A discrete choice experiment to determine patient preferences for injection devices in multiple sclerosis

Abstract

Objective:

Relapsing-remitting multiple sclerosis (MS) is usually managed with disease modifying drugs (DMDs), most commonly administered via self-injection. The aim of this study was to estimate the influence that different treatment-related attributes have for MS patients on their choice of MS DMD device. By establishing the relative importance of these characteristics for patients it should be possible to better understand the acceptability of a given device and to optimize the development of future devices.

Methods:

A discrete choice experiment (DCE) survey was developed on the basis of a review of published literature. Attributes identified for inclusion in the survey were: ease of use; comfort of use; presence of additional functions, needle visibility; practicality and efficacy. Choice sets were presented as pairs of hypothetical treatments based upon a fractional factorial design. One-hundred device-using MS patients completed the survey online. Analysis was conducted using a mixed-logit approach.

Results:

Analysis of the DCE data revealed that all attributes significantly predicted treatment choice. Efficacy exhibited the largest effect on treatment selection and this provided context for understanding the magnitude of impact for the other attributes. Reducing the discomfort associated with device use and eliminating the necessity for assembly or drug reconstitution were highly valued. The addition of reminder and time-stamping functions, improved needlestick injury prevention, and reduction in device size were secondary concerns but still deemed desirable.

Conclusion:

Efficacy is of primary importance to MS patients, but characteristics of drug delivery devices can play an important role in treatment decision-making. Not all device characteristics could be included, and results are based upon 100 participants only. Findings suggest there is significant potential value in developing self-injection devices that are not only efficacious but also convenient and comfortable to use. Reducing barriers to adherence could potentially translate into improved treatment outcomes for patients with MS.

Keywords::

• Multiple sclerosis
• Patient preference
• Discrete choice experiment
• Health-related quality-of-life

Introduction

Multiple Sclerosis (MS) is a chronic, inflammatory, degenerative neurological condition of the central nervous system, which affects ∼2,500,000 people worldwide1. A majority of patients (70–80%) present with a relapsing-remitting form of MS (RRMS)2. Currently, there is no cure for the disease, but RRMS is usually treated with disease modifying drugs (DMDs). These drugs are designed to affect the long-term progression of MS by reducing the number and severity of relapses2. Where oral administration is not available or appropriate these drugs can be administered via regular self-injection on a long-term basis3. Research has shown that self-injection of such treatments can be stressful and physically demanding for MS patients, especially where symptoms of the disease itself can make the task more difficult4. The ability to self-inject has been shown to be a key factor in influencing patient treatment adherence, especially in terms of convenience and discomfort5. Improving the tolerability of these medications or improving the mode of their administration may help reduce patient burden and potentially improve adherence5.

Various self-injection devices exist, providing patients with options regarding their treatment administration. Devices can vary in a number of ways such as in terms of providing differing levels of comfort, practicality and needle visibility. Existing studies have assessed patient acceptability and adherence with such injection devices2,4,6. They suggest that convenience and ease of use of electronic autoinjection devices are highly valued by patients and improving these attributes may promote increased patient adherence2. New technological improvements to injection devices (such as dose monitoring) may also help patients overcome the barriers to self-injection, improve adherence, and even improve the communication between patient and healthcare professional6.

Where there is more than one medically reasonable treatment option, the active participation of patients in decision-making is becoming more important; especially in terms of patient preference and values7. The differing aspects of injection devices may be important to patients, and they must also be of interest to decision-makers who need to place a value on all new available devices and treatments. This is often achieved by capturing data on patients’ health-related quality-of-life (HRQoL), via instruments such as the EQ-5D8, for use in quality-adjusted life year (QALY) models. However, it is not always clear that the benefits conferred by optimizing the design of such devices is adequately captured using such an approach. A mechanism that is potentially more useful for understanding the value of different treatment characteristics (or attributes) is through the use of stated preference methods such as Discrete Choice Experiments (DCEs)9.

