Stem Cells and Type 1 Diabetes: What the Future Has in Store

Stem cell


A pancreas transplant has always stood out as a possible ‘cure’ for type 1 diabetes (T1D), but one problem has been obvious: there just are not enough organ donors-on the order of 10,000 a year-while there are between 1 and 2 million people with T1D in the U.S. In a kidney transplant, a healthy donor can donate one of two functioning kidneys with a generally low-risk surgery, and still have normal kidney function. A similar approach with part of the pancreas would be unsafe. In addition, a pancreas transplant is generally less successful than a kidney transplant, and there are higher risks of serious side effects after pancreatic transplant surgery. The math is even worse when trying to transplant insulin-producing islets, because more than one donor is needed per recipient, which has stopped islet cell transplant from taking hold outside of a few centers. Furthermore, transplants of any sort require lifelong use of powerful and expensive medications that suppress immune function and can also cause serious side effects.

But what if we could transplant insulin-producing cells made in the lab? Wouldn’t that solve the donor dilemma? Yes, but the recipient with by far the most common form of T1D would still require immune suppression. Their immune system already destroyed, and is continuing to destroy, their insulin- producing beta cells. This would be true even if the insulin producing cells were derived from their own tissue. But what if we could protect new insulin-producing cells from the recipient’s immune system another way?

It is now possible to manufacture insulin-producing cells in the lab, using multiple different techniques developed by a multitude of researchers (Type 1 Diabetes Treatments Based on Stem Cells, Arana et al., Current Diabetes Reviews, 2018, 14, 14-23). That is a huge step forward, and a tribute to the benefit of supporting basic and applied research. Researchers are working on ways to ‘hide’ the new cells from the recipient’s immune system by altering the cells immune ‘appearance’, or more selectively suppressing the immune attack by the host. Hopefully, those efforts will pay off someday. But how about putting the new cells behind a barrier that the immune system cannot get through?

ViaCyte, a privately-held bioresearch company, reported some intriguing results at this year’s American Diabetes Association Scientific Sessions: the two-year data from the ongoing Safety, Tolerability, and Efficacy of PEC-Encap™ Product Candidate in type 1 diabetes (STEP ONE) clinical trial. The PEC-Encap consists of stem cell-derived cells that can develop into insulin-producing cells, encapsulated in a delivery device that is surgically implanted under the skin, called the Encaptra® Cell Delivery System. This system is designed to block immune access to the new cells but allow insulin, glucagon, glucose and other nutrients to pass through the membrane. The results indicate that the PEC-Encap product did not trigger a specific immune response against the new cells or the device itself, and it appeared to be safe. That’s the good news. Unfortunately, few of the implanted devices allowed enough new blood vessel growth from the host to sufficiently nourish the new cells, so in most cases, the new insulin-producing cells did not last. This appeared to result primarily from a foreign body reaction, a non-specific response of the recipient’s immune system that is similar to what one might find develop around a splinter. ViaCyte is now working on modifying the system to improve the potential for long-term survival of the manufactured insulin-producing cells.

If these or other similar efforts are successful, a large percentage of those with T1D could ultimately receive a functional ‘cure’. In addition, those with long-term type 2 diabetes (T2D) who can no longer produce much insulin, a common state that makes blood sugar management very difficult, might also benefit from this promising new therapy.

A second, perhaps less ambitious device is also under development, PEC-Direct™, one which would still require the use of immunosuppression medication. However, since the cells can be generated in a lab in potentially unlimited numbers, there is no need for organ donors. Thus, a much larger group of people might be able to benefit from transplanted insulin-producing cells, albeit with the need for immunosuppression. The current plan is to consider such a transplant for those with T1D who suffer from recurrent severe hypoglycemia episodes or have hypoglycemia unawareness, conditions which are life-threatening. Those who are unable to manage T1D effectively due to highly variable blood glucose levels, so-called ‘brittle’ diabetes, could also benefit. Together, such groups are thought to represent about 10% of all people with T1D.

In summary, there is great news in the stem cell arena; insulin-producing cells can be made in unlimited numbers. While not yet ready for clinical use in people with diabetes, rapid progress is being made. We waited for finger sticks to become available, so we could finally see what we so desperately needed to see–where is my blood glucose, right now. We waited for insulin pumps and better insulins, so we could do what we so desperately needed to do, right now-tame T1D’s wild blood glucose fluctuations. We waited for continuous glucose monitoring, so we could know what we so desperately needed to know- where is my blood sugar going, right now. Stems cells have the potential to deliver what we all still so desperately want- relief from the 24/7/365 burden of thinking and acting like a beta cell. Stay tuned, T1D nation!

Nicholas B. Argento, MD, Diabetes Technology Director, Maryland Endocrine and Diabetes

8 Reasons You’re Waking Up Mid-Sleep, and How to Fix Them

Talk about a rude awakening.
woman laying in bed at night on her cell phone

One minute you’re snoozing peacefully, the next you’re wide awake in the dead of night. Sound familiar? Unless you’re blessed enough to conk out like the most determined of logs, you may have experienced this form of sleeplessness before. Waking up during the night isn’t uncommon—a study of 8,937 people in Sleep Medicine estimates that about a third of American adults wake up in the night at least three times a week, and over 40 percent of that group might have trouble falling asleep again (this is sometimes referred to as sleep maintenance insomnia).

