Flexible insulin dosing improves health-related quality-of-life (HRQoL): a time trade-off survey

Abstract

Purpose:

People with insulin-treated diabetes often face strict regimens with inflexible dose timing, frequent injections, and frequent self-measured blood glucose (SMBG) testing. The objective of this study was to estimate the health-related quality-of-life (HRQoL) impact of these aspects using time trade-off (TTO) methods.

Methods:

HRQoL was examined via a TTO survey in the UK, Canada, and Sweden with separate analyses of 2465 respondents from the general population, 274 people with type 1 diabetes, and 417 people with type 2 diabetes. Respondents evaluated health states with diabetes, SMBG testing, and basal injections that were once-daily time flexible, once-daily at a fixed time, and twice-daily at a fixed time in a basal or basal–bolus regimen.

Results:

Time-flexible basal injections were associated with 0.016 and 0.013 higher utility vs a fixed time of injection for basal-only and basal–bolus regimens, respectively, as evaluated by the general population. The diabetes respondents confirmed the basal-only results with 0.015 higher utility, but the difference in utility was non-significant for basal–bolus. Once-daily injections had higher utility compared with twice-daily injections for basal (0.039 and 0.042) and basal–bolus (0.022 and 0.021) regimens, as evaluated by the general population and people with diabetes, respectively. Increased frequency of SMBG negatively affected health utility.

Limitations:

This study has the limitation that it measures hypothetical health states rather than the HRQoL of people with these health states; furthermore, it could be suggested that the web-based nature of this survey is biased towards literate respondents with internet access and IT competence.

Conclusions:

Flexible dosing and fewer injections have a positive HRQoL impact, which potentially may enhance therapy adherence and could contribute to improved long-term outcomes. The impact of flexibility is greater in people treated with basal-only insulin regimens, and diminishes if bolus injections are part of the treatment regimen.

Keywords::

• Diabetes
• Flexibility
• Insulin
• Quality-of-life

Introduction

Insulin therapy remains the most efficacious glucose-lowering therapy available for patients with diabetes1. However, patients view the rigidity of glucose-lowering treatment as a concern2 and many are reluctant to initiate insulin therapy as it is perceived to be inflexible3,4. Insulin therapy can be particularly cumbersome with the requirement for basal insulin dosing at the same time every day, and for some people twice a day. Fixed dosing regimens may have a negative impact on health-related quality-of-life (HRQoL) and potential non-adherence. Moreover, for patients with type 1 diabetes and those with type 2 diabetes requiring treatment intensification, frequent injections of fast-acting bolus (meal-time) insulin are needed in a basal–bolus regimen. The potential burden of insulin therapy on HRQoL is associated with greater injection frequency2. Several studies have investigated barriers to insulin adherence, as reviewed by Davies et al.5, with patients finding it difficult to incorporate insulin injections into daily life, due to changing daily routines, thus making therapy adherence a challenge5–7. Indeed, a recent study showed that one third of patients reported at least one occurrence of insulin omission or non-adherence during a 1-month period. The most common reasons for this were: too busy, travelling, skipped meals, stress/emotional problems, or public embarrassment7.

Poor glycemic control and diabetic complications are positively associated with missing insulin doses8,9, while decreased medication adherence has been associated with increased healthcare service use and poor health-related outcomes in patients with type 2 diabetes10.

Insulin treatment of diabetes often involves drawing blood from a finger prick to conduct self-measured blood glucose (SMBG) testing at various intervals throughout the day, which is regarded as burdensome by patients and an interruption to daily activity; ultimately, this can have a negative impact on HRQoL2. Treatment guidelines recommend that people with diabetes using multiple daily insulin injections or insulin pumps perform SMBG testing before meals at the very least, which equates to three or more measurements per day11. People using less frequent insulin injection therapy or non-insulin glucose-lowering therapy are also recommended to perform SMBG testing to help guide treatment decisions and patient self-management11,12. Furthermore, blood glucose test strips represent a significant direct cost to healthcare systems, accounting for 24–41% of the total pharmacy-related costs in insulin-treated patients with diabetes in the US, Canada, and Germany13–15.

