Quantitative data

Tables 65–67 show QoL outcomes at 6, 12 and 24 months, respectively. Table 68 shows DTSQ data at the same time points.

TABLE 65

Quality-of-life secondary outcomes: MD in change from baseline at 6 months

TABLE 66

Quality-of-life secondary outcomes: MD in change from baseline at 12 months

TABLE 67

Quality-of-life secondary outcomes: MD in change from baseline at 24 months

TABLE 68

Quality-of-life secondary outcomes: DTSQ change from baseline at 6 and 24 months, DTSQc raw scores at 12 months

Improvement was seen across most psychosocial outcomes and time points for both treatment groups. There were no statistically significant differences at 6 months between the pump and MDI cohorts on any psychosocial measure. Participants in the pump group had better improvement in treatment satisfaction at all time points using DTSQ, but not using DSQOL; however, the difference was statistically significant only at 12 and 24 months (p = 0.067 at 6 months; p < 0.001 at both 12 and 24 months). Furthermore, participants in the pump group reported statistically improved diabetes-specific QoL at 24 months compared with the MDI group (p = 0.006); however, this was not the case at 6 or 12 months and could be due to chance rather than the treatment effect. We note that, if 6-month treatment satisfaction is reanalysed using a mixed-effects linear regression adjusted for baseline score, HbA_1c, centre and course, as was done for the other Qol measures (rather than a non-parametric test), the treatment difference is similar and is statistically significant (p = 0.004), in part due to the increased precision from covariate adjustment.

There were some statistically significant differences on some subdomains, using p < 5% as the level for statistical significance. A caveat is required concerning the number of variables examined and tests performed, as multiple testing was not adjusted for. A statistically significant difference was observed on the social relations domain of the WHOQoL-BREF generic QoL measure in favour of the pump participants (p = 0.026), but this was seen only at 6 months, was one of 12 tests of significance and is likely to be a chance finding.

The DSQOL results (see Table 69) at 24 months showed statistically significant improvements (reductions) in both the pump (mean reduction of 8.2, 95% CI 5.84 to 10.50; p < 0.00001) and MDI (mean reduction 4.2, 95% CI 1.71 to 6.61; p = 0.001) groups, but with greater improvements in the pump group in the overall score (difference 3.8; p = 0.006) and some subdomains. The improvement in DSQOL diet restrictions was larger for the pump group than the MDI group at both 12 and 24 months (12-month adjusted MD in change from baseline –4.1, 95% CI –7.2 to –1.0; p = 0.010; 24-month adjusted MD in change from baseline –5.1, 95% CI –8.6 to –1.6; p = 0.004: lower scores represent better outcomes). A slightly smaller difference was observed at 6 months, which was not statistically significant (adjusted MD –3.3, 95% CI –6.9 to 0.2; p = 0.061). The pump group also had better improvement in DSQOL daily hassle or functions at both 12 and 24 months: at 24 months the score had decreased by 9.6 points in the pump group compared with 3.6 points in the MDI group (adjusted MD –6.3, 95% CI –10.9 to –1.8; p = 0.006).

However, there was a wide spread of changes in DSQOL, with some patients in both groups reporting deterioration at 24 months compared with baseline, as shown in Figure 19.

FIGURE 19

Changes in overall DSQOL at 24 months. (a) Pump; and (b) MDI.

The HFS showed no difference in behaviour score but less worry about hypoglycaemia in the pump arm at 24 months only (p = 0.01). Higher treatment satisfaction by DTSQ was reported by pump users at all time points, and although this was not statistically significant at 6 months, statistical significance was reached at 12- and 24-month follow-up periods, although the absolute difference at 24 months was small at 4.0. EQ5D, SF-12, WHOQOL-BREF and HADS scores showed no differences between groups at any time.

Per-protocol results

Some patients switched from pump to MDI and vice versa, and they may be atypical. We therefore carried out an exploratory per-protocol analyses of psychosocial outcomes after excluding those who switched, and obtained, the following results.

Both groups showed statistically significant improvements in DSQOL as shown in Table 69.

TABLE 69

Overall change in DSQOL within arms at 24 months

The findings were similar with HADS-anxiety – both groups showed small improvements but this reached statistical significance only in the pump arm (Table 70). Large improvements would not be expected because baselines scores were quite low (pump 6.8, MDI 6.1)

TABLE 70

Overall change in HADS-anxiety within arms at 24 months

Hospital Anxiety and Depression Scale-depression scores improved in both groups but the change only reached statistical significance in the pump group (Table 71). Again, baseline scores were low (pump 4.4, MDI 3.7).

