Perceptions of Usability and Design for Prefilled Insulin Delivery Devices for Patients With Type 2 Diabetes

The study was conducted in two phases, as illustrated in Figure 1. Patients who were insulin-naive and current or previous prefilled insulin pen users were recruited to the study. The HCP sample comprised diabetologists and diabetes nurse educators (DNEs).

For both phases, the inclusion criteria for patients were as follows:

• Insulin-naive patients: < 80 years of age, at least 2 years' duration of type 2 diabetes, only receiving oral antihyperglycemic agents, not using injectables for any other condition, comfortable and open to discussing insulin injection at interviews, able to complete simulated injections without assistance from others.

• Prefilled pen users: < 80 years of age, at least 2 years' duration of type 2 diabetes, currently injecting prandial insulin using a prefilled pen delivery system with or without oral antihyperglycemic agents or a basal (long-acting) insulin, not using injectables for any other condition, comfortable and open to discussing insulin injection at interviews, able to complete simulated injections without assistance from others.

To ensure that HCPs could speak about patients' perceptions with confidence, the inclusion criteria for HCPs were as follows:

• Diabetologists: Qualified for more than 4 years, consultants (specialists) only, workload of > 100 patients per month with at least 60 patients per month consulting on type 2 diabetes, personally makes the decision to initiate insulin therapy and chooses the insulin brand to prescribe, currently prescribing prandial insulin delivered through at least two of the three pen brands of interest (FlexPen, KwikPen, and SoloSTAR).

• DNE: Qualified for more than 4 years, workload of > 20 diabetes patients per month, personally trains patients on insulin injections or devices, currently prescribing prandial insulin delivered through at least two of the three pen brands of interest.

The qualitative phase of the study aimed to identify concepts related to insulin pens that were important to measure and therefore to optimize the questionnaire design for the quantitative phase. Patients were recruited in the United Kingdom via fieldwork agencies that used HCPs and patient associations to identify potential patients from their databases. With their permission, patients were screened for eligibility.

The study was completed in accordance with current British Healthcare Business Intelligence Association guidelines.³⁷  Semi-structured interviews were conducted with a small cohort of patients with type 2 diabetes (n = 4 insulin-naive, n = 4 pen users), DNEs (n = 5), and specialist physicians (n = 5). Patients and HCPs gave written informed consent. After open-ended concept elicitation to understand important features of insulin pens for prandial insulin delivery, an existing questionnaire containing 18 items on the features of insulin pens³⁸  was provided to participants, who were asked about its comprehensiveness and ease of understanding. The interviews were conducted as 30-minute telephone calls with HCPs and 60-minute face-to-face meetings with patients.

The interviews were digitally recorded and analyzed thematically for missing concepts or features, lack of clarity, and redundancy in the questionnaire.³⁹  The questionnaire³⁸  was then adapted according to the qualitative results to provide a comprehensive survey of the key features of prefilled insulin pens. Two versions of this questionnaire were created: one to evaluate the importance of the features and another to evaluate the perception of the insulin pens on these features.

Using the same inclusion criteria, additional patients were recruited through fieldwork agencies to participate in a simulation study. These patients had not participated in the qualitative phase. The simulation study aimed to quantify the importance of different insulin pen features to patients, quantify patient perceptions of these pens, and quantify prescribers' perceptions of what patients feel is important in insulin pens.

At the start of the study, patients who provided written informed consent were asked to complete a sociodemographic form and a questionnaire to rate the importance of pen features identified in phase 1. Patients were then asked to simulate the preparation and delivery of insulin using SoloSTAR, KwikPen, and FlexPen disposable pens. For each device in turn, patients were asked to prime the pen (i.e., remove air bubbles that may be in the pen and dial the dose) and then deliver 20 units of insulin into an injection pad. The order of device use was rotated through a simultaneous algorithm to minimize order bias. After each simulated injection, patients were asked to complete a questionnaire to rate the features of the pen they had just used.

HCPs were recruited through fieldwork agencies and, after provision of consent, completed an Internet-based sociodemographic form and a questionnaire to rate their perception of the importance of features of insulin pens to their patients. No simulation or pen rating was performed with HCPs. An overview of the process followed for the quantitative phase is provided in Figure 1.