So, what’s causing you to wake up in the middle of the night, and how can you stop it from happening? Here are eight common reasons, plus what you can do to get a good night’s rest.

1. Your room is too hot, cold, noisy, or bright.

Your arousal threshold—meaning how easy it is for something to wake you up—varies depending on what sleep stage you’re in, Rita Aouad, M.D., a sleep medicine physician at The Ohio State University Wexner Medical Center, tells SELF.

When you sleep, your body cycles through different sleep stages: 1, 2, 3, 4, and rapid-eye movement (REM). (Some schools of thought lump together stages 3 and 4.) The first stage of sleep is the lightest, Dr. Aouad explains. That’s when you’re most likely to startle awake because a door slams, a passing car’s headlights shine into your window, or because of some other environmental factor like your room being too hot or cold.

Ideally, your room should be dark, comfortably cool, and quiet when you sleep. This might not all be under your control, but do what you can, like using earplugs and an eye mask to block out errant noise and light, or buying a fan if your room is stifling.

2. You have anxiety.

Anxiety can absolutely wake you up at night,” Nesochi Okeke-Igbokwe, M.D., a physician in New York, tells SELF. In fact, trouble sleeping is one of the most common symptoms of an anxiety disorder, according to the Mayo Clinic. That’s because you can experience anxiety-induced issues that are severe enough to rouse you, like a galloping heartbeat or nightmares.

“Additionally, there are people who may experience what are called nocturnal panic attacks, meaning they may have transient episodes of intense panic that wake them up from their slumber,” Dr. Okeke-Igbokwe says.

If your anxiety regularly wakes you up, Dr. Okeke-Igbokwe recommends mentioning it to your doctor, who should be able to help you get a handle on any underlying anxiety or panic disorder at play. Doing so may involve cognitive behavioral therapy, anti-anxiety medication, or a combination of the two. “Meditation and deep-breathing exercises can also sometimes alleviate symptoms in some people,” Dr. Okeke-Igbokwe says.

3. Your full bladder can’t wait until the morning.

Nocturia—a condition that’s generally viewed as getting up to pee at least once during the night, though some experts say that’s not often enough to qualify—appears to be fairly common. A study in the International Neurourology Journal found that out of the 856 people surveyed, around 23 percent of women and 29 percent of men experienced nocturia.

Causes of nocturia include drinking too much fluid before bedtime, urinary tract infections, and an overactive bladder, per the Cleveland Clinic. Untreated type 1 or type 2 diabetes may also be a factor; having too much sugar in your bloodstream forces your body to extract fluid from your tissues, making you thirsty and possibly prompting you to drink and pee more, according to the Mayo Clinic.

If cutting back on your evening fluid intake doesn’t reduce your number of nightly bathroom trips, consult a doctor for other possible explanations.

4. You had a couple of alcoholic drinks.

Sure, alcohol can make it easy to drift off—even when you’re, say, on a friend’s couch instead of tucked into your bed—but it also has a tendency to cause fitful sleep. This is because alcohol can play around with your sleep stages in various ways. For instance, it seems as though alcohol is associated with more stage 1 sleep than usual in the second half of the night. Remember, stage 1 sleep is the period in which you’re most likely to wake up due to environmental factors. So if you’re looking for quality, sleep-through-the-night rest, it’s worth taking a look at how much alcohol you’re consuming.

Everyone metabolizes alcohol differently depending on factors like genetics, diet, and body size. However, Alexea Gaffney Adams, M.D., a board-certified internist at Stony Brook Medicine, recommends that people stop drinking at least three hours before going to bed to give their bodies time to process the alcohol. Since drinking often happens at night, we realize that can be an optimistic time cushion. Based on your personal factors and how much you drank, you might not need that much. But having some kind of buffer—and drinking plenty of water so you’re more likely to booze in moderation—may prevent alcohol from interfering with your sleep.

Also, Dr. Gaffney Adams notes that drinking alcohol too soon before bed will make you need to pee, increasing the likelihood you’ll wake up in the night to use the bathroom. Double whammy, that one.

5. You’ve got sleep apnea.

If you find yourself jolting awake and feeling like you need to catch your breath, sleep apnea might be the culprit. This disorder slows and/or stops your breathing while you are asleep.

If you have obstructive sleep apnea, the muscles in your throat relax too much, which narrows your airway, causing your oxygen levels to drop, the Mayo Clinic explains. If you have central sleep apnea, your brain doesn’t send the right signals to the muscles controlling your breathing, again causing this potentially harmful drop in oxygen. Complex sleep apnea features characteristics of both conditions.

To diagnose sleep apnea, your doctor may have you do an overnight sleep study that monitors your breathing, according to the Mayo Clinic. The most common treatment for sleep apnea is a continuous positive airway pressure (CPAP) machine, which is basically a mask you wear during sleep to help keep your airways open, but your doctor can help you explore the alternatives if necessary.

6. You have an overactive thyroid gland.

“This gland controls the function of several other organs,” Dr. Gaffney Adams tells SELF. When it’s overactive (also called hyperthyroidism), it creates too much of the hormone thyroxine, which can have ripple effects on many different systems in your body, according to the Mayo Clinic. Common symptoms of an overactive thyroid include trouble sleeping, an increased heart rate, sweating (including at night), anxiety, tremors, and more.

