Decisions in the Psychology of Glucose Monitoring

Abstract

This commentary aims to discuss the parallels between nonadherence to continuous glucose level monitoring and nonadherence to medication in people with diabetes and to investigate specific reasons for the difficulties involved in glucose monitoring. To this end, examples are given from both continuous and discontinuous glucose monitoring (CGM and SMBG, respectively).

In general, when people decide to do something, they generally have a reason to do so. As Figure 1 shows, such reasons involve knowledge, skills, beliefs, emotions, desires, and visceral states such as pain and pleasure. ¹ There are also routines that people become used to, avoiding the need for a decision; for instance, in a study investigating the motives for self-monitoring of blood glucose (SMBG) in people with type 1 diabetes, 33% of patients tested on suspicion of hypo- or hyperglycemia; 44% conducted routine checks; and 23% tested on a combination of both motives. Moreover, people may carry out routine checks for a reason or because it is associated to a certain behavior: in this study, it was found that severe hypoglycemia in the preceding year and smoking, respectively, were positively and negatively associated with routine testing. ²

Figure 1.

Reason, decision, and action.

Sometimes, while people have a good reason for doing something, they do not accomplish the corresponding action. When people with chronic conditions demonstrate this type of behavior toward their treatment, it is referred to as “nonadherence.” ³ Two types of nonadherence have been identified. ⁴ The first is intentional nonadherence, which involves the patient making an active reasoned decision not to adhere to the medication advice by, for example, not buying the medicine. The second is unintentional nonadherence, which is a passive act being caused by forgetting to take the medicine, having no access to the medicine, or misunderstanding the prescription. A high rate of unintentional nonadherence has been found in patients who have low health literacy and difficulty grasping the meaning of medical concepts, while high rate of intentional nonadherence has been found in people with adequate health literacy. ⁵ Vrijens et al proposed another medication adherence taxonomy that distinguished initiation (first dose), implementation (how the medication is taken), and discontinuation, with persistence being the time spent between initiation and discontinuation, and implementation being quantified on the basis of the following: (1) the proportion of the prescribed drug taken, (2) the number of days that the correct number of doses is taken, (3) the proportion of doses that are taken on time in relation to the prescription-defined time interval between successive doses, (4) the distribution of the interdose intervals, (5) the number of drug holidays, and (6) the longest interval between two doses. ⁶

This commentary suggests that all these definitions could also be applied to glucose monitoring, a field wherein nonadherence has recently attracted considerable research interest.^7-10 Therefore, the aim of this commentary is to present the parallels between the nonadherence to continuous monitoring and the nonadherence to medication and to investigate specific reasons for the decision-making difficulties involved in glucose monitoring. While this commentary mainly focuses on continuous glucose monitoring (CGM), some examples are also given for the discontinuous self-monitoring of blood glucose (SMBG).

Nonadherence to Medications and Glucose Monitoring: A Parallel Overview

Progressive Decline in Adherence Rates

Medication adherence is often quantified using the medication possession ratio, which is the ratio of the total supply of medication dispensed divided by the number of days in the evaluation period. Whatever is the medication, this ratio has been found to progressively decrease to less than 80%—the threshold that often defines nonadherence—within 6 months of treatment initiation. ¹¹ The same figure has been found for CGM adherence, which was defined when a working sensor was available and was switched on as a proportion of the available time over six months (Figure 2). ¹² A study on medication adherence found that it was unintentional nonadherence that tended to increase during the first six months of therapy, ¹³ indicating that nonadherence was not a deliberate patient choice but owing to factors such as simply forgetting. Therefore, it could be interesting to determine how this type of nonadherence also applies to glucose monitoring, since this kind of adherence may be improved by reminders.

Figure 2.

Decline in adherence to medication and sensor use over 25 weeks (175 days, 6 months) from initiation. Adherence to medication, dotted lines: data (medication possession ratio, %) were drawn from Figure 2 of Curtis et al, ¹¹ for diabetes medications (squares), thiazide diuretics (diamonds), and glaucoma medications (triangles). A percentage below 80% defines nonadherence. Adherence to CGM, circles, continuous line: data (defined by having a working sensor available and by the time it was switched on as a proportion of available time, mean of the week) were drawn from Figure 1 of De Bock et al. ¹²

Effect of Age on Adherence

It has been found that older people tend to adhere more to medication regimes,^14,15 and it has been suggested that taking medication for long periods of time can be discouraging for younger people. ¹⁶ In the same vein, in the study quoted above, when the motives for SMBG were investigated, it was found that the people who routinely checked their blood glucose were older than those who only checked their blood glucose when they suspected hypo/hyperglycemia. ² Adolescents have been found to be especially at a risk of nonadherence in an evaluation of the number of days of insulin supply per annum in people with type 1 diabetes. ¹⁷ Similarly, the rate of CGM adherence (sensor available and switched on) was lower in patients aged 12-18 years (69%) than in children aged <12 years or in people aged ≧ 18 years (72% and 88%, respectively). ¹²