Because this was an exploratory study, no power calculations were conducted. Sample sizes were chosen to provide meaningful data from a variety of patient and HCP subgroups (i.e., insulin-naive patients vs. current insulin users and users of the three different pen types) and therefore to allow subanalyses to be conducted. A target sample of 100 patients, 75 diabetologists, and 75 DNEs was determined; the final sample size of the study was 111 patients, 75 diabetologists, and 76 DNEs. The sample chosen was considered to be sufficient for the exploration of the data.

All analysis variables were assessed descriptively; discrete data were summarized presenting counts and percentages, whereas for continuous data counts, means with standard deviations (SDs), medians, quartiles, and ranges were calculated. To aid interpretation of the descriptive analysis, tests for statistical significance (t tests) were performed at a 5% significance level to assess differences in perceptions of pen design and usability. Summary statistics were performed using SPSS software (version 14, IBM, Armonk, N.Y.).

Eight patients (n = 4 insulin-naive, n = 4 pen users) and 10 HCPs (n = 5 specialists, n = 5 DNEs) participated in the qualitative questionnaire development phase.

After qualitative thematic analysis³⁹  according to a predefined coding frame, several changes were made to the original 18-item questionnaire.³⁸  Three items were added to the questionnaire after their importance as attributes was highlighted: “how well you can grip it,” “hear the clicks as you dial the dose,” and “tell how much is left in the cartridge.” Other items were expanded to provide further detail (e.g., “its size” was expanded into three items “its length,” “its width,” and “its diameter”; and “changing the pen needle” was changed to “putting on the pen needle,” “taking off the needle cap,” and “taking off the needle [after injection]”). Some items were deleted because of perceived redundancy or irrelevance to the quantitative study. Further changes were made to the wording of items to increase clarity and avoid misinterpretation of statements.

Finally, statements within the questionnaires were re-ordered to match the process followed when patients use the prefilled insulin pens. The final questionnaire was 26 items long and included 14 items on design and 12 items on usability. Two versions of this questionnaire were created:

1. The “importance questionnaire,” through which respondents rated the importance of design and usability features of prefilled insulin pens for the delivery of prandial insulin on a scale of 1 to 10 (where 1 was “not at all important” and 10 was “extremely important”). This questionnaire had different introductory text for insulin-naive patients, current pen users, and HCPs, although the items were the same. The questionnaire was designed so that patients could rate the importance of the features to them personally and HCPs could answer with reference to what they believed is important to their patients (rather than to themselves).

2. The “rating questionnaire,” through which respondents evaluated each pen on the same features on a scale of 1 to 10 (where 1 was “extremely dislike” [design questions] or “extremely difficult” [usability questions] and 10 was “extremely like” [design questions] or “extremely easy” [usability questions]). Because they used the same items, the rating questionnaire could be mapped directly to the importance questionnaire. In addition, patients were asked to rank the three pens overall to provide their preferred choice of pens.

A total of 111 patients (33 insulin-naive and 78 current prefilled pen users) with a mean age of 60 years completed the quantitative phase. Overall, 54% were male, 97% were Caucasian, and close to 50% were retired. Thirty-six percent of patients had visual impairment. Sixty-eight percent used concomitant oral antihyperglycemic agents. Approximately 50% had hypercholesterolemia or hypertension (Table 1). There were only two significant differences between the patient groups, related to the duration of diabetes (significantly more patients with a duration of diabetes > 12 years in the group of current pen users) and the number of patients aged 60–80 years (higher in the group of current pen users).

Among pen users (n = 78, 70.2%), the mean number of injections per day was 3.6 (SD 1.48, range 1–11). The average number of units per injection varied depending on type of insulin administered, but overall for the first injection (n = 78) was 24.4 (SD 17.87, range 3–80) and for the second injection (n = 76) was 34.2 (SD 38.78).

A total of 151 HCPs completed the quantitative phase, 50% of whom were diabetologists (n = 75) and 50% were DNEs (n = 76) (Table 2).

Among patients, the importance of usability features was largely similar between insulin-naive patients and experienced pen users, with the two most important usability features for both groups being knowledge that the entire dose was delivered (mean score 9.3, SD 1.29, out of 10, where 10 was “extremely important”) and ability to read the dose (mean score 9.1, SD 1.26). All other usability features (in terms of ease of use) were also considered important to patients; the items with the lowest score in both groups (priming the pen and replacing the pen cap) still had scores on average closer to the “extremely important” anchor (score of 10) than to “not at all important” (score of 1) (mean scores 7.6, SD 2.3, and 7.2, SD 2.37, respectively) (Table 3).