Different insulin analogs vary in terms of the number of injections, flexibility, and the necessity of SMBG testing. This should be taken into account when evaluating the cost-effectiveness of one insulin treatment over another. Insulin regimens that are more flexible may be easier for patients to adhere to and could improve HRQoL. HRQoL can be quantified using health utility values, which are on a scale from 0–1, where 0 = death and 1 = perfect health. Boye et al.16 previously estimated HRQoL values for different features of diabetes treatment using standard gamble (SG) methods, but the study did not specifically include insulin-treated patients.

In this study, the time trade-off (TTO) method was used to examine the HRQoL impact of various features of insulin treatment. It was conducted in three countries and included a sample of the general population as well as people with diabetes. The TTO method estimates the health utility of a certain disease state and is a well-recognized and extensively used tool in health-economic assessments17–19. Utility values are obtained by asking respondents to ‘trade off’ a portion of their remaining life-span for an improved health state. Methods such as TTO are recognized by the Health Technology Assessment (HTA) agencies and are used to calculate quality-adjusted life years (QALYs) and in processes to estimate the cost-effectiveness of particular treatments20–24. The aim of this study was to assess the impact on HRQoL of flexible timing of basal insulin doses, the number of basal insulin doses, and the frequency of SMBG testing, using a TTO utility approach in a population.

Patients and methods

Study design

The TTO methodology has been well described elsewhere and the methodology for this study follows that described by Evans et al.24. Briefly, each respondent is asked to make a choice or be indifferent about two health states: a defined health state for the remainder of their lifetime or perfect health for a shorter lifetime. The shorter lifetime is then changed and the question repeated until the point of indifference is reached (where both options are equally acceptable)19. For example, option A: you have diabetes and give yourself one injection at the same time every day and will live for 40 years, or option B: you will live in full health for 36 years. If the respondent chooses option B they are willing to trade 4 years to have full health, and the utility is 0.9 (36/40 years). To identify the point of indifference (i.e., where two health states are equally acceptable) respondents were asked the same questions repeatedly, varying the number of years to trade off for the improved health state. This procedure followed standard bisection methodology, using a starting point of utility of 0.6 and after four questions reducing the utility value to an interval of 0.05. Particular attention was given to the distribution tails. Respondents who either chose not to trade lifetime at a utility value of 0.95, or who were willing to trade a very high proportion of their remaining lifetime (0.875) to be restored to full health, were carefully screened and received additional questions. Responses were excluded if the respondents refused to trade on ethical or religious grounds or if they did not understand the question. However, those who believed the health state manageable or who stated a desire to live as long as possible due to obligations (e.g., caregivers) were retained. Negative utilities were not allowed in this study. To ensure the trade-off values were as accurate and relevant as possible, the time horizons used in the survey were based on each respondent’s projected life expectancy. These were calculated using World Health Organization life tables accounting for respondent age and gender25.

Survey description

Data were collected via an internet-based survey, using existing email panels of respondents. All respondents had previously agreed to participate in online surveys, were >18 years old and participated at their discretion, remaining anonymous throughout the study. The patients with diabetes were also sampled from online panels as the panel providers have panels with disease characteristics such as diabetes from which the respondents can be sampled from.

Depending on the country of origin, remuneration (nominal value €1–2) for participating was offered via ‘points’ for online shopping or entry in a draw. In order to improve the quality of the answers and prevent unconsidered responses, a delay feature was introduced to pages containing a large amount of text such that respondents were unable to click the ‘Next’ button until 10 s had passed.

The survey was conducted in the UK, Sweden, and Canada and was performed in Swedish, English, and French. Canadian respondents could choose which version of the questionnaire to answer (French or English). The questionnaire was back-translated to ensure the accuracy of each translation. Respondents were asked to provide demographic information (age, sex, employment, household, regional details). Those with diabetes were asked to disclose details on their disease duration and medication.