TABLE 71

Overall change in HADS-depression within arms at 24 months

Qualitative interpretation: cross-cutting improvements in quality of life

We turn now to qualitative data to (1) help explain the general improvements found across most psychosocial outcomes for both treatment groups and (2) aid the interpretation of those findings that reached statistical significance at more than one time point (i.e. findings relating to treatment satisfaction, dietary restrictions and daily hassles of function).

The overarching improvements in QoL observed in this study mirror those experienced by other cohorts of patients who have attended the DAFNE programme^2,78,124,153 and, arguably, are largely attributable to conversion to a DAFNE approach. Indeed, when they were interviewed after their courses and 6 months later, patients in both arms reported very similar benefits and improvements to their lives. For example, patients in both arms – like other DAFNE graduates who have taken part in longitudinal qualitative research¹⁵⁴ – reported a renewed enthusiasm for managing their diabetes after attending their courses and being more open to discussing aspects of their condition and self-management practices with family and friends. As a consequence, patients also discussed being more open to seeking and accepting support from these family members and friends.

Patients in both arms – like other DAFNE graduates¹⁴³ – also reported feeling more in control of their diabetes/blood glucose levels and more committed to adhering to their treatment regimens (e.g. undertaking SMBG, administering insulin to cover the carbohydrate content of meals/snacks). Notably, however, although participants in the MDI arm tended to attribute these kinds of benefits and improvements to the education and instruction in DAFNE principles received during their courses, those in the pump arm – such as the participant quoted below – tended to accredit them to use of the insulin pump:

Because the pumps given me more awareness, like well if I do eat this and I give myself some insulin for it I’ll need to know what my blood sugar is then, so I will test, so I’ve been doing more tests as a result of doing more insulin with the pump.

P43.2

In addition, patients in both arms reported similar improvements in QoL arising from use of their automated bolus advisors. As described in detail elsewhere,¹⁴³ patients who lacked confidence in their mathematical skills, or whose concentration could be compromised by high/low blood glucose, described the benefits and ‘peace of mind’ that arose from having the advisor to calculate their insulin doses for them. Those who were more confident about their mathematical ability also described liking and benefiting from using their advisor as these devices saved time and effort when calculating doses. Others still reported liking the data storage facility, as this reduced the burden of maintaining a paper diary.¹⁵⁵

Preconceptions about insulin pumps

Many patients described having had misconceptions prior to the trial about how the pump worked and how it could be used to manage their diabetes. Specifically, some described how they had thought that the pump would be a small device implanted under the skin. Others had perceived the pump as being more akin to a closed-loop system, which would alleviate much of the burden of diabetes management by monitoring blood glucose and calculating and administering insulin doses:

I actually thought the pump was some kind of implant . . . and I thought it was something you connected . . . some kind of pipe or cannula and you filled up this implant and then once it was full you disconnected it and then you just had like a remote [control].

P09.1

I think my preconceived ideas were slightly wrong . . . I thought it would be a continuous monitoring system and adjust accordingly . . . And I didn’t realise that you had to keep on testing yourself.

P14.1

Despite some such patients’ initial hopes and expectations not being met, most of those who used pump therapy during the trial described the pump as offering benefits over a MDI regimen. Although some of the benefits described by these patients were also highlighted by those in the MDI arm of the trial (and, hence, arguably were due to the use of the DAFNE approach rather than pump therapy per se), some did appear to arise specifically from use of an insulin pump and these are considered below.

Drip-feeding basal insulin and altering basal rates

Most patients using pump therapy described feeling that they had better control over their blood glucose levels because the device supplied a constant drip-feed of basal insulin, which, as they suggested, more accurately mimicked the natural release of insulin by the pancreas.

Now, because it’s such a little trickle, it’s really, I think, that’s made a huge difference, because it’s made me operate, my body operates more like somebody that’s got a, you know, a pancreas that works.

P24.2

Some patients also highlighted the benefits of being able to set different rates of basal insulin infusion during the day and night. This included P09, who described using a lower basal rate for a specific period of time to counter recurring nocturnal hypoglycaemia, and P19, who reported using higher basal rates to counter rises in blood glucose during periods of inactivity and lower rates when more active (e.g. at weekends):

. . . it’s a lot easier, like, at the moment, my blood sugar tends to dip between midnight and four in the morning, so the pump slightly reduces the insulin . . . whereas on the pen [MDI] then I’d have to reduce the whole of the insulin from before I go to bed until I get up in the morning.

P09.2

[During the week] I’ll sit at my desk until lunchtime, whereas obviously at the weekend I’ll get up, have breakfast, and then I’ll probably go out and about and do something active, so that was . . . weekends were my problem for blood sugar. But that, you know, now I’ve changed that, I’ve put on temporary, er, temporary basals for then, during the morning, and er, it’s been fine.