The only statistically significant difference between non-naive and naive patients was in being able to hear the clicks when they dialed the dose; experienced pen users did not rate this feature to be as important as naive patients (7.0 [SD 2.74] vs. 8.8 [SD 1.52], P < 0.05). Although not statistically significant, naive patients ranked most usability features to be more important than experienced pen users, with the exception of putting the needle on the pen (8.6 for both) and applying the force to inject (8.6 [SD 1.81] for pen users vs. 8.2 [SD 2.13] for naive patients) (Table 3).

When examining results by demographic group, importance rankings for “hearing the clicks as you dial” were significantly higher for patients > 60 years of age (n = 65) than for younger patients (n = 46) (8.0 [SD 2.43] vs. 7.0 [SD 2.68], P < 0.05) and for female patients (n = 60) compared to male patients (n = 51) (8.2 [SD 2.02] vs. 7.0 [SD 2.86], P < 0.05). Female patients also ranked the following parameters significantly higher than males: “dialing a dose of insulin” (9.0 [SD 1.34] vs. 8.1 [SD 2.13], P < 0.05), “taking the needle off after you have injected” (8.8 [SD 1.38] vs. 8.2 [SD 2.01], P < 0.05), and “replacing the pen cap” (7.8 [SD 2.11] vs. 6.8 [SD 2.48], P < 0.05). Patients with visual impairment (n = 41) scored “hearing the clicks as you dial the dose” as significantly more important than patients without visual impairment (n = 70) (8.2 [SD 2.27] vs. 7.2 [SD 2.68], P < 0.05). There were no significant differences in the importance scores between patients with different levels of education or work status.

Statistically significant differences (P < 0.05) were noted between DNEs (n = 76) and diabetologists (n = 75) across all usability features, with DNEs consistently rating pen usability features as being of greater importance to patients. However, when the scores for each usability feature were ranked in order of importance, results were similar: as with patients, both DNEs and diabetologists ranked “knowing the entire dose has been administered” and “reading the dose” as being the most important usability features, whereas “priming the pen” and “replacing the pen cap” were seen as the least important (Table 3).

There were few significant differences in the scores for the usability features between patients and HCPs. Of note was that patients scored the importance of putting the needle on the pen significantly higher than HCPs (8.6 [SD 1.54] vs. 8.2 [SD 1.33], respectively, P < 0.05). Taking the needle off after injection was also scored as significantly more important by patients compared to HCPs (8.5 [SD 1.77] vs. 8.3 [SD 1.47], respectively, P < 0.05) (Table 3).

Patients generally scored design features as less important than usability features, with the exception of “how well you can grip while using it” (mean score 8.4 [SD 1.78]), pen robustness or durability (mean score 7.8 [SD 1.97]), and “how it feels in your hand when using it” (mean score 7.8 [SD 2.04]), which were the three most important design features to patients (Table 3). Female patients scored the importance of several features significantly higher than males: “its discreetness in carrying it with you” (7.6 [SD 2.28] vs. 6.1 [SD 2.83], P < 0.05), “its discreetness when using it” (8.1 [SD 1.85] vs. 6.0 [SD 3.04], P < 0.05), and “how it feels in your hand when using it” (8.2 [SD 1.84] vs. 7.2 [SD 2.50], P < 0.05). Patients with visual impairment also scored the importance of three design features significantly higher than those who did not have visual impairment: “how it feels in your hand when using it” (8.2 [SD 2.10] vs. 7.3 [SD 2.31], P < 0.05), “how far the dose button sticks out to dial 40 units” (7.2 [SD 2.81] vs. 6.0 [SD 2.87], P < 0.05), and “the number of turns it takes to dial 40 units” (6.9 [SD 2.78] vs. 5.6 [SD 2.77], P < 0.05). There were no significant differences in scores by age, different levels of education, or work status.

HCPs' perceptions of the importance of different design features to patients were largely similar to patients' own rankings of importance.

After testing the pens in the simulation exercise, patients scored each pen on each of the key features. Findings are detailed in Table 4, along with the ranking compared to the other pens for each usability or design feature.

The FlexPen scored significantly lower than the other two pens on the attribute “knowing if you have injected the entire dose” (paired t test, P < 0.05). No other significant differences among the pens were noted in usability or design. The KwikPen scored highest on more of the usability features compared to the FlexPen and the SoloSTAR (5 vs. 3 and 0, respectively, out of 12). The SoloSTAR scored highest on more of the design features (7 vs. 2 and 2, out of 14).