Your primary care physician or an endocrinologist (a doctor specializing in hormones) can test your blood to evaluate your hormone levels. If you do have an overactive thyroid, your doctor can walk you through the potential ways of treating it, including medications to slow your thyroid’s hormone production and beta blockers to reduce symptoms like a wild heartbeat.

7. You ate right before bedtime, or you didn’t eat recently enough before you went to sleep.

“Eating too heavy of a meal too close to bedtime can make it difficult to fall asleep or stay asleep,” Dr. Aouad says. One potential reason behind this is acid reflux, which is when your stomach acid moves up into your throat and causes painful nighttime heartburn. And if you eat food right before bed that makes you gassy, the resulting abdominal pain could drag you out of dreamland, too.

On the flip side, going too long without eating before you sleep can also cause this type of insomnia, Dr. Aouad says. There’s the simple fact that your growling, crampy stomach can wake you up. Hunger could also mess with your blood sugar while you sleep, especially if you have diabetes. Going too long without eating can provoke hypoglycemia, which is when your blood sugar drops too low. This can lead to restless sleep, per the Cleveland Clinic, along with issues like weakness or shaking, dizziness, and confusion. Although hypoglycemia can happen to anyone, it’s much more likely in people with diabetes. If you have the condition, work with your doctor on a plan for keeping your blood sugar stable, including during sleep.

8. You have restless legs syndrome.

Restless legs syndrome, or RLS, may make your lower extremities feel like they are throbbing, itching, aching, pulling, or crawling, among other sensations, according to the National Institute of Neurological Disorders and Stroke (NINDS). If you have RLS, you’ll also feel an uncontrollable urge to move your legs. These symptoms are most common during the evening and night and become more intense during periods of inactivity, like…you guessed it, sleep.

Experts aren’t totally sure what causes RLS, but it seems as though there’s a hereditary factor in the mix, according to the NINDS. Researchers are also investigating how issues with dopamine, a neurotransmitter your muscles need to work correctly, may cause RLS. Sometimes there are other underlying issues bringing about RLS as well, such as iron deficiency.

After diagnosing you with RLS via questions and lab exams, your doctor may prescribe medications to increase your dopamine levels or other drugs, such as muscle relaxants. They may also be able to counsel you on home remedies to soothe your muscles, like warm baths.

To sum it up, there are a bunch of possible reasons you are waking up at night. Some are pretty easy to change on your own, others not so much.

If you think all you need to do to fix this is tweak a habit, like falling asleep with the TV on or chugging a liter of water before bed, start there. If you’ve done everything you can think of and still don’t see a change, it’s worth mentioning your nighttime wakeups to an expert who can help you stay put after you drift off.


Managing type 1 diabetes in adolescents and kids : New position statement by ADA

Do Germs Cause Type 1 Diabetes?

More Coffee May Up Metabolic Syndrome Risk in Type 1 Diabetes

In a snapshot cross-sectional analysis of study participants with type 1 diabetes in Finland, drinking three or more cups of filtered coffee a day was associated with increased odds of having metabolic syndrome, in contrast to previous findings in the general population.

The article, by B Stutz of the University of Helsinki, Finland, and colleagues, based on data from the Finnish Diabetic Nephropathy (FinnDiane) study, was published online February 1 in Nutrition, Metabolism & Cardiovascular Diseases.

“In contrast to the previous observations in other populations, coffee consumption was associated with higher odds of [metabolic syndrome] in the current study” in people with type 1 diabetes, the researchers report.

Among these individuals with type 1 diabetes, those who drank moderate or high amounts of coffee had a significantly increased risk of metabolic syndrome compared with those who did not drink coffee, after adjusting for multiple variables.

Those who drank any amount of coffee were more likely to have hypertension.

The reasons for these findings remain to be determined and more study is also needed to see how coffee consumption might affect health outcomes in people with type 1 diabetes, the researchers conclude.

Association Mainly Driven by Hypertension

In the general population, regular coffee drinkers have been reported to have a lower risk of developing metabolic syndrome, and thus a lower risk of cardiovascular disease and early mortality.

However, it has not been clear whether patients with type 1 diabetes who are regular coffee drinkers would also have a lower risk of metabolic syndrome.

To investigate this, researchers examined 1040 participants from the FinnDiane study who had type 1 diabetes — defined as diabetes onset before age 35 years and permanent insulin therapy within a year of diagnosis — and complete data from a diet questionnaire along with information on metabolic syndrome criteria.

Metabolic syndrome was defined as having three of the following five criteria: waist circumference ≥ 94 cm in men and ≥ 80 cm in women; triglycerides ≥ 1.7 mmol/L or on lipid-lowering medication; high-density lipoprotein cholesterol (HDL-C) < 1.0 mmol/L in men and < 1.3 mmol/L in women, or on medication to increase HDL-C; blood pressure ≥ 130/85 mmHg or use of antihypertensive medication; and fasting blood glucose ≥ 110 mg/dL.

Participants were classed as having the following coffee consumption levels:

  • None: < 1 cup/day, 134 participants (13%)
  • Low: 1–2 cups/day, 230 participants (22%)
  • Moderate: 3–4 cups/day, 371 participants (36%)
  • High: ≥ 5 cups/day, 305 participants (29%)

Of the 906 coffee drinkers, 825 (91%) drank filtered coffee, so researchers pooled data for the different types of coffee.

On average, individuals who did not drink coffee were 40 years old and had diabetes for 21 years, whereas coffee drinkers were older (mean age 46 to 49 years) and had diabetes longer (mean duration 27 to 30 years).

