Post-market Review Products Used in the Management of Diabetes Report to Government Stage 2: Insulin Pumps February 2015

4.4 Second literature review: observational studies

4.4.1 Methodology and limitations

A systematic review was undertaken to assess the efficacy and safety of subcutaneous insulin infusion therapy versus multiple daily insulin injections in observational studies of adults and children with type 1 diabetes published between January 2008 and October 2014. Outcomes considered were blood glucose control (HbA_1c and glycaemic variability) and hypoglycaemia. Studies with a follow-up of less than three months, cross-sectional studies, and studies focussing on specific sub-groups (pregnant women, peri-operative period, cystic fibrosis) were excluded. Evidence synthesis was carried out using a narrative review for each research question.

The literature review focussed on four research questions that are relevant to ToR 8:

1. 
   What is the evidence that continuous subcutaneous insulin infusion therapy is more effective than multiple daily injections for blood glucose control (HbA_1c) in children and adolescents under 18 years of age with type 1 diabetes?
   
2. 
   What is the evidence that continuous subcutaneous insulin infusion therapy is safer than multiple daily injections with regard to the risk of severe hypoglycaemia in children and adolescents under 18 years of age with type 1 diabetes?
   
3. 
   What is the evidence that continuous subcutaneous insulin infusion therapy is more effective than multiple daily injections for blood glucose control (HbA_1c) in adults with type 1 diabetes?
   
4. 
   What is the evidence that continuous subcutaneous insulin infusion therapy is safer than multiple daily injections with regard to the risk of severe hypoglycaemia in adults with type 1 diabetes?
   

The results of observational studies need to be interpreted with caution because of the higher risk of bias, including publication bias, selection bias and unmeasured confounding resulting from lack of randomisation. In before-and-after studies, the changes associated with time (historical data being the comparison), make it impossible to assess whether the effect observed is due to the intervention alone (i.e. use of an insulin pump) or other factors, such as intensity of blood glucose testing or education after introduction of the insulin pump. Conversely, the increase in HbA_1c levels that has been observed in patients with diabetes over time may have partially masked the potential benefit provided by insulin pumps.

In many of the cohort studies identified, the statistical tests compared the end-of-study HbA_1c levels between the two treatments without explicit adjustment for baseline HbA_1c. Given that with both therapies, reduction in HbA_1c may progressively increase as baseline HbA_1c rises (Retnakaran 2005), and that poor glycaemic control is often a selection criterion for initiation of insulin pump therapy, the improvements in glycaemic control seen with insulin pump therapy may result from higher initial HbA_1c. The variation in definition of severe hypoglycaemia outcome across studies adds difficulty to the interpretation of the synthesised results for addressing any difference in benefit. Nevertheless, observational studies do provide useful data on effects observed in clinical practice.

A 0.5% reduction or more in HbA_1c is generally accepted to be of clinical importance (Clar 2010; Cummins 2010). However, there is no consensus on this issue and a smaller reduction might be considered important from a public health perspective if achieved on a wide scale (Farmer 2012).

4.4.2 Results

A total of 37 observational studies were identified, including 23 before-and-after studies, nine retrospective cohort studies and five prospective cohort studies. The studies were conducted in 24 different countries and the majority (62%) had a followup period of two or more years (range of three months to nine years).

There were 21 studies in children and adolescents under 18 years old, one study in adolescents and adults, 14 studies in adults, and one study including all age groups. The number of patients included in the studies varied between 10 and 2709, with eight studies including over 200 patients.

A variety of pumps were used in the studies; most studies (21) did not report the type of insulin pumps used. Twenty-three studies did not report the source of funding, seven received public funding, three industry funding, and four both public and industry funding.

Indications for insulin pump therapy were broadly similar between studies: poor glycaemic control, frequent hypoglycaemia, dawn phenomenon, need for lifestyle flexibility, and needle phobia. Limited information was reported on the education provided to patients at either the time of initiating insulin pump therapy or subsequently.

4.4.2.1 What is the evidence that continuous subcutaneous insulin infusion therapy is more effective than multiple daily injections for blood glucose control in children and adolescents under 18 years of age with type 1 diabetes?

HbA_1c and glycaemic variability

Of 13 before-and-after studies, 12 showed a decrease in HbA_1c (0.1-1.04%) with insulin pump therapy that was statistically significant in nine studies. Long term studies generally showed positive results for insulin pump therapy, with statistically significant HbA_1c reductions of 0.4-0.7% at 2–5 years follow-up in four studies, a nonsignificant reduction in one study (Abaci 2009), and no reduction in a sixth study (Knight 2011).

Of nine cohort studies, seven showed statistically significant decreases in HbA_1c with insulin pump therapy; however, the changes were less than the 0.5% decrease that is usually considered clinically meaningful (Cummins 2010) in all but one Australian study (Johnson 2013). Most studies did not explicitly adjust for pre-study HbA_1c. One study of around 100 children found a statistically significant mean difference in HbA1c of 0.6% (95% CI 0.1-1.2) between the multiple daily injection group and the insulin pump group after adjusting for baseline differences in HbA_1c (Senniapan 2012). The Australian study surveyed a total of 345 children including 129 children with insulin pumps matched to 129 children with multiple daily injection. It found a mean improvement in HbA_1c over a five-year period of 0.7% for those using insulin pumps. In the largest study in 868 children and adolescents a significant decrease was observed only in the first year and not in the following years (Jakisch 2008).