When asked to rank the pens overall in terms of preference, current pen users were more likely to prefer their own pens to competitor pens (Figure 2).

However, when results are viewed for the whole patient population (n = 111), including both insulin-naive patients and current pen users, the mean ranking (from 1 to 3) of each of the pens was similar (KwikPen, 2.0; FlexPen, 2.1; and SoloSTAR, 1.9).

The development of insulin pens represents an important achievement in the management of diabetes, with 77% of patients reporting that pen devices facilitate adherence with their regimen.⁴⁰  However, delay to progression of insulin therapy and nonadherence to therapy is an ongoing and increasing problem.^16,41  It is therefore important to understand patients' views on all of the components of insulin delivery, including the insulin delivery device, to enable steps to address the issue of nonadherence. This study sought to understand the relevant features of prefilled disposable insulin pens from the perspectives of insulin-naive patients and current pen users, in addition to HCPs' perceptions of what is important to their patients. Attributes tested were based on a design and usability questionnaire adapted, with patient and HCP input, from Haak et al.³⁸ 

The three most important features to patients all relate to the usability of the pen: knowing the entire dose has been injected, reading the dose correctly, and correcting if the dose is over-dialed. It is also note-worthy that these are the top three features that HCPs (both DNEs and physicians) perceived to be most important to patients. This suggests that HCPs understand patient preferences and therefore may be able to adjust their prescribing decisions to meet these preferences.

Few significant differences were noted between HCP specialties. Although there were few discernible differences between what was perceived as important to current pen users and to insulin-naive patients, importance scores for current pen users tended to be lower. Although HCPs perceived design features to be significantly more important to patients than patients themselves indicated, further analysis revealed that, although the importance score attributed to a particular feature was significantly different, the relative importance of that feature was similar. Therefore, it is important that the scores given to the different usability and design features be considered within the context of the ranking with other features.

As a second objective, this study aimed to elicit patients' and HCPs' perceptions of prandial insulin delivery when simulating an injection using three prefilled disposable pens: SoloSTAR, KwikPen, and FlexPen. After the simulation, patients were asked to score each of the three pens on their usability and design features.

Only one significant difference was noted among the pens; furthermore, the absolute difference between the scores for each pen was very similar, as every feature was ranked > 7 on a 1–10 scale. In terms of usability, the KwikPen ranked as the top pen of the three examined, and the highest design scores were given to the SoloSTAR. Overall, usability features appear to be of greater importance than design features (such as width, color, length, design, and weight) to adult patients with type 2 diabetes. This is also recognized by HCPs.

Findings from this study are consistent with and augment findings from previous studies examining patient preferences with insulin pen use.^42–44  Our results highlight the importance patients place on knowing that the entire dose has been injected, reading the dose correctly, and correcting if the dose is over-dialed.

The fact that these three features were the most important to patients would appear to support previous reports that the majority of patients using vial-and-syringe insulin delivery may incorrectly administer insulin because of timing and dosing errors.^42–44  A previous study³⁸  found that patients considered insulin pens to be more socially acceptable than vials and syringes. This study confirms that patients feel aspects of social acceptability are an important design feature of insulin pens (patients' overall mean score for a pen's discreteness in carrying it and using it were both 7.0 on a 1–10 scale). However, patients considered other features, primarily related to usability, to be more important than the discreetness of the insulin pens.

It is important to note that patients also find glucose control, cardiovascular risk, and weight gain to be key features to consider when deciding whether to choose a glucose-lowering therapy.⁴²  This was not evaluated in the current simulation study, but it is likely that an understanding of patients' perceptions and expectations about glucose control and side effects, as well as their needs regarding key pen features, will affect adherence.

When asked for an overall ranking of the pens, more than 50% of patients who were previous pen users ranked their own pen as the preferred choice. There is a potential bias in the scores toward pens already used by individual patients that may result from familiarity with a specific pen (patients in this study were asked to simulate only one injection with each pen) and that may limit the generalizability of findings.

However, these results also highlight the implications of pen preference as a barrier to switching insulin delivery device, if required. Further research will be valuable to understand the requirements for switching insulin delivery devices in patients with type 2 diabetes and any barriers to doing so in relation to patient preferences for pen features (usability and design).

We identified more statistically significant differences (P < 0.05) between DNEs and diabetologists than between insulin-naive patients and pen users. This could be partially the result of different sample sizes (75 DNEs vs. 76 diabetologists compared to 33 insulin-naive patients vs. 78 pen users). HCPs were not asked whether they themselves had diabetes; however, it is unlikely that this would have confounded results given that the questionnaire clearly asked what pen features they deemed to be important for their patients.