There were more current smokers in the group that had a high consumption of coffee (22%) than in the groups that drank less (up to 12%) or no coffee (3%).

There were also more men in the group who drank 5 or more cups of coffee a day (60%) than in the groups that drank less or no coffee (32% to 45%).

Coffee drinkers were more likely to be taking antihypertensive or lipid-lowering medications than nondrinkers.

The prevalence of metabolic syndrome increased stepwise from 51% to 64% to 65% to 70% among people whose coffee consumption was none, low, moderate, or high, respectively.

Compared with noncoffee drinkers, those who drank 3 or 4 cups a day had a 1.8-fold increased risk of having metabolic syndrome, and those who drank 5 or more cups a day had a 2.1-fold increased risk (P < .05 for both), after adjusting for age, sex, total calories, alcohol, physical activity, and smoking.

Similarly, compared with noncoffee drinkers, those who drank 1 or more cups of coffee a day had a 2.2- to 2.8-fold increased risk of hypertension, after adjusting for multiple variables (P < .05).

“The association between coffee consumption and [metabolic syndrome] seems to be mainly driven by the blood pressure component,” Stutz and colleagues observe.

Methyldopa found effective for preventing onset of type 1 diabetes

Making Smart Food Choices With T1D

Making healthy food choices can be very cumbersome for those with type 1 diabetes (T1D) because many different guidelines exist and parameters are constantly changing. Moreover, for children with the disease there is limited research examining their different nutritional needs compared with children without diabetes.

People with diabetes are encouraged to eat low glycemic index foods to help prevent microvascular complications. As a result, many patients limit carbohydrate intake and potentially choose food options higher in fat or protein.

As a person with type 1 diabetes, this line of thinking certainly happens when I become frightened of blood sugar surges. For example, I may eat that extra meatball in place of additional pasta. Dietary choices to specifically avoid “evil carbohydrates” and consequently consume more protein or fat from calories may help prevent acute blood sugar fluctuations, but could have unintended consequences such as increased risk of cardiovascular disease

The United States Department of Agriculture’s (USDA) current guidelines, MyPlate, advise the general population to make healthy dietary choices from five food groups – fruits, vegetables, grains, proteins and dairy. MyPlate states that total fat intake should be 30%-35% of calories for children 2-3 years of age, 25%-35% of calories for children 4-8 years of age, and for those > 19 years old 20%-35%. The 2013 American Diabetes Association (ADA) guidelines recommend total fat intake <30% and saturated fat intake <7% in children and <10% in adults . Additionally, much of ADA recommendations encompass guidelines for both type 1 and type 2 diabetes for which disease management varies greatly.

A review in the Diabetes Educator examined 9 studies that looked at dietary intake of children with T1D and found some results showing potential dietary inadequacies. Fruit, vegetable, and fiber intakes were low, but did not differ statistically from children without diabetes. However, in three of the studies, mean intake for total fat were 11%-15%, which exceeded ADA recommendations (>7%). Another study in Diabetes Carecompared dietary recall between 132 adolescents with T1D and 131 adolescents without T1D and found significant difference in fat intake between the groups. Specifically, those adolescents with T1D consumed more of all kinds of fat and male subjects with substantially more saturated fat compared with males without diabetes.

As I noted before, it is likely that anxiety associated with counting carbohydrates can lead inadvertently to choosing a diet higher in fat, and specifically saturated fats. Additionally, less consumption of carbohydrates can potentially lower insulin requirements. For example, when I am busy I may opt for a snack not requiring an insulin injection such as cheese. This may be okay on an occasion, but a diet high in saturated fats does have its consequences. But many children are not properly educated about which foods are even high in fat and what alternatives are available. In one quantitative study of children with T1D, foods such as cheese, bacon and steak were perceived as good choice foods because they do not contain carbohydrates.

So what can be done to help improve dietary choices amongst individuals with diabetes and specifically children and adolescents?

  • They should have access to diabetes educators allowing for formal teaching on how to make healthy choices. Advise them to choose more plant-based alternatives (beans, lentils, nuts and nut butters, seeds, peas, and soy foods) or lean meats (fish, seafood, chicken, turkey, and yogurt)
  • Along with education describing good fat such as avocado, teach that there are good carbohydrates. Complex carbohydrates include brown rice, whole wheat, quinoa, oatmeal, fruits, vegetables, beans, and lentils. This is a great resource to reference.
  • Nuts are an excellent source of vitamins and nutrients and can easily be supplemented in daily snacks such as oatmeal and yogurt. Additionally, they are low in carbohydrates, high in protein and while high in fats they keep you full longer and can prevent overeating. However, there is such a thing as too much of a good thing as nuts are very caloric. They contain about 160 to 200 calories per ounce, which is about 24 almonds, 18 cashews, 48 pistachios or 14 walnut halves. Thus, nuts should be appropriately portioned to prevent weight gain. has an excellent resource for this.

Basically, from my experience and research, dietary recommendations are not one-size-fits-all for anyone, and this is particularly true and critical for everyone with diabetes, no matter their age.

What Can Exercise Do for People With Type 1 Diabetes?

exercise and diabetesexercise and diabetes

In a meta-analysis done to look at exercise training in those with type 1 diabetes, researchers report which benefits were observed.

They sought to “establish the relationship between exercise training and clinical outcomes in people with type 1 diabetes.”