Effect of Nonadherence on HbA1C

Nonadherence to oral antidiabetic agents has been shown to reduce the decrease in HbA1c when the medication possession ratio is lower than 80% (justifying a posteriori this threshold for a definition of nonadherence). ¹⁸ Similarly, in a study on sensor-augmented insulin pump therapy, Hirsch et al observed that HbA1c improvements were significantly associated with sensor adherence, which was defined as the amount of actual sensor use over the expected use. Specifically, it was found that for every 10% increase in adherence, there was a 41% increase in the probability of a 0.5% reduction in HbA1c. ¹⁹ As is known, the first trial that evaluated the impact of sensor use on HbA1c found that improvements were only observed in adult patients and not in teenage patients, who were less adherent to the sensor use. ²⁰ A recent Swedish study found that less than 50% of people with type 1 diabetes performed SMBG ≧ 4 per day, which is the recommendation in the current ADA guidelines, and that there was an inverse relationship between the rate of daily SMBG and HbA1c. ⁸

Nonadherence and Health Expenses

Annually, between $100 and $300 billion of avoidable health care costs have been attributed to medication nonadherence in the United States, which is approximately 3-10% of total US health care costs. ²¹ There is also a concern that nonadherence to glucose monitoring could result in healthcare resource waste, which has been found to be lower for early discontinuation, where waste is calculated as per the unutilized portion of upfront costs than for continuing nonadherence, where waste results from both upfront and ongoing costs. ¹⁰

Adherence to Glucose Monitoring as a Marker of Adherence in General

Thus, it is possible to draw a parallel between nonadherence to glucose monitoring and that to medication because nonadherence is related to all prescriptions, ³ including those for discontinuously or continuously measuring blood glucose. For instance, Safford et al found that the frequency/day of SMBG self-testing was associated with the time spent on foot care, exercise, shopping and cooking, and total self-care and that people who never tested their blood glucose spent less time in self-care than those who measured it more than twice/day: 42 (4.64) vs 71 (35.95) min/day, respectively, mean (range) (P < .05). ²² Similarly, Telo et al compared two CGM patient groups: a “CGM group” of 120 youths interested in initiating CGM and a “Standard group” of 238 youths (from an initial cohort of 455) who participated in the study, but were not interested in initiating CGM. The CGM group was found to have lower HbA1c, higher SMBG (frequency/day), higher usage percentage of continuous subcutaneous insulin infusion (CSII), higher youth quality of life, less diabetes-specific family conflicts, higher percentage of two-parent families, and a greater adherence to diabetes care. ²³

Specificity of Decisions in Glucose Monitoring

However, patient decisions regarding glucose monitoring are more complex than those for taking medication (Figure 3). Specifically, glucose monitoring, regardless of whether it is continuous or discontinuous, has some steps that can be painful (finger prick or sensor insertion) or need reflection, and, as found in Tanenbaum et al’s article, there are complex adherence tasks that need to be performed at different time intervals (daily, weekly, monthly, and even yearly). ⁹

Figure 3.

Decisions in the psychology of glucose monitoring.

Reasons for Not Initiating CGM

In a recent survey ²⁴ of 533 adult people—84% of whom were being treated with an insulin pump—32% were male, 76% had a college degree, 85% had private insurance, and 55% had an income >$75,000. The main reasons for not initiating CGM were that it was too expensive (55.3% of participants), not covered by their insurance (39.5%), likely to be uncomfortable (35.5%), requiring a device to be attached to the body (27.6%), and likely to be painful to wear (9.2%). A further 13.2% said they were satisfied with SMBG; 10.5% considered the CGM to be less accurate than the SMBG; and 9.2% said that they were unfamiliar with CGM. These diverse reasons could be seen to be consistent with the resources, pain, beliefs and knowledge associated with the decision-making described on Figure 1. This study also found that the reasons people were using CGM were an expectation of better control and lower HbA1c, the possibility of avoiding hypoglycemia or hyperglycemia, and HCP recommendations.

This last point is critical because clinicians and patients may have different views on CGM. For instance, Tanenbaum et al found that the percentage of clinicians who were concerned by the cost, discomfort, or difficulties in understanding what to do with the information was higher than that of patients who expressed these concerns. ⁹

Invasiveness as a Barrier to the Implementation of Glucose Monitoring

Wagner et al designed a score that evaluated the degree of invasiveness associated with SMBG, which used questions such as how often do you skip checking your blood sugar level because of fear of pricking yourself, having to “milk” the fingertip for blood, or pain from the finger stick. They observed that this invasiveness score was inversely correlated to the percentage of SMBG adherence (P < .01). ²⁵ The nonadherence for these reasons could explain the success of FreeStyle Libre, which does not require any capillary blood glucose measurements as there is no need for calibration. It also supports the argument for implantable sensors ²⁴ and other attempts (unfortunately unsuccessful so far) to develop noninvasive glucose monitoring technologies.