P19.2

Others highlighted the clinical and personal benefits gained from being able to use a temporary basal rate to accommodate sporting and other physical activities (see Study aims 1 and 2) or, in P18’s case, to minimise the risk of hypoglycaemia after drinking alcohol:

I’d set a temporary basal on it because I was having a drink and so I lowered the basal so as that I could, to stop me hypo-ing through the night sort of thing.

P18.1

Fine-tuning and administering small doses of bolus insulin

As well as being able to alter basal rates, some patients reported additional benefits arising from being able to administer very small and/or precise bolus doses of insulin. Reflecting back on their experiences using an injection regimen, such patients described how this feature had enabled them to more precisely match insulin doses to carbohydrate intake in order to fine-tune their blood glucose control:

I love that you can, you can just give 0.1 of a unit now and before I was on, like, you know, 1 unit, so the accuracy’s much better . . . I’m excited that you can just fine-tune it so much . . . the control that it’s given me already is just fab.

P04.1

. . . it’s more clinical isn’t it, so, you know, it’s easier to be, to be able to drill down into it and to fine-tune it, which is, which is what really I need to do, it doesn’t need to be massive changes, it just needs to be slight, you know, slight changes to make it that much better.

P19.2

As a consequence of being able to administer very precise and small doses, some patients who were sensitive to insulin also described how using the pump had lessened their perceived risk of hypoglycaemia:

If you’re on the edge of going hypo[glycaemic] and you’re having something to eat . . . so you take your insulin, that half a, extra half a unit can send you down again. Whereas on this [pump] you can, like I say, you can fine-tune it to half a unit, so you know exactly what you’re taking. If you need one and a half units for a sausage roll, you’re not trying to think, ‘well, do I take 1 or do I take 2?’. You can take one and a half.

P13.1

Advanced settings: dual- and square-wave boluses

A small number of patients also suggested that they benefited from using advanced pump settings, such as the dual- or square-wave function, to offset the delayed effect of carbohydrate-dense foods, such as pasta, or when eating a meal over an extended period of time:

Then there was the dual wave, you know like when we’ve, if we’ve had pasta and you know your carb[ohydrate]s are going to be long acting and things like that, I think that’s brilliant, whereas before when I were having injections, you just had your injection and then 3, 3 hours later your blood sugar would still be really high.

P05.2

At Christmas time, parties, right? Buffets and things like that, this is a lot easier because you can put it on a dual wave or a square bolus or something and you can forget, you know, right, I’ve dealt with the insulin, and then you can just eat little bits over however many hours, um, and I did that and it worked . . . you couldn’t do that with injections.

P07.1

Wearing the pump prompts patients to perform self-management practices

Although some patients described disliking being connected to the device (see Study aim 5) an additional benefit identified by some individuals was that the pump’s presence prompted them to undertake DAFNE-specific self-management practices, such as SMBG:

. . . it’s a very useful kind of physical manifestation of the fact that, ah, you have this, you have this condition and you’re eating right now, and so do something about your blood test, do something about what you’re eating . . . It’s a very, kind of very useful as a, as a way of, er, reminding you to, you to employ the, er, the techniques . . . that we’re that we’re taught on the DAFNE course.

P23.2

No need to inject

Aside from perceived clinical benefits, many patients reported personal benefits arising from no longer having to inject. Despite having to insert a cannula every 2–3 days, most suggested that this procedure was much less onerous than having to inject five or six times a day:

I know you have to mess about with putting the cannulas in every 3 days, but that’s the biggest hardship. It’s still, you know, going from that . . . er, to like four injections a day, morning, lunchtime, teatime and night-time, when you’re out you’ve got to pull the injection out, stick it in you and stuff like that, it’s, it’s totally different.

P10.1

As reported earlier, in Diabetes Treatment Satisfaction Questionnaire: treatment satisfaction (better for patients using pump therapy) and Diabetes-specific quality of life: daily hassles of functions (better for patients using pump therapy), patients using pumps also described benefits and satisfaction arising from being able to administer insulin without having to inject in front of others and/or to find somewhere private to administer an injection when in a public place. As such, and like the adolescent pump users studied by Lowes et al.,¹⁵⁶ patients also described feeling less noticeable, stigmatised and, hence, detached from others as a consequence:

It’s actually more discreet . . . one person thought it was an iPod [Apple Inc., Cupertino, CA, USA] . . . I find it more discreet because you can take a bolus before a meal without having to expose your skin, which, not everybody likes you injecting in public, it’s easier to take your dose in that way, and it means it’s much easier to fit in.