The study authors searched for prospective randomized or controlled trails involving exercise training in people with type 1 diabetes for 12 or more weeks though MEDLINE, Cochrane Controlled Trials Registry, CINAHL, SPORTDidscus, and Science Citation Index.

What Does Exercise Help With if You Have Type 1 Diabetes?

In those who exercised, researchers found that exercise lowered daily insulin needs, BMI (body mass index), peak VO2, resting heart rates, resting systolic blood pressure (the top number), LDL cholesterol, and triglycerides.

Children who exercised, specifically had lowered insulin doses, waist circumference, and triglycerides.

They didn’t find any effects from exercise on A1c levels however, nor fasting blood glucose, body mass, or HDL cholesterol levels.

What About  You?

If you have type 1 diabetes, what does exercise personally help you with? Share in the comments!

Many Cases of Type 1 Diabetes Happen in Adults and These Diagnoses Are Tough

adults diagnosed with type 1

Years ago, type 1 diabetes was often called “juvenile diabetes” which is a term that wasn’t accurate because of how not all type 1 diabetes diagnosis happen in childhood. Over time we’ve learned just how inaccurate that term indeed was.

Among those with a high genetic risk for type 1 diabetes, at what age do they most often develop type 1 diabetes well into adulthood and what are the clinical characteristics of those patients?

Researchers conducted a cross-sectional analysis using a type 1 diabetes genetic risk score that was based on 29 common variants to pinpoint individuals of white European descent in the UK Biobank “in the half of the population with high or low genetic susceptibility to type 1 diabetes, ” wrote the researchers in their study abstract.

They looked at the number of cases of diabetes in both of those groups over their first 60 years of life and then “genetically defined type 1 diabetes as the additional cases of diabetes that occurred in the high genetic susceptibility group compared with the low genetic susceptibility group”.

Remaining cases were defined as type 2 diabetes and the clinical characteristics of the groups with genetically defined type 1 or 2 diabetes were investigated.

Study Results

Researchers found that 13,250 of the 379,511 individuals from the UK Biobank had developed diabetes in their first 60 years of life. They write that 537 or 42% of individuals were diagnosed between age 31-60 which comprised 4% of the total diabetes cases diagnosed after age 30.

The clinical characteristics in those diagnosed with type 1 diabetes after age 30 were similar to those diagnosed with type 1 under age 30 and different from those diagnosed with type 2 diabetes. All those diagnosed with type 1 diabetes generally had a lower BMI, were more likely to use insulin in their first year of diagnosis, and were more likely to develop diabetic ketoacidosis.

Diagnosing Type 1 in Adults is Daunting

These findings show that type 1 diabetes often develops in individuals after age 30, but since these cases of type 1 diabetes are such a small percentage of cases, with type 2 diabetes being the overwhelming majority and therefore likelihood, properly identifying type 1 diabetes in this age group is very challenging.

Yet, failing to do so can lead to “serious consequences because these patients rapidly develop insulin dependency”.

The researchers state that tests that check C-peptide and islet-specific autoantibodies can be used to figure out if a patient has type 1 or 2 diabetes but these tests aren’t routinely done and sometimes aren’t enough.

They write that “Progression to absolute insulin deficiency, defined by measurement of serum C-peptide concentration, can be used to identify type 1 diabetes, but is only useful 3–5 years after diagnosis.”

Sometimes, to diagnose type 1 diabetes, healthcare providers use a test that looks for autoantibodies to the GAD islet antigen but that isn’t a sure thing because only about 70% of type 1 patients have those autoantibodies.

All this despite the other complications for professionals trying to figure out which diabetes a patient has like the way that patients diagnosed later in life tend to have a slower progression of type 1 diabetes compared to children which can make them look like a type 2, at least initially.

In conclusion, the researchers explain that half of the population has a very low genetic risk for type 1 which means that if they develop diabetes, it is most likely to be type 2 diabetes. The other half of the population with a high risk for type 1 diabetes has a likelihood for either type 1 or 2 diabetes, leaving medical professionals struggling to properly identify patients.

Were you diagnosed with type 1 diabetes as an adult? Was there trouble getting a diagnosis? Please share in the comments.

In Type 1 Diabetes, Patient Training More Important Than Choice of Multiple Insulin Injections vs. Pump

When it comes to managing type 1 diabetes, patient education and training in insulin delivery — rather than choosing between an insulin pump and injections — may be key, according to a study in The BMJ.

Over 250 adults with type 1 diabetes were randomized to receive multiple daily insulin injections or insulin pump therapy, plus structured group training on flexible insulin dosing based on diet, activity, and blood glucose levels. The training took place over five consecutive days. At baseline, over 90% of participants had hemoglobin A1c levels of 7.5% or higher.

At 24 months, both groups had achieved clinically significant reductions in mean HbA1c level, with no significant difference between the groups. In addition, about a quarter of patients in each group had HbA1c levels below 7.5%.

The authors write, “These results do not support a policy of providing insulin pumps to adults with poor glycemic control until the effects of training on participants’ level of engagement in intensive self management have been determined.”



Objective To compare the effectiveness of insulin pumps with multiple daily injections for adults with type 1 diabetes, with both groups receiving equivalent training in flexible insulin treatment.

Design Pragmatic, multicentre, open label, parallel group, cluster randomised controlled trial (Relative Effectiveness of Pumps Over MDI and Structured Education (REPOSE) trial).