Discomfort, Pain, Allergic Reactions, and Invasiveness as Reasons for Stopping CGM

In the study quoted above, ²⁴ previous patients gave several reasons for stopping CGM such as discomfort (47%), skin irritation due to the adhesive (41%), pain caused by sensor application (31%), and invasiveness (23%). In particular, the issue of skin irritation ²⁶ and allergies to the adhesives associated with the pumps and sensors, especially with FreeStyle Libre,^27,28 cannot be overlooked.

The Major Impact of Trust in the Result

As Figure 1 shows, beliefs is heavily involved in decision-making. In this conceptual framework, trust can be seen as the beliefs related specifically to the patient’s evaluation of the accuracy of the glucose measurement. De Bock et al found that any 1% increase in the sensors’ mean absolute relative difference (MARD) was associated with a 0.35% decrease in adherence (P < .001). ¹² A lack of trust in the results could also be a reason for discontinuing CGM. For instance, ex-users have reported significantly less satisfaction with the device’s accuracy (P < .001), with the main reasons for quitting real-time CGM being that the numbers could not be trusted; there were too many false alarms; or the device often stopped working. ²⁹ Tanenbaum et al also found that the primary reasons for quitting real-time CGM were a lack of accuracy and trust in the system. ³⁰

Decisions as the Consequences of a Deliberation

As with any technology, CGM has benefits—continuous data, better control, provision of trend information and identification of hypo and hyperglycemia—and hassles—discomfort, pain, invasiveness, too many alerts, and skin irritations. ³¹ Continued use may depend on how people weigh the benefits and hassles: a study that evaluated the frequency of sensor use by using a benefit subscale and a lack of drawbacks subscale of the CGM-SAT score found that people who used the device infrequently focused more on the hassles than on the benefits. ³²

As Figure 3 shows, besides the decision to use or not to use the monitoring system, patients are also required to make decisions based on glucose monitoring, such as modifying their insulin doses as per the glucose levels. Therefore, patients are required to not only have knowledge (see Figure 1) but also special skills, referred to as health literacy and numeracy, both of which affect the patient’s ability to correctly perform SMBG. For instance, if people with low literacy or numeracy levels are less able to identify values within the target range of 60-120 mg/dl (3.33-6.66 mmol/l), ³³ consider if they would be able to understand the “time-in-range” concept. Surprisingly, the impact of health numeracy and literacy on CGM use ³⁴ has not been fully evaluated yet.

However, decisions such as adjusting insulin doses as per glucose measurements are also emotionally difficult (Figure 1). We observed that in response to a high glucose result people with type 1 diabetes were less willing to increase their insulin doses in real-life conditions than in theoretical exercises, ³⁵ possibly because of previous fearful hypoglycemia experiences. Indeed, these bad instances may have a greater memory impact on the patient than normo- or hyper-glycemia (usually asymptomatic) instances, ³⁶ and this effect, in turn, can lead to decision avoidance.^37,38

Finally, these difficulties point to the need for the development of closed-loop insulin delivery systems so that this decision task is transferred to the system. However, here also, the decision to use such a system will largely depend on how patients weigh its benefits and hassles. For instance, terms such as “closed-loop” and “artificial pancreas” could be misleading, as people may expect these systems to reduce the burden of self-care. Here again, patient trust in the “decisions” made by the system will be critical.^39,40

Conclusions: Implications for Clinical Practice

Figure 4 illustrates the practical value of the adherence model shown on Figure 1: it makes it possible to propose avenues for optimizing decisions in the area of glucose monitoring. First, the caregiver must know the content of the mental states (ie, knowledge, skills, beliefs, emotions, and desires); this is the assessment phase of any patient esducation. ⁴¹

Figure 4.

Targets of patient education in glucose monitoring.

Next, patient education can improve the patient’s knowledge, for instance, by explaining why glucose measured by a subcutaneous system may differ from the blood glucose level, which will avoid a mistrust response that, as shown above, is a cause of stopping the use of the system. Patient education will develop skills concerning decision-making based on the results of glucose monitoring. One may modify patient’s beliefs by explaining the benefits of the use of the system: this can reinforce the desire to use it. Positive emotions such as pride will be used and negative emotions such as fear or shame will be combated. Miniaturized systems, less painful to use, can facilitate their use, and it is important to solve the problem of irritation and/or allergy. Regarding the resources available, the reimbursement of the systems can improve the decision to use them.

Finally, motivational interviewing techniques, reinforcing the sense of self-efficacy, and the use of shared medical decision-making can improve adherence. However, as discussed above, nonadherence to advices concerning glucose monitoring may also be unintentional: it may be possible to fight against the simple forgetting to use the system by using external help (family members) or electronic systems. Concerning adherence to medications, unintentional nonadherence occurs within the six first months following initiation. ¹³ If this holds true for adherence to glucose monitoring, this period should be used to demonstrate to the patient all the benefits of using the glucose monitoring system, and if possible help him/her to form habits that may alleviate the burden.