There remains a debate in the literature, and there is also a difference in opinion between HTA bodies, as to whether patient or public preferences carry the most influence when determining the value attached to a particular health state26–28. In order to determine whether responses differed between the general population and those respondents who may have personal experience of using insulin, a second population with type 1 diabetes or type 2 diabetes were also identified from the available panels. Therefore, the study included three samples: general population (including some respondents with diabetes), a sample of respondents with type 1 diabetes (including those with type 1 diabetes who were also included in the general population and answered the basal–bolus questionnaire), and a sample of respondents with type 2 diabetes (including those with type 2 diabetes also included in the general population) (Figure 1). A similar study design was used by Evans et al.24.

Figure 1. Patient consort flow diagram. pop, general population; TTO, time trade-off; T1D, type 1 diabetes; T2D, type 2 diabetes. There were a number of ‘reusable respondents’ who were included in the general population but also declared that they had diabetes. Of the 31 respondents in the general population with type 1 diabetes; seven were also included in the type 1 diabetes analysis because they had answered the basal–bolus questionnaire and the six daily SMBG questionnaire. Additionally, all 176 respondents in the general population cohort with type 2 diabetes were also included in the relevant analyses (basal-only or basal–bolus) of the diabetes populations.

To validate the functionality of the questionnaire, a pilot survey of 200 respondents was conducted in Canada in English. This pilot survey showed that a warm-up question was needed in order to make respondents familiar with the TTO methodology. It was observed that respondents were answering differently to the first TTO question compared to the following ones, making the results of the first TTO exercise unreliable. Accordingly, a warm-up TTO exercise was added to the questionnaire and to further improve the reliability of the results the order of the subsequent three TTO exercises about injections and flexibility were randomly asked. To test the revised questionnaire a second pilot survey with another 200 respondents was performed (also in Canada); results were as expected, so the main survey was initiated.

Two versions of the questionnaire were used (examples of the health states assessed are depicted in Table 1):

• Basal-only: in this version respondents were to imagine having baseline diabetes and administer basal insulin injections according to the description in the health state.

• Basal–bolus: in this version the respondents were asked to imagine having baseline diabetes and administer basal insulin injections according to the description in the health state, whilst also taking a bolus insulin injection with every meal.

Table 1. Description of health states.

Introduction to the health states (all respondents)	In the following pages you will be presented with 3 different health states. In all health states you will have diabetes, as described. The only differences will be with respect to the flexibility and frequency of your insulin treatment.
	Some diabetes medications have to be taken at the exact same time of day every day, e.g., at 10 pm, 7 days a week. Other diabetes medications are more flexible and can be taken at any time of day and at different times of day from day to day. If preferred, the flexible medication can of course also be taken at the same time of day every day.
	If you do not take the medication as described, the insulin treatment may not be as effective and you will/may increase the risk of developing long-term complications, such as eye damage, kidney problems, nerve damage (numbness or pain in your hands and feet) and heart disease.
Baseline diabetes (all respondents)	Now, we would like you to imagine that you have a health condition called diabetes.
	• This means that your body cannot keep your blood sugar at a constant level.
	• To control this, you often follow a special diet, are careful about eating regularly and take steps to have something sweet nearby.
	• You take medication on a daily basis and sometimes need to plan your life around food and medicine.
	• You need to check your blood sugar from time to time.
	• You need to consider your diabetes when you are planning to exercise, travel, go out with friends, and drive.
	• Your diabetes does not affect your work/study, and you don’t have any problems looking after yourself.
	• You occasionally worry about the effects of your diabetes on your day-to-day life.
	Your diabetes is well-controlled and you do not experience hypos (too low blood sugar levels).
Bolus description (included in all three health states, but only the bolus version of the questionnaire)	In order to control your blood sugar levels, you must give yourself insulin injections with each of your meals.
One fixed injection  health state	You must give yourself one insulin injection each day. It is important that the injection is taken at the same time of day each day.
Two fixed injections  health state	You must give yourself two insulin injections each day, one injection in the morning and one injection in the evening.
	It is important that the injections are taken at the same time of day each day both in the morning and in the evening e.g., 7 am and 10 pm.
One flexible injection  health state	You must give yourself one insulin injection each day. This can be taken at a time of your choosing.
Self-monitored blood glucose testing	Imagine that you have diabetes as described. In order to control your blood sugar levels, you must give yourself insulin injections with each of your meals. You must also give yourself one additional insulin injection each day. It is important that the injection is taken at the same time of day each day. To control your blood sugar, you would have to test your blood sugar 3-times a day.