Setting Eight secondary care centres in England and Scotland.

Participants Adults with type 1 diabetes who were willing to undertake intensive insulin treatment, with no preference for pumps or multiple daily injections. Participants were allocated a place on established group training courses that taught flexible intensive insulin treatment (“dose adjustment for normal eating,” DAFNE). The course groups (the clusters) were then randomly allocated in pairs to either pump or multiple daily injections.

Interventions Participants attended training in flexible insulin treatment (using insulin analogues) structured around the use of pump or injections, followed for two years.

Main outcome measures The primary outcomes were a change in glycated haemoglobin (HbA1c) values (%) at two years in participants with baseline HbA1c value of ≥7.5% (58 mmol/mol), and the proportion of participants achieving an HbA1c value of <7.5%. Secondary outcomes included body weight, insulin dose, and episodes of moderate and severe hypoglycaemia. Ancillary outcomes included quality of life and treatment satisfaction.

Results 317 participants (46 courses) were randomised (156 pump and 161 injections). 267 attended courses and 260 were included in the intention to treat analysis, of which 235 (119 pump and 116 injection) had baseline HbA1c values of ≥7.5%. Glycaemic control and rates of severe hypoglycaemia improved in both groups. The mean change in HbA1c at two years was −0.85% with pump treatment and −0.42% with multiple daily injections. Adjusting for course, centre, age, sex, and accounting for missing values, the difference was −0.24% (−2.7 mmol/mol) in favour of pump users (95% confidence interval −0.53 to 0.05, P=0.10). Most psychosocial measures showed no difference, but pump users showed greater improvement in treatment satisfaction and some quality of life domains (dietary freedom and daily hassle) at 12 and 24 months.

Conclusions Both groups showed clinically relevant and long lasting decreases in HbA1c, rates of severe hypoglycaemia, and improved psychological measures, although few participants achieved glucose levels currently recommended by national and international guidelines. Adding pump treatment to structured training in flexible intensive insulin treatment did not substantially enhance educational benefits on glycaemic control, hypoglycaemia, or psychosocial outcomes in adults with type 1 diabetes. These results do not support a policy of providing insulin pumps to adults with poor glycaemic control until the effects of training on participants’ level of engagement in intensive self management have been determined.

Trial registration Current Controlled Trials ISRCTN61215213.


People with type 1 diabetes mellitus require lifelong treatment with insulin to prevent diabetic ketoacidosis and to optimise blood glucose levels to minimise vascular complications.1 Insulin is generally administered by multiple daily subcutaneous injections, using different insulins to cover background and meal requirements. Doses are adjusted according to eating, physical activity, and blood glucose level. This approach and its integration within flexible lifestyles is promoted in “dose adjustment for normal eating” (DAFNE)2 and similar structured training courses.3 Despite this training and best efforts, many people struggle to achieve glycaemic targets and a considerable proportion go on to develop serious complications, reducing both the length and the quality of their lives.1 4

In continuous subcutaneous insulin infusion, a pump delivers insulin continuously under the skin through a small plastic tube and cannula.5 6 Pumps are filled with quick acting insulin to supply both background insulin and insulin replacement after meals.

Potential advantages include more precise insulin delivery and the ability to adjust basal insulin levels. Observational studies have reported improved glucose control, reduced risk of hypoglycaemia, and enhanced quality of life. Pump treatment is more expensive than multiple daily injections, with pumps costing around £2500 ($3041; €2800) each plus £1500 a year for consumables (cannulas, reservoirs, and batteries).7

In the UK, pump use is approved in adults with type 1 diabetes who have high glycated haemoglobin (HbA1c) values (>8.5%) or an inability to achieve reasonable control without “disabling hypoglycaemia.”8 An estimated 6% of UK adults with type 1 diabetes use pumps, which is lower than in many comparable countries.9 Around 40% of people with type 1 diabetes in the USA use pumps,10 and proponents of pumps suggest that far more people should be offered them in the UK.11

One weakness of existing evidence is that patients allocated to pumps are likely to have received more training and attention than those using multiple daily injections. A recent observational study12 of pump treatment and injections, where both groups received intensive education in insulin usage, concluded that the training might have made the most difference. To our knowledge, no randomised trials in adults have compared pump treatment with multiple daily injections where the same structured training in insulin adjustment has been provided, so the added benefit of pump technology remains unclear.

In the Relative Effectiveness of Pumps Over MDI and Structured Education (REPOSE) trial we assessed the effectiveness of adding pump treatment to high quality equivalent structured education in flexible intensive insulin treatment for people with type 1 diabetes, comparing pump plus education with multiple daily injections plus education. Our hypothesis was that much of the benefit of pump treatment might come from the re-training and education in insulin use given to enable patients to use pumps safely. We present the clinical effectiveness and quantitative psychosocial results of this pragmatic trial.


Trial design

The study protocol has been previously published.13 In brief, we conducted a multicentre, parallel group, open label, confirmatory, cluster randomised controlled trial. Eight secondary care centres (three in Scotland and five in England) recruited up to 40 participants to three pump and three multiple daily injection courses (with five to eight patients on each course) over 11 months. Participants were allocated a place on a one week DAFNE skills training course, with a further visit at six weeks. The course groups (clusters) were randomly allocated in pairs to either pump or multiple daily injections. After the courses, participants received the trial treatment for two years. We collected outcome measures at baseline (up to three weeks before the DAFNE course) and at six, 12, and 24 months. A cluster design was chosen to address the challenge of randomising participants and then finding suitable times for their attendance on a course of the correct allocation. We believe this approach reduced both recruitment bias and attrition rates before the course.