Age

Hughes (2012) observed that the highest HbA_1c reductions occurred in children under 6 years (-0.9% at six months follow-up, p < 0.01). In another study, the mean HbA_1c levels were significantly lower for those who initiated insulin pump therapy before age 6 throughout the entire follow-up period (p = 0.02) (Levy-Shraga 2013). These findings suggest that insulin pump therapy may be more beneficial for children under 6 years. However, these two studies included low numbers of children.

4.4.2.2 What is the evidence that continuous subcutaneous insulin infusion therapy is safer than multiple daily injections with regards to the risk of severe hypoglycaemia in children and adolescents under 18 years of age with type 1 diabetes?

Children and adolescents under 18 years of age with type 1 diabetes appear to have a lower risk of severe hypoglycaemia when using an insulin pump than with multiple daily injections.

Of five before-and-after studies identified that considered severe hypoglycaemia, four reported decreases in severe hypoglycaemia, which were statistically significant in three studies (Campbell 2009; Hasanbegovic 2009; Rabbone 2009). Only one study of 17 adolescents reported an increase in the rate of severe hypoglycaemia with insulin pump therapy (Abaci 2009). In the before-and-after studies, statistically significant differences in the rate of severe hypoglycaemia per 100 patient-years in children and adolescents using insulin pump therapy compared to multiple daily injections ranged from -6.4 to -62. The definitions of severe hypoglycaemia varied, making comparison difficult.

Seven cohort studies found a decrease in the rates of severe hypoglycaemic events with the use of insulin pump therapy, which was statistically significant in three studies (Berghaeuser 2008; Jakisch 2008; Johnson 2013). In the cohort studies, statistically significant differences in the rate of severe hypoglycaemia per 100 patient-years in children and adolescents using insulin pump therapy compared to multiple daily injections ranged from -2 to -17. Again, the definitions of severe hypoglycaemia varied.

One study reported significant decreases in the rates of symptomatic and nocturnal hypoglycaemic events with insulin pump therapy (Çamurdan 2008).

4.3.2.3 What is the evidence that continuous subcutaneous insulin infusion therapy is more effective than multiple daily injections for blood glucose control in adults with type 1 diabetes?

HbA_1c and glycaemic variability

Adults with type 1 diabetes appear to achieve a greater reduction of HbA_1c when using an insulin pump than with multiple daily injections.

Ten before-and-after studies assessed HbA_1c. With the exception of one study, all showed a statistically significant decrease in HbA_1c with insulin pump therapy at all follow-up periods with significant mean decreases ranging from 0.4–2.6%. Aberle (2008) and Shalitin (2010) showed that patients with a higher initial HbA_1c at the start of insulin pump therapy showed greater decreases in HbA_1c, compared to those with lower baseline HbA_1c.

Three cohort studies, involving 220, 247 and 2709 patients respectively, showed statistically significant decreases in HbA_1c levels with insulin pump therapy at one, two, three and five years of follow-up (Lepore 2009; Carlsson 2013) or at 6 months, 1 and 2 years of follow-up (Cohen 2013). In the largest cohort, Carlsson (2013) found that the reduction in HbA_1c observed with insulin pump therapy compared to multiple daily injections was greatest in the first two years, and was estimated to be 0.20% at five years (95% CI 0.07% to 0.32%), which is less than the 0.5% reduction usually considered clinically significant. A similar trend was observed in a cohort of 247 patients where the decrease in HbA_1c between the two groups declined progressively over time and was only significant up to 24 months (Cohen 2013).

Four studies found a significant decrease in glycaemic variability before insulin pump therapy to after insulin pump therapy (Aberle 2008; Lin 2011; Gimenez 2010; Maiorino 2014). Two cohort studies found a significant reduction in glycaemic variability with insulin pumps compared to multiple insulin injections (Maiorino 2014; Cohen 2013).

4.4.2.4 What is the evidence that continuous subcutaneous insulin infusion therapy is safer than multiple daily injections with regards to the risk of severe hypoglycaemia in adults with type 1 diabetes?

Adults with type 1 diabetes appear to have a lower risk of severe hypoglycaemia with insulin pump therapy than with multiple daily injections. Five before-and-after studies reported decreased rates of severe hypoglycaemic events with insulin pump therapy in adults; three with significant results (Marmolin 2012; Shalitin 2010; Gimenez 2010). Statistically significant differences in the rate of severe hypoglycaemia per 100 patient-years in adults using insulin pump therapy compared to multiple daily injections ranged from -3 (where severe hypoglycaemia was defined as coma, seizures or administration of glucagon injection or intravenous glucose)(Shaitlin 2010) to -120 (where severe hypoglycaemia was defined as required assistance from another person) in a study which included patients at high risk of hypoglycaemia ( Gimenez 2010).