To avoid fatigue, each respondent answered one version of the questionnaire (basal-only or basal–bolus). Those in the population with type 1 diabetes all answered the basal–bolus questionnaire (as this would be the most relevant for them). Half of those in the general population and the population with type 2 diabetes answered the basal-only questionnaire and the other half answered the basal–bolus questionnaire.

Health states

The health states used are shown in Table 1. All respondents were shown the baseline diabetes health state which was used to familiarize participants with the TTO method. The baseline diabetes health state was developed based on patient questionnaires described in Evans et al.24. This description of the health state with bolus insulin injections was only shown in the bolus version of the questionnaire. Three different health states were evaluated to assess the impact of flexibility and the number of daily basal insulin injections:

• Once-daily flexible injection,

• Once-daily fixed injection, and

• Two fixed injections.

The disutility of injections and flexibility were calculated by subtracting the utility values of these different health states. To assess the impact of SMBG testing, respondents were randomly allocated to evaluate one of the following health states: once-weekly, once-daily, three times daily, six times daily SMBG tests.

Exclusion criteria

The exclusion criteria were applied sequentially, with respondents failing the test question excluded first. The test question asked respondents to choose between living for a longer time in full health or living with diabetes for a shorter time. Those who prefer to live with diabetes for a shorter time are screened out because they probably have not read or understood the question properly. Second, respondents completing the general population version of the questionnaire in less than 5 min or the diabetes version in less than 7.5 min were excluded. Third, respondents could be excluded if there was an inconsistency in their responses. It was required that a valid utility value exist for all three health states concerning flexibility and injections for each respondent to be included in the analysis (Figure 1).

Ethical approval

This study was performed in accordance with the code of conduct of the European Pharmaceutical Market Research Association. In Canada, ethical approval was gained from Institutional Review Board Services (Aurora, Ontario, Canada). In the UK and Sweden, an ethical application was not necessary for this type of survey, as it was not a clinical trial, and did not collect human or biological samples or identifiable personal information.

Statistical analysis

All statistical analyses were performed using SAS® version 9.2 statistical software. A utility value was assigned to each health state based on each individual response; the mean utility value across respondents was then calculated for each health state.

For SMBG measurements, an ordinary least squares (OLS) linear regression model was estimated with individual level data to capture the utility value of one less SMBG test. The model estimated utility as a function of annual number of SMBG tests with a binary variable for regimen. A number of tests were conducted and a non-parametric bootstrapping method was used to make sure the distributional assumptions were not violated. As the data were not normally distributed, bootstrapping was used to calculate standard errors and confidence intervals (CI)29. As recommended, the estimation of each utility was repeated 10,000 times by drawing the original number of observations from the dataset with replacement 10,000 times. The CI and standard deviation (SD) were then taken from these distributions.

Results

Respondent demographics

The patient consort flow is summarized in Figure 1, including details of those who were excluded. A total of 3384, 420, and 385 respondents in the general, type 1 diabetes, and type 2 diabetes populations, respectively, started the questionnaire. Data from 2465 respondents (73%) from the general population, 274 respondents from the population with type 1 diabetes, and 417 respondents from the population with type 2 diabetes were used for the analysis (Figure 1). In the general population, 31 (1.3%) and 176 (7.1%) respondents disclosed that they had type 1 or type 2 diabetes, respectively. The numbers of respondents used for the diabetes population analysis also includes some respondents included in the general population as the diabetes data was used in both samples; hence the higher number of respondents with type 2 analyzed than started the questionnaire. Of the 31 respondents in the general population with type 1 diabetes; seven were also included in the type 1 diabetes analysis because they had answered the basal–bolus questionnaire and the six daily SMBG questionnaire. Additionally, all 176 respondents in the general population cohort with type 2 diabetes were also included in the relevant analyses (basal-only or basal–bolus) of the diabetes populations. The median answer time was 13 min for the general population, 18 min for those with type 1 diabetes, and 17 min for those with type 2 diabetes.