The DCE is based on the premise that all ‘goods’ can be described by their attributes and that, secondly, the extent to which an individual values a good or service can be described in terms of the levels of these characteristics9. From these, a value can be derived10. When used in the context of healthcare, the DCE methodology is a powerful means of capturing data concerning the strength of preference for attributes of a given intervention, through the possibility of multiple trade-offs11. Furthermore, one of the major benefits of this methodology is that it enables us to understand the relative importance of attributes with respect to the degree to which patients are willing to trade between them (estimated using marginal rates of substitution). The importance of each attribute can be expressed as an Odds Ratio (OR), or if cost is included as an attribute, using willingness-to-pay (WTP) estimates. To elicit such values, the varying level of each attribute of the device or treatment can be bundled together to form a profile depicting either actual or hypothetical interventions, and these can be compared against other such profiles.

The present study was designed to estimate the influence that a number of different treatment-related attributes have for MS patients on their choice of MS DMD injection device. By establishing the relative importance of these characteristics for patients it should be possible to better understand the acceptability of a given device and to optimize the development of future devices.

Methods

Study design

This DCE study involved the development of hypothetical scenarios associated with injection devices for MS treatments. These scenarios consisted of a series of choices between two hypothetical treatments and were presented via the web. Participants were asked to make a decision as to which of the two treatments they preferred (described in terms of attributes and their associated levels). The results provide insight as to whether an attribute was a significant predictor of treatment choice (and therefore instrumental in treatment decision making), to what extent particular treatment characteristics were important, and the rate at which individuals were willing to trade between characteristics12.

Attribute identification

The attributes describing DMD injection devices were identified through examination of existing devices and a review of published literature. Potential attributes identified included device ease of use, reliability, associated pain levels, safety, ease of device preparation, and confidence in using the device3,13,14. The final attributes selected for inclusion in the scenarios reflected a number of likely concerns for users which covered the spectrum of commonly available devices. These were: ease of use; comfort of use; needle visibility; the presence of a reminder function; and practicality. An attribute which captured the efficacy of the treatment was also selected for inclusion, disability (expressed as chance of disability progression in the next 12 months), to help establish the relative value of the other attributes. The associated levels for these attributes were also identified through examination of published literature and clinical input. Efficacy was expressed in terms of the prevention of increasing disability level (as reported through the use of a MS disability scale ranging from 1 ‘I have no problems and no limitations’ to 11 ‘I am bedridden and unable to sit in a wheelchair for more than 1 h’). This scale was a modified version of the Patient Determined Disease Steps (PDDS), which is a functional self-assessment scale for use in MS patients15. The PDDS is based on the earlier measure, the Disease Steps Scale16, which correlates strongly with the Expanded Disability Status Scale (EDSS)17, which is a physician rated scale16.

Final DCE development

After identifying the attributes for inclusion, a draft version of the DCE was developed. Semi-structured cognitive debriefing interviews were conducted over the phone with five MS patients with the aim to assess the relevance of the suggested attributes and to ensure that the draft survey was relevant and comprehensible.

Two MS specialist clinicians were also interviewed over the phone and were asked to review the complete draft survey to ensure the descriptions were clinically realistic and that the survey appeared suitable for patient completion. Clinicians were prompted to give their thoughts on the attributes selected and how they believe they related to patients.

The final attributes and their associated levels were confirmed following the interviews and are outlined below:

Table

Ease of use:	• Device requires assembly. Medication requires reconstitution.
	• Device requires assembly. Medication is pre-mixed.
	• Device needs no assembly. Medication is pre-mixed.
Comfort of use:	• I cannot adjust any aspects of the injection. The device uses a large needle which is uncomfortable.
	• I cannot control the speed of injection and the device injects the treatment quite rapidly. The device uses a very small needle and I can partially control the depth of injection.
	• The injection is more comfortable because the needle and injection speed are adjustable. The device uses a very small needle and I can easily control the depth of injection.
Reminders:	• No reminders.
	• Provides reminders.
	• Records time and provides reminders.
Needle visibility:	• Visible the entire time.
	• Visible but protected by guard.
	• Cannot be seen.
Practicality:	• Large device. Requires sharps bin and medication needs refrigeration.
	• Quite large device. No sharps bin necessary and can be stored at room temperature.
	• Compact device. No sharps bin necessary and can be stored at room temperature.
Disability:	• Your condition progresses two levels on the MS Disability Scale.
	• Your condition progresses one level on the MS Disability Scale.
	• Your condition remains stable on the MS Disability Scale.