Although this study provides useful data regarding patient preferences and the preferences that HCPs perceive their patients to have, there are a number of limitations to consider. The study focused on patients with type 2 diabetes and was limited to the delivery of prandial insulin. Therefore, the results are not generalizable to patients with type 1 diabetes (recent research has highlighted the need to improve and refine the functionality of insulin pen devices for use in children and adolescents⁴⁵ ), to other types of insulin regimens (e.g., basal insulin or premixed insulin), or to reusable pens.

Although the overall patient sample size was large enough to perform many of the comparisons of interest (e.g., experienced pen users compared to insulin-naive patients), the small subsets of patients using the different types of pens (e.g., only 20 SoloSTAR users were included) prevented further analyses to determine differences in design and usability scores by insulin pen brands. Furthermore, the sample was one of convenience, and the generalizability is unknown. The patient characteristics were as expected, although the study purposefully over-sampled people with visual impairment (36%)⁴⁶  to gain an understanding of whether patients with vision problems place differing emphasis on the importance of various pen features than those without visual impairments.

Because the numbers of patients using the different pen brands differed, the higher ratings for patients' “own” pens may have introduced a bias in ratings for those pen brands that had a greater proportion of users in the study. It is important to note that we did not adjust for multiple comparisons because this was meant to be a descriptive study.

The study used newly developed questionnaires (based on an existing questionnaire³⁸ ) that, although content-valid, have not undergone psychometric testing. However, given that the questionnaires were reviewed at an item-by-item level, the results can still be considered scientifically robust. Nevertheless, future work to assess the reliability and validity of the questionnaires used would be valuable.

This study found few differences between insulin-naive patients and experienced pen users. However, this was a cross-sectional study. Further research examining how patients' perceptions change over time as they move from being insulin-naive to becoming experienced insulin users may provide additional insights. Linking such information to adherence and satisfaction may also provide added benefit to HCPs when making prescribing decisions.

Our study asked patients to rate the importance of features and asked HCPs to rate the importance of features from their patients' perspective. Future work could examine what HCPs themselves find important when prescribing and then compare this to patients' perceptions. This study shows that HCPs have a clear understanding of what is important to patients, but it would be useful to examine what they themselves find important and whether their importance criteria differ from those of patients.

Future research into patient-reported outcomes in diabetes focusing on insulin delivery systems should consider the results of this study (in terms of the features that were important to patients and also the questionnaires used) when designing protocols. Future research may also wish to explore the demographic and clinical variables associated with pen preference in more detail, using, for example, multivariate modeling or multi-attribute utility theory analysis to determine preferences. This would provide further evidence regarding the significance of the findings of a similar study and may help to identify groups of patients for whom certain features are more important, thereby enabling physicians to recommend the most appropriate pen for their patients.

Patient preferences elicited from this research have highlighted certain pen usability and design features that are of importance to patients. Their preferences can be used to identify which of the existing insulin pens (and those in development) have these features to aid physicians in ensuring that they choose the most appropriate pens for their patients.

In a recent review of the link between satisfaction and adherence, 20 studies were found to show a positive association between treatment satisfaction and adherence, compliance, or persistence. The most satisfied patients were the most compliant or persistent, and the least satisfied were the least compliant or persistent.⁴⁷  In an observational study on subcutaneous versus oral treatment for iron chelation therapy, greater satisfaction in the oral treatment group was found and “never thinking about stopping treatment” was associated with less treatment burden,⁴⁸  indicating the importance and relevance of delivery device features in patient adherence. Better understanding patient preferences may ultimately lead to greater patient adherence to insulin therapy, an important issue given the clinical and economic consequences of nonadherence. Future research examining predictive factors in patient preferences and the link between preference and adherence is warranted.

This study was sponsored by and designed and conducted in part by Eli Lilly, the manufacturers of the KwikPen. Authors Gregg, Abetz, Hermanns, and Reaney designed and conducted the research. Authors Heron, Gregg, Abetz, and Reaney interpreted results. Authors Heron and Abetz developed the manuscript, and authors Reaney, Hermanns, and Gregg provided critical comments on the manuscript drafts. All authors finalized and approved the manuscript. The authors acknowledge the assistance of Helen Smith and Lauren Lee in developing the study design and manuscript, respectively.