The demographic characteristics of the final sample of respondents used for statistical analysis are summarized in Table 2. Approximately equal numbers of respondents came from the three participating countries: Canada, Sweden, and the UK. The characteristics of the general population and diabetes populations were similar, except that the respondents with type 1 diabetes had a younger average age (mean 43.15 years) than the general population (mean 47.51 years), whilst those in the type 2 diabetes population were on average older (58.59 years) than the general population. The mean (±SD) disease duration for respondents with type 1 diabetes (n = 274) was 16.6 (±14.4) years and for those with type 2 diabetes (n = 417) was 8.7 (±7.9) years.

Table 2. Background characteristics of the final sample.

	General population		Type 1 diabetes		Type 2 diabetes
	n	(%)	n	(%)	n	(%)
Gender
Male	1205	49	143	52	249	60
Female	1260	51	131	48	168	40
Total	2465	100	274	100	417	100
Age group
18–29	477	19	63	23	16	4
30–44	586	24	86	31	35	8
45–59	648	26	88	32	139	33
60+	754	31	37	14	227	54
Total	2465	100	274	100	417	100
Labor market status
Going to school or college full-time	159	6	13	5	2	0
In paid employment or self-employed	1064	43	134	49	131	31
Doing unpaid work for own business	42	2	7	3	5	1
Looking for paid work or a government training scheme	111	5	16	6	13	3
Permanently unable to work (sickness/disability)	158	6	32	12	46	11
Retired from paid work	543	22	33	12	175	42
Looking after home or family	214	9	23	8	27	6
Doing something else	174	7	16	6	18	4
All	2465	100	274	100	417	100
Household
One adult, no children	584	24	47	17	95	23
One adult, children	113	5	19	7	11	3
Two adults, no children	985	40	103	38	224	54
Two adults, children	628	25	86	31	68	16
Other	155	6	19	7	19	5
All	2465	100	274	100	417	100
Diabetes
Yes	224	9	274	100	417	100
No	2241	91	0	0	0	0
All	2465	100	274	100	417	100
Type of diabetes
Type1	31	14	274	100	0	0
Type2	176	79	0	0	417	100
Another type	5	2	0	0	0	0
Don’t know	12	5	0	0	0	0
Total	224	100	274	100	417	100
Country
UK	800	32	103	38	129	31
Sweden	818	33	69	25	127	30
Canada	847	34	102	37	161	39
Total	2465	100	274	100	417	100

TTO results

In the general population, once-daily, time-flexible basal insulin injections were associated with 0.016 [95% CI: 0.011; 0.022] and 0.013 [95% CI: 0.007; 0.020] higher utility vs a once-daily fixed time of injection for basal-only and basal–bolus regimens, respectively (Table 3). The diabetes respondents confirmed the basal-only results (0.015 [95% CI: 0.004; 0.027]), but elicited a smaller utility for time-flexibility in a basal–bolus regimen (0.004 [95% CI: −0.006; 0.04]) (Table 3).

Table 3. Utility differences from the time trade-off survey.

	Basal-only			Basal–bolus
	n	Utility	95% CI	n	Utility	95% CI
Population
One flexible vs one fixed	1121	0.016	0.011; 0.022	1144	0.013	0.007; 0.020
One fixed vs two fixed	1121	0.039	0.032; 0.046	1144	0.022	0.016; 0.028
One flexible vs two fixed	1121	0.055	0.048; 0.063	1144	0.036	0.029; 0.043
Diabetes populations*
One flexible vs one fixed	192	0.015	0.004; 0.027	465	0.004	−0.006; 0.014
One fixed vs two fixed	192	0.042	0.025; 0.061	465	0.021	0.013; 0.031
One flexible vs two fixed	192	0.057	0.040; 0.076	465	0.025	0.015; 0.035

*Diabetes population with basal-only questions: 192 patients with type 2 diabetes; diabetes population with basal–bolus questions: 265 patients with type 1 diabetes; 200 patients with type 2 diabetes.