The final attributes and levels were combined into paired choice sets using a published orthogonal array which had been folded over (www.research.att.com/~njas/oadir/). The final survey included 19 pairs of choice sets. The electronic version of the survey was developed in a format suitable for administration via the web.

Sample

One hundred participants residing in the UK were recruited via a specialist recruitment agency. All participants were diagnosed with MS, were ≥18 years of age, and currently self-injecting one of the following DMDs (which may be delivered via a number of different devices):

• Interferon beta 1a (Avonex, Biogen Idec, Research Triangle Park, NC);

• Interferon beta 1b (Betaferon, Bayer Pharma AG, D-13342 Berlin, Germany);

• Glatiramer acetate (Copaxone, Teva Pharmaceuticals Ltd, London, UK); and

• Interferon beta-1a (Rebif, EMD Serono, Inc. Rockland, MA).

Data collection and analysis

Prior to data collection taking place, ethical approval was sought from an Investigational Review Board. Eligible participants were sent the study-specific URL by a specialist patient recruitment agency and asked to provide informed consent before participating in the study. Participants completed a basic socio-demographic and clinical MS history form (comprising of nine items on ethnicity, employment status, time since diagnosis, etc.) followed by the EQ-5D. Information regarding the participants’ current injection device was also collected. Finally, all participants were asked to read the DCE instructions before being presented with the 19 DCE choice sets. Participants were asked to select whether they would choose scenario A or B, and then whether they would choose their preferred choice over their current injection device. Data from the DCE was analysed using a mixed logit model. This approach evaluates choice responses based on attributes of the treatment alternatives and also takes into account repeated choices by the same individual. The regression coefficients obtained from the mixed logit model provided an estimate of the (log) odds ratio of preference for a unit increase in the treatment attribute (outlined in the Results section). In other words, the mean coefficient represent the relative utility (or value) of each attribute to the respondent conditional on other attributes.

Results

One hundred participants completed the online survey. Data from 99 of these participants were analysed, due to one participant not meeting the clinical criteria.

Table 1 shows the sample demographics and their MS clinical history. The study sample included a higher proportion of female participants (70.7% female vs 29.3% male) which reflected higher prevalence of MS among females. A little over a third of the sample had completed degree level education. Over half the participants had been diagnosed with MS more than 5 years ago (56.5%), with a majority having received DMDs for 3 years or more (64.7%). The mean health-related quality-of-life of the sample (expressed a utility value ranging from 0 for dead to 1 for full health) was 0.58 (SD = 0.29), with a mean visual analogue score of 63.72. The self-reported level of disability amongst the respondents varied, with the most common level of disability (25.3%) being reported as being able to walk 100–200 m. Only a small number of respondents (2%) reported the most severe level of disability, i.e., wheelchair bound.

Table 1. Socio-demographic and clinical information.

Demographic and clinical information	n = 99
Gender
Male	29 (29.3%)
Age
Mean (SD)	41.90 (7.98)
Ethnicity
White	96 (97%)
Non-white	3 (3%)
Highest level of education completed
No formal qualifications	6 (6.1%)
Completed university	36 (36.4%)
Employment status
Employed, full or part time	44 (44.4%)
Employed, currently on long-term sick-leave	5 (5.1%)
Disabled	25 (25.3%)
How long have you been diagnosed with MS?
<10 years	75 (75.8%)
>10 years	24 (24.2%)
How long have you been injecting DMD for?
<5 years	60 (60.7%)
5 years or more	39 (39.3%)
Current medications
Avonex	17 (17.2%)
Betaferon	12 (12.1%)
Copaxone	39 (39.4%)
Rebif	28 (28.3%)
Self-reported quality-of-life
Mean EQ-5D score (SD)	0.58 (0.29)
Mean VAS sore (SD)	63.72 (19.76)
Self-reported level of disability (MS disability scale)
No problems or limitations	8 (8.1%)
Mild symptoms	11 (11.1%)
Noticeable symptoms	15 (15.2%)
Symptoms plus limit on activities	7 (7.1%)
MS interferes with activities	15 (15.2%)
Can walk 100–200 m	25 (25.3%)
Requires stick/crutch	10 (10.1%)
Requires 2 canes/crutches	6 (6.1%)
Wheelchair main form of mobility	2 (2%)
Restricted to bed or chair	0 (0%)
Bedridden	0 (0%)