P12.2

Added benefit of pump therapy

All staff were keen to emphasise that a MDI regimen, taught in conjunction with a DAFNE or similar educational approach, presented a very good and effective toolkit for undertaking diabetes self-management. Hence, educators also suggested that, if they were taught to use a MDI regimen effectively, most patients would neither need nor gain added clinical benefit from using pump therapy:

I think we can maximise most people on DAFNE and it’s wonderful, we really are DAFNE advocates and we’ve had a lot of improvements and reductions in hypos.

D4

However, all educators also noted that, because of the constant drip-feed of insulin, the ability to alter basal rates, and also the ability to titrate and deliver very small insulin doses, pump therapy could potentially help certain groups of patients to improve and/or fine-tune their glycaemic control. As educators described, these individuals were principally those who met current NICE criteria for pump referral,¹³ such as those who suffered from the dawn phenomenon, were very insulin sensitive and/or who undertook a lot of sporting activities that exposed them to risk of hypoglycaemia:

People whose insulin requirements are really small, really low, where sort of injected longer-acting insulin, background insulin, you just can’t adjust them finely enough . . . a pump is great for them because you’ve got the really, you know, minute basal adjustments.

N02

Those that are maybe quite intense when it comes to exercise, you know, there’s definitely a potential for them. Equally, those that are maybe finding that they are on really small doses of insulin because it [the pump] does give them that opportunity to fine-tune.

D05

However, all educators pointed out that, to gain added clinical benefit from using a pump, patients had to be willing and able to their use the pump’s features otherwise, as N3 suggested:

They will just sit on the pump and use it as another method of delivering insulin and they’ll be no better off than on injections.

N3

As is described further later (see Study aim 4), educators also highlighted the difficulties of predicting which patients, or groups of patients, would have this willingness and ability to use the pump to optimal effect.

Pre-trial views about pump candidacy

As indicated earlier (see Study aim 3) educators described having had preconceptions in advance of the trial about the kinds of individuals who would do well on a pump. Specifically, educators discussed how, in their routine clinical practice, in addition to using NICE and other clinical criteria, they had tended to recommend individuals for pump therapy based on tacit and informal assumptions about whether or not they had the right aptitude and technical ability to use the pump to optimal effect. These individuals, as educators also noted, had tended to be those who were younger, technologically savvy and academically able:¹⁵⁷

. . . people who [are] more numerate and the more, the more intelligent, the more, you know, sort of educationally able to take on board all the information.

D03

For similar reasons, educators also described how, despite meeting clinical criteria for pump referral, they had not generally recommended individuals for pump therapy in routine clinical practice if they had a poor history of diabetes self-management, were older or were less academically able.¹⁵⁷ This was a result of their concerns that such individuals would be unwilling or unable to ‘put in the extra work required to use a pump properly’ (N11) and, hence, would not gain any added clinical benefit from the pump as compared with MDI.

Revising preconceptions as a result of trial participation

Educators also described how, as a result of their participation in the REPOSE Trial (for which a randomisation process rather than their own judgement was used to determine who was moved onto the pump), they had been exposed to individuals using pumps who they would not have put forward for this regimen in routine clinical practice. As educators further noted, this kind of exposure had led them to reconsider which kinds of people might gain clinical benefit from using a pump. Specifically, and as detailed elsewhere,¹⁵⁷ educators recounted experiences during which they had observed individuals during the trial ‘doing really, really well on pump therapy who we would have predicted would have really struggled’ (D2), as well as those ‘such as the likes of the young lad who was desperate for a pump and he’s just not using it’ (N9). As a consequence, some educators described how they ‘had stopped having preconceptions about who it will suit and who it won’t’ (N3), whereas others suggested that, in light of their trial experiences, they now thought that motivation – rather than age, technological aptitude or academic ability – should be used as the main criterion (alongside clinical criteria) for determining future pump referrals:

I’ve found that when you actually sit down, show them it, work way through it, actually they become more efficient. So in a way I don’t think there’s anybody that shouldn’t do well on a pump as long as they are keen and motivated.

D4

Others still noted from their experiences of observing patients during the REPOSE Trial that use of a pump could itself act as a tipping point for increased disease self-management among some erstwhile seemingly demotivated patients. As a consequence, such individuals described having reached the conclusion that pumps ‘should potentially be made available to everyone [meeting clinical criteria] because you simply can’t predict, so maybe you need to give everyone a chance?’ (D1).¹⁴⁷

Summary

Educator accounts thus highlight the difficulties of identifying and using patient characteristics to predict potential clinical success using an insulin pump, thereby reinforcing the findings of the quantitative analysis, which showed that it is not possible to determine which patients, compared with others, are likely to do better on the pump.