95% CI, 95% confidence intervals.

Once-daily injections (both fixed and flexible timing) had significantly higher utility compared with twice-daily, as evaluated by all populations (Table 3). The greatest increase in utility was estimated by the diabetes population for a basal-only regimen, where a once-daily flexible dose had a 0.057 [95% CI: 0.040; 0.076] higher utility compared with two fixed-time injections. The effect of flexible dosing was less pronounced for all groups investigated when the flexible basal insulin dose was part of basal–bolus therapy (where timing of bolus doses was fixed) (Table 3).

Frequency of SMBG testing decreased health utility, with the annual number of tests significantly correlated with reduced utility in basal–bolus regimens (Table 4, Figure 2). One additional SMBG test per annum had a negative impact of 0.000,022,1 in annual utility [95% CI: 0.000,011; 0.000,033,4].

Figure 2. Utility and annual SMBG tests.

Table 4. Regression of utility on number of self-monitored blood glucose tests.

	Estimate	95% CI	SE
Intercept (basal-only)	0.84
Basal–bolus regimen	−0.051,131	−0.069,824; −0.032,554	0.009,363,1
Number of annual SMBG-tests	−0.000,022,1	−0.000,033,4; −0.000,011	0.000,005,8

SMBG, self-monitored blood glucose.

Discussion

This is the first study to use the TTO survey, an established health-economic methodology, to show that a large cohort of people considered that the ability to dose basal insulin flexibly had an influence on health utility in diabetes. This study supports the notion that flexible dosing of basal insulin and dosing once rather than twice daily would increase health utility in people with diabetes, particularly in relation to basal-only therapy. Therefore, the ability to dose basal insulin flexibly (by changing the time of administration from day-to-day according to patient requirements) could positively impact on HRQoL, thereby enabling patients to more easily incorporate therapy into their existing lifestyles, and potentially contributing to greater therapy adherence and long-term outcome benefits.

The results from the basal-only questionnaire were consistent for the general population and the population with diabetes. However, the population with diabetes did not value flexibility to the same degree as the general population when part of a basal–bolus regimen. When the basal injections are part of a regimen with meal-time bolus injections, the flexibility does not mean as much as when the basal injection is the only injection per day. The difference in this result may reflect the better understanding of the treatments by the respondents in the population with diabetes who have experience of fixed-time dosing regimens. Another study has shown that increasing flexibility of bolus insulin doses with insulin lispro vs human insulin can also improve QoL30. Therefore, it should be recognized that, although the benefits of flexible dosing would be greater for patients using basal-only therapy, increased flexibility would be advantageous also to patients receiving basal–bolus therapy and could potentially make therapy adherence easier. Furthermore, the estimated HRQoL increment associated with flexible basal insulin dosing observed in this study may under-estimate the effects on health utility in specific sub-groups where such an attribute may be particularly important, e.g., people with varying shift working patterns. A large-scale survey investigating the issue of psychological insulin resistance in people with type 2 diabetes found that 45.2% cited ‘restrictiveness’ as a reason to avoid insulin therapy4. Adherence to therapy is vital for maintaining good glycemic control8,9, with simulations using patient data suggesting that missing an average of 2.1 basal insulin injections per week would lead to an increase in glycosylated hemoglobin (HbA_1c) of 0.3–0.4%31.

Boye et al.16 investigated the impact of flexible insulin dosing on health utility in 151 patients with type 2 diabetes using standard gamble interviews. The average added utility associated with flexible vs inflexible dosing was 0.006, which, whilst significant, was lower than the utility values observed in the current study16.

This study showed that an increased frequency of blood glucose monitoring decreased health utility. The SMBG-associated disutility appears to be quite low (relative to that associated with injection frequency), but many tests are taken each year, so the increment has to be small to make sense when they are aggregated. The Digem study showed an annual utility of −0.029 (for a scheme with SMBG tests approaching five per week) for patients with type 2 diabetes. If constant utility per SMBG test is assumed, the utility for an annual SMBG test can be calculated by dividing the −0.029 with the number of tests per year (5*365.25/7). This amounts to a small decrease in utility of −0.000,111 per test32, being greater than that found in the present study (equivalent to −0.0011 for one additional test per week). Additionally, more frequent SMBG testing is associated with higher healthcare costs and reduced patient quality-of-life33, without significantly contributing to improved HbA_1c29.