Data was collected on participants’ current injection devices. A third of participants reported that their current devices were easy to use, 34% reported that their current device had a good level of comfort, and 47% reported that the needle couldn’t be seen in their current device. Only 14% of participants reported that their current device provided reminders, and only 14% reported that their device was compact and practical.

Table 2 shows the results of the mixed logit model of patient preferences of injection devices for MS. The OR expresses the likelihood of choosing one attribute/level over another. As expected, treatment efficacy, evaluated in terms of ability to reduce disability progression, was the biggest concern to participants, who expressed reduction in odds of treatment preference by a factor of 0.117 for a device that was likely to result in disease progression by one level on the MS Scale (p ≤ 0.001, OR = 0.117). As anticipated, a single level change of this attribute had the most pronounced effect on choice of treatments. Patients placed the greatest preference on this particular attribute compared to the others included in this survey. Participants were significantly less likely to choose a device that is hard to use compared to one that is easy (p ≤ 0.001), or a device that was not comfortable to use (‘moderate’ OR = 0.689, p = 0.001, or ‘poor’ OR = 0.251, p < 0.001) when compared to one that offered a good degree of comfort. The OR demonstrate that comfort was found to be a highly important attribute in the decision-making process. The very low OR (0.251) and relatively tight confidence intervals (0.171–0.367) show that a device with a poor level of comfort is less preferable to patients than a device with moderate levels of comfort, which in turn is less preferable than a device with a good level of comfort. A reminder function was also preferred by participants (OR = 1.313, p = 0.04), as were reminders plus time-stamping (OR = 1.499, p = 0.001) over devices that offered no additional functionality. Participants preferred a device with a completely hidden needle compared to a needle that was ‘always exposed’ (OR = 0.691, p = 0.004). Finally, the respondents preferred to avoid a large device that required both a bin and a refrigerator when compared to a device that was compact and required no bin or refrigeration (OR = 0.632, p < 0.001). It is difficult to provide an order of which attribute is most or least important from those included in this DCE. This is because the attributes are different concepts and so difficult to compare. However, based purely upon the odds ratios in Table 2, the attributes of disability, device comfort, and ease of use were all deemed as highly important by patients when selecting an injection device. Although still influential in the decision-making process, needle visibility and the inclusion of reminders were not as important.

Table 2. Results of mixed logit model of patient preference for injection devices for MS.

Variables	Odds ratios	CI	Coefficients, Mean (SD)	Significance (p)
Ease of use (reference: Easy: no assembly or reconstitution)
Moderate: Assembly required, injection pre-mixed	0.823	0.660–1.026	−0.194 (0.022)	0.084 (0.916)
Hard: Assembly and reconstitution required	0.484	0.363–0.645	−0.726 (0.544)	0.000 (0.002)
Comfort of use (reference: Good, injection more comfortable, speed adjustable, depth easily controllable, very small needle)
Moderate: Speed not controllable, depth partially controllable, very small needle	0.689	0.551–0.860	−0.373 (0.034)	0.001 (0.905)
Poor: No aspect adjustable, large and uncomfortable needle	0.251	0.171–0.367	−1.383 (1.142)	0.000 (0.000)
Reminders (reference: No reminders)
Reminder only	1.313	1.012–1.704	0.272 (0.341)	0.040 (0.084)
Records time and provide reminders	1.499	1.171–1.919	0.405 (0.017)	0.001 (0.929)
Needle visibility (reference: Cannot be seen)
Visible but protected	0.993	0.795–1.240	−0.007 (0.041)	0.950 (0.784)
Always visible	0.691	0.537–0.889	−0.370 (0.117)	0.004 (0.550)
Practicality (reference: Compact device, no bin or refrigeration)
Large device, bin and refrigeration required	0.632	0.492–0.813	−0.458 (0.062)	0.000 (0.766)
Quite large device, no bin or refrigeration required	0.810	0.648–1.012	−0.211 (0.034)	0.064 (0.871)
Disability (reference: stable condition)
Progresses by one level on MS scale	0.117	0.067–0.204	−2.145 (2.148)	0.000 (0.000)