We recruited adults with a diagnosis of type 1 diabetes for at least 12 months, and who were willing to undertake intensive insulin treatment with self monitoring of blood glucose levels, counting of carbohydrate intake, and insulin self adjustment, with no preference for either pump or multiple daily injections. Participants were those with clinical indications for structured education in insulin treatment to optimise diabetes control who had not participated previously in structured training. We excluded those with a strong desire for pump treatment, those already using optimised multiple daily injections and meeting the criteria of the National Institute for Health and Care Excellence for pump treatment, those needing a pump in the opinion of the investigator, those with serious diabetic complications, and those unable to communicate in English. Courses (clusters) required between five and eight participants to maintain optimal course dynamics.


Participants using multiple daily injections attended standard DAFNE structured education courses,2 which were conducted over five consecutive days and delivered to groups of five to eight adults as outpatients. The participants took insulin aspart, a quick acting insulin analogue, for meals, and twice daily injections of insulin detemir for background replacement. They used the Accu-Chek Aviva Expert Bolus Advisor System (Roche Diagnostics Ltd, Burgess Hill, UK) as a bolus calculator, as there is evidence that bolus advisers can improve glycaemic control, presumably by helping patients calculate the appropriate meal related bolus.14

Participants allocated to pump treatment attended a modified DAFNE course, previously validated in pump users.15That course maintained the five day structure and the principles of insulin dose adjustment in the standard DAFNE course, but also incorporated the practical skills and learning outcomes needed to use pumps successfully. This necessitated an additional group session, delivered one to three weeks before the main course. Standard DAFNE includes a rigorous quality assurance programme. For the pump courses, fidelity testing was undertaken to assess incorporation of appropriate pump training. Participants used a Minimed Paradigm Veo insulin pump (model X54; Medtronic, Watford, UK) with insulin aspart. The bolus wizard in the pumps was activated as part of the course.

The curriculum (and associated patient workbook) was adapted specifically to make better use of pump features, compared with standard (multiple daily injections) DAFNE. This included general pump management (infusion sets, cannulas, filling reservoirs, changing batteries, and troubleshooting). During the five day course participants were also taught use of the bolus wizard, basal testing and adjustment (including fasting during the day and overnight glucose profiles), use of temporary basal rates for physical activity or alcohol intake (reduced) and illness (increased), use of alternative bolus “waves” (extended, multiwave), prevention of diabetic ketoacidosis (“sick day rules” eg, when to use a pen to inject insulin in case of cannula/pump failure, how much insulin to give and how often).

All participants (multiple daily injections and continuous subcutaneous insulin infusion) were invited to the course follow-up session at six to eight weeks, and those who required additional input were offered further one-to-one appointments. These appointments were supported by meter or pump downloads (Diasend:; CareLink:, depending on local availability. Participants were encouraged to maintain paper record diaries to facilitate discussion and adjustments, according to the principles taught on the course and supported by the workbook. Some might have already sought additional help in between these planned appointments. We recorded diabetes related contact with educators outside the course and at the six week follow-up.


Primary outcomes

We specified two primary outcomes. One was the change in HbA1c (%, measured centrally) after two years in participants whose baseline HbA1c was ≥7.5% (58 mmol/mol). HbA1c is the accepted ideal surrogate measure of glycaemic control and provides a measure of efficacy and a means of modelling long term cost effectiveness. Our choice of this primary outcome was based on our concern that HbA1c values might not decrease in those who entered the trial with low baseline values, but who might be experiencing problematic hypoglycaemia. Success for such individuals would be an HbA1c value that was maintained or even increased but with a reduced frequency of severe hypoglycaemia (an important secondary endpoint).

The other primary endpoint was the proportion of participants reaching the 2004 NICE target of HbA1c ≤7.5% (58 mmol/mol).

Secondary outcomes

Secondary biomedical outcomes measured at six, 12, and 24 months were moderate hypoglycaemia (an episode that could be treated by the individual, but where hypoglycaemia caused a significant interruption of current activity leading to impaired performance, or embarrassment, or being woken during nocturnal sleep); severe hypoglycaemia (an episode leading to cognitive impairment sufficient to cause either coma or requiring the assistance of another person to recover); total and high density lipoprotein cholesterol levels; proteinuria; insulin dose; and body weight. Diabetic ketoacidosis was recorded through the assessment of serious adverse events throughout the trial.

Ancillary outcomes

Quantitative psychosocial self completed questionnaires assessed generic and diabetes specific quality of life: SF-12 (12 item short form health survey); WHOQOL-BREF (World Health Organization quality of life–BREF); EQ-5D (EuroQol five dimensions questionnaire); DSQOL (diabetes specific quality of life), fear of hypoglycaemia (HFS: hypoglycaemia fear scale), satisfaction with treatment (DTSQ: diabetes treatment satisfaction questionnaire), and emotional wellbeing (HADS: hospital anxiety and depression scale) at the same time points.13

A health economic evaluation to address the question “What is the cost effectiveness of pump treatment compared with multiple daily injections in patients receiving the DAFNE structured education programme?” was undertaken. It has been submitted for publication. It included a within trial and a modelled patient lifetime analysis, the latter being the primary focus of the evaluation.

