidence-based medicine (EBM), provided this statistical method is used correctly. No study is structured exactly the same. Even if the intervention is comparable per se, the studies differ, for example, in the number of patients. As a rule, in valid meta-analyses, this heterogeneity is corrected using statistical means, usually with an I2 statistic according to Higgins and Thompson. However, if the interventions are completely different, the statements made by these statistics are just as distorting and irrelevant as determining a mean and a standard deviation for data that is not normally distributed - you are comparing apples with oranges. It is precisely this effect that can be observed in the meta-analyses on BZSM that have been carried out so far. If you compare the RCTs used as a basis, you can see that in some studies there was hardly any training for the BZSM group, while in other studies this was intensive. In some studies, patients were allowed to change their therapy or behavior based on the measurement results, but in others they were not. In addition, when using meta-analyses, only the change in HbA1c value is taken into account. For the reasons mentioned above, the results of the cost-effectiveness studies almost always fall into the upper right-hand quadrant of the cost-effectiveness level (Fig. 2).
Impact on the German healthcare system
However, if one also takes into account that with a structured BZSM (e.g. Kempf et al. 2009) additional positive effects that are causally linked to the BZSM occur beyond the influence on the HbA1c value, a different picture emerges: the patients gain years of life, but cause less costs, which corresponds to the right, lower quadrant in Figure 2. Expressed in economic terms, the "incremental cost-effectiveness ratio" (ICER) becomes negative1 (Table 2).
In addition to the cost-benefit analysis, an essential element of an economic analysis should always be a “budget impact analysis” (BIA) (Mauskopf et al. 2007), since the cost-effectiveness analysis is usually only a relative measure, namely an ICER provides only limited information (Birch and Gafni 2006), since it makes no statement about the “affordability” of a measure. It may well be that a new therapy method proves to be cost-effective in a cost-effectiveness calculation according to national standards, but ultimately proves to be "unaffordable" due to the large number of patients who should or could then benefit from the new therapy. According to such a calculation based on the results of the cost-effectiveness calculation shown in Table 2, there is even significant savings potential for the German healthcare system (Fig. 3).
Conclusion
Hardly any other diagnostic method has been examined as frequently in randomized clinical studies as blood glucose self-monitoring. Nevertheless, the results remain contradictory, which is essentially due to the different ways of use. With a structured blood sugar measurement, which can provide the patient with direct feedback on his behavior due to the measurement scheme, not only the blood sugar level but also other clinical parameters can be favorably influenced. However, the prerequisite is an intensive training program and the opportunity for the patient to react adequately to the measured values. In this case, self-monitoring of blood sugar is not only cost-effective, it is also associated with potential savings for health insurance companies.