Table 3 provides information on the participants’ current injection devices and associated levels. As demonstrated, a small number of participants have devices that are hard to use with a poor level of comfort, with over half reporting moderate levels of comfort in devices that are moderately easy to use. The majority (over 80%) of devices currently used do not provide reminders, and over half the devices have a visible needle. A very high number of participants (79%) reported a low level of practicality, where the device is large, a sharps bin is needed, and medication needs to be kept cool.

Table 3. Participants’ current injection devices and level of disability.

Injection device questions	Description of level	n = 99 (%)
Ease of use
Easy	‘My injection device does not need to be assembled prior to the injection. The medication is pre-mixed in a cartridge. An electronic device inserts the needle’	33 (33.3%)
Moderate	‘My injection device needs to be assembled before I can inject myself. The medication is pre-mixed in a cartridge’	59 (59.6%)
Hard	‘My injection device needs to be assembled before I can inject myself. The medication has to be mixed with water to reconstitute it before it can be injected. The medication is injected manually’	7 (7.1%)
Comfort of use
Good	‘The injection is more comfortable because the needle and injection speed are adjustable. The device uses a very small needle and I can easily control the depth of injection’	34 (34.4%)
Moderate	‘I cannot control the speed of injection and the device injects the treatment quite rapidly. The device uses a very small needle and I can partially control the depth of injection’	53 (53.5%)
Poor	‘I cannot adjust any aspects of the injection. The device uses a large needle which is uncomfortable’	12 (12.1%)
Reminders
Records and reminders	‘My injection device records time of last injection and provides a reminder’	14 (14.1%)
Doesn’t record but gives reminders	‘My injection device doesn’t record time of last injection but does provide a reminder of when the next injection is due’	4 (4%)
No reminders and doesn’t record	‘My injection device doesn’t provide any reminders’	81 (81.8%)
Needle visibility
Can't be seen	‘Needle cannot be seen’	46 (46.5%)
Visible but protected	‘Needle is visible but protected by a guard’	31 (31.3%)
Visible	‘Needle is visible the whole time’	22 (22.2%)
Practicality
Device compact	‘Device is compact, no sharps bin and can be stored out of the fridge’	14 (14.1%)
Device large but no sharps bin needed	‘Device is quite large, but does not require a sharps bin and can be stored out of the fridge’	7 (7.1%)
Device large and need sharps bin	‘Device is large. You also need a sharps bin and the medication needs to be kept cool’	78 (78.8%)

Discussion

This study employed DCE methodology to explore patient preferences for different attributes of self-injection devices for MS. Data from 99 patients was analysed to explore the potential influence of a number of characteristics on their choice of treatment (ease of use, comfort of use, presence of a reminder/records function, needle visibility, practicality, and efficacy). The findings suggest that all of these attributes predicted device choice and were, therefore, considered by patients as part of their evaluation process.

The attribute that exerted the greatest degree of influence on patients’ treatment selection was efficacy (as expressed in terms of ‘disability’). This is not a surprising finding, as disease progression in MS can be associated with severe consequences and is an indication of a poorer future prognosis. This finding is not unexpected as the other attributes included in the DCE, which centre on the convenience and practicality issues of drug delivery, would be unlikely to be considered as important by patients. Despite this fact, the analysis revealed that the remaining attributes were all integral to the decision-making process of participants and the individuals were not simply selecting devices that offered the best outcome in terms of disease progression.