Whilst several studies have used cost-utility models to examine the value of treatments for diabetes34–36, there is a lack of data specifically addressing the effects of insulin-related attributes, such as injection frequency, on HRQoL, which become increasingly important during the course of therapy intensification in people with type 2 diabetes. This study has an advantage in that it looked specifically at the characteristics of insulin therapy, namely injection frequency and timing, but also the frequency of SMBG testing. Furthermore, this study used a large cohort of respondents from the general population and with type 1 and type 2 diabetes. The similar response from the general population and the population with diabetes increases the face validity of the results. Despite the small increase in utility estimated in this study, dose flexibility and reducing the number of SMBG test strips are important factors to consider in cost-utility models, due to the large number of patients they affect and the frequency at which they are utilized13.

TTO is a preference-based instrument that is relatively easy to use and understand, and was, therefore, considered to be more appropriate for this study than the widely used standard gamble instrument, which can be difficult for the respondents to understand26. The use of an age-dependent, life-expectancy adaptation of the TTO questionnaire may have increased the relevance of the questions in this study, compared with the fixed 10- or 30-year trade-offs used in other studies. The TTO approach that specifically estimates the influence of basal-insulin dose flexibility on an individual may be more sensitive and more applicable than generic measures of QoL, such as the Short Form (36) Health Survey (SF-36)® or EQ-5D™ that lack a treatment- and disease-specific dimension. This study has the limitation that it measures hypothetical health states rather than the HRQoL of people with these health states; however, the preferences of the general population are often recommended as the most relevant17,27,28. The repetitive nature of TTO questionnaires may give rise to respondent fatigue. To avoid this, a limited number of health states were evaluated by each respondent, and for each health state only four iterative questions were given plus two follow-up questions to clarify extreme answers. It could be suggested that the web-based nature of this survey is biased towards literate respondents with internet access and IT competence. However, the literacy rates and proportion of internet users is very high in the three countries investigated here. Due to the web-based nature of the survey no help was available to respondents when answering the questions, which could lead to inconsistencies and misunderstandings in the responses. The drop-out rate of >10% is quite high; however, this is typical of many of the web-based surveys that are commonly used in health-economic assessments37–39.

Conclusion

This large TTO-based survey of a general population and a population with diabetes showed that utility values derived for flexible dosing of basal insulin were higher than for once- and twice-daily fixed-time basal insulin injections. This effect was most pronounced in the context of basal-only therapy, which is used by a majority of patients with type 2 diabetes. Respondents to this survey also perceived SMBG testing to be a burden, with an increase in the frequency of testing associated with reduced health utility.

Transparency

Declaration of funding

This study was sponsored by Novo Nordisk.

Declaration of financial/other relationships

ME has been an advisory panel member and received speaker honoraria for Novo Nordisk, Sanofi, Bristol Myers Squibb, Novartis, and MSD. KK has received an NHS R&D study grant and served on advisory boards for Novo Nordisk, Eli Lilly, Merck, Sharp & Dohme, Bristol-Myers Squibb, Roche, Novartis, Boehringer Ingelheim-Lilly, Astra Zeneca, Takeda, Janssen, and Sanofi. MB and HHJ have received consultancy fees from Novo Nordisk. BC and JG are employees at Novo Nordisk.

The sponsor was involved in the study design, data collection, data review, and data analysis. All authors had full access to the data and were involved in data interpretation and manuscript writing. As stated in the ethics applications and approvals, the raw data are stored at Incentive, with access restricted to HHJ and MB, who take responsibility for the integrity of the data and the accuracy of the data analysis. ME had the final decision to submit the manuscript for publication and takes responsibility for its content.

Acknowledgments

The authors take full responsibility for the content of this manuscript. Adele Norman and Mark Nelson at Watermeadow Medical (supported by Novo Nordisk) provided medical writing and editing support.