MS is a lifelong chronic condition and the prospect of having to self-inject on a regular basis could potentially be burdensome for individuals. The DCE results reveal that participants expressed a preference for devices which offered a combination of increased practicality and convenience as well as reduced discomfort in day-to-day use. Devices which were compact, ready to use, did not require refrigeration and that could easily be disposed of were strongly favoured. This finding is consistent with the fact that the majority of people are diagnosed with MS at a young age (between 20–30) and so therefore the ability to continue normally with daily activities (including work, socializing, and caring for the family, etc.) is typically of great importance. The stigma associated with MS can also be a potential cause of embarrassment and may necessitate modification of social behaviour. This can be especially true when the illness is presented in public18. Devices that are large, require preparation before use, or facilities such as refrigeration, can cause unnecessary burden and attention for self-injectors. A smaller, practical, and more convenient device may permit individuals with MS to lead lives less burdened with the need for behaviour modification to manage their disease, both in terms of disruption to social and daily activities. This development may in turn promote increased adherence to therapeutic regimens and, therefore, potentially improved treatment outcomes. It may also be worth noting the types of device that are currently in use by respondents. If a participant has a device which does not currently provide reminders, does that make them more or less likely to place value on this attribute?, i.e. does it change your valuation for that feature compared to people who have experienced that attribute in their current device? The answer is probably yes, although this is difficult to analyse and interpret.

Adherence to medication is important for a variety of conditions, but for diseases such as MS, non-compliance with treatment can have devastating repercussions for individuals. Research suggests that reasons and predictors for non-adherence in MS patients who self-inject include low levels of self-efficacy and self-injection competence19. Where failure to adhere to the treatment regimen is of genuine oversight on the part of patient, rather than a deliberate effort, attributes such as reminders on devices may prove to be important. Research has shown that the main reason for MS patients to be non-adherent is due to forgetting to administer their injection14. This could explain why the reminders and time-stamping functionality was significantly valued by respondents in the results of the current study, and may be indicative of their desire to ensure proper treatment adherence. Having access to a device which provides reminders may directly translate to improved adherence and potentially improved drug efficacy. A device with a good level of comfort was also highly valued by participants in this study. This attribute has also been shown to impact on adherence levels, with injection site pain being cited as a common reason for non-adherence in MS patients14.

There are a number of limitations to the current study that should be considered when assessing the findings. As with all DCE surveys, not all aspects of treatment which could potentially affect or influence preference were included. The attributes that have been selected for inclusion in this survey were developed from previous literature and were sufficient to adequately characterize currently available treatments. It may be that there were additional attributes that individuals were considering when making decisions that were inferred from the participant’s perspective. Although a cognitive debriefing process was undertaken with the selected attributes, and that these were all found to significantly predict treatment choice, we cannot be certain that participants were not considering other issues when making treatment choices.

It is difficult to establish to what extent the sample population is illustrative of ‘typical’ MS patients, although the information collected does appear to suggest that they were broadly representative. The use of an online web-based survey and recruitment from a panel of multiple sclerosis patients may have led to some bias in the study population. Only patients who were sufficiently well enough to participate and who have internet access were able to respond. It would be interesting to capture preferences of a more severely affected population (as measured by the MS Disability Scale). A device that is easier and more convenient to use may potentially have more benefit in this population, by prolonging their ability to self-inject DMDs and, therefore, their independence. This does not diminish the value of the study findings, but it would be prudent to consider them within the context that the data was collected.

Conclusion

The information gained from this study is relevant both for the future development of injection devices and for clinicians responsible for managing MS. Understanding which aspects of devices are important to patients could significantly reduce patient burden and translate into an improvement in treatment adherence. This in turn may potentially lead to increased treatment efficacy. The findings indicate that all attributes included in the DCE survey exerted influence on treatment selection and, therefore, represent issues that should be considered when prescribing the use of DMD injection devices.

Transparency

Declaration of funding

This study was funded by Merck Serono.

Declaration of financial/other relationships

Sarah Shingler, Paul Swinburn, Shehzad Ali, and Andrew Lloyd have disclosed that they are all employees of Oxford Outcomes, an ICON plc Company that received funding from Merck Serono to conduct this research. Rodolphe Perard has disclosed that he is an employee of Merck Serono. JME Peer Reviewers on this manuscript have no financial or other relationships to disclose.