Sample size

We calculated the sample size using a minimally clinically important difference of 0.5% (5.5 mmol/mol) in HbA1c values. To detect this difference with a standard deviation of 1% at 80% power and 5% two sided significance required 64 participants with an HbA1c of ≥7.5% in each group. To allow for clustering of educators, an average of seven participants per group, a within course intraclass correlation coefficient of 0.05, and a 10% dropout rate, we required a sample size of 93 in each group. In the DAFNE database,15 75% of participants had an HbA1c value ≥7.5%, so we required 124 participants per group. We planned to recruit 280 participants, which increased the power to 85% but allowed for some variation in dropout rates and the proportion of participants with HbA1c of ≥7.5%. However, monitoring of baseline data showed the actual proportion of participants with an HbA1c of ≥7.5% was around 90%. A modelling exercise undertaken during recruitment with conservative estimates of 85% (HbA1c ≥7.5%) and dropout rate of 15% suggested the trial would require at least 240 participants with primary outcome data at two years to preserve a power of at least 85%.


After providing consent, participants were allocated to a training course, depending on their availability. Courses were randomised in pairs either to DAFNE plus pump or to DAFNE plus multiple daily injections. Simple randomisation in a block size of two, stratified by centre, was used for the first four courses. Courses 5 onwards were allocated in pairs using minimisation; the number of participants with baseline HbA1c values ≥7.5% and the total number of participants were used as minimisation factors. A statistician within Sheffield Clinical Trials Research Unit conducted the randomisation by a user written Stata code produced to generate allocation. The trial coordinator revealed the allocation to study sites. Participants who were unable to attend their original course were allowed to attend a later course in the same treatment arm, to reduce selection bias.

Statistical analysis

Analyses were performed in Stata 13 after a prespecified approved statistical analysis plan. All analyses were by intention to treat, with participants analysed in the groups to which they were randomised, unless otherwise specified. Participants were included in the intention to treat analysis if they had at least one post-baseline HbA1c measure. Those who dropped out before receiving the intervention were substituted where possible, to ensure the courses were run with adequate numbers of participants, but these individuals were not included in the analysis. Statistical tests were two sided at the 5% significance level.

We analysed the change in HbA1c (%) at two years using a mixed effects model, with centre and baseline HbA1c treated as fixed effect covariates, and course (cluster) as a random intercept. For the primary analysis we used multiple imputation to impute missing HbA1c data for participants with at least one follow-up HbA1c measure but without two years outcome. We also performed a per protocol sensitivity analysis that excluded participants who had switched treatment. Changes in HbA1c values at six months and one year were analysed in the same way.

The proportion of participants reaching an HbA1c of <7.5% was compared between groups using a logistic regression model adjusted for baseline HbA1c, and centre and modelling separate courses within centre as random intercepts.

We used negative binomial mixed effects regression (to account for over-dispersion and clustering) on the number of moderate hypoglycaemic episodes in the four weeks before each follow-up, and occurrence of at least one moderate episode in the four weeks before courses as a covariate and the same covariates described previously. The number of severe hypoglycaemic episodes in two years was analysed as for moderate hypoglycaemic episodes, with the addition of study follow-up as the exposure. Incidence rate ratios were calculated using negative binomial random effects regression with participant as the random effect, adjusted for baseline HbA1c value and centre, based on the full intention to treat set (n=260).

Secondary continuous outcomes (insulin dose, body weight, high density lipoprotein and total cholesterol levels) were analysed as for the primary outcome. We categorised proteinuria as macroalbuminuria, microalbuminuria, or normal and analysed using mixed effects ordered logistic regression adjusted for clustering by course (random effect), centre, and baseline HbA1c (fixed effects).

Changes in psychological outcomes were analysed using a mixed model adjusted for course (random), centre, baseline HbA1c, and baseline score, with the exception of the diabetes treatment satisfaction questionnaire, which we compared between groups using a non-parametric Wilcoxon-Mann-Whitney U test. No adjustments for multiple testing were made to the significance level for all exploratory secondary objectives.

A retrospective subgroup analysis used mixed effects regression modelling with the primary outcome, change in HbA1c (%), including main effects of treatment group and subgroup, an interaction term between treatment and subgroup, and covariates of centre (fixed effect) and course (random effect). All categories for the subgroup analysis were prespecified in the statistical analysis plan but not in the original protocol, and are reported in full.

Patient involvement

Fifteen people, who had previously attended DAFNE courses (including pilot courses on pump treatment) but were not participating in the trial, were recruited to act as a user group and contribute to different aspects of the work. We invited two members to join both the steering group and the other investigator meetings. In addition, one of the project team (a doctor) is a pump user. They provided input to the trial design, implementation, and dissemination, including all participant materials. This included a discussion of the most appropriate research questions and whether individuals who were willing to try pump treatment could be successfully recruited into a trial where they could be randomised to multiple daily injections.


Study participants

Participants were recruited between November 2011 and April 2013. Follow-up continued until June 2015. The CONSORT flowchart (fig 1) shows the flow of patients through the trial. Forty six courses were randomised. Of the 317 participants included in the randomisation, 50 were excluded from any analysis; 40 withdrew before giving baseline data and 10 before their course. All randomised courses were delivered. Of 267 participants randomised and attending the baseline assessment and the course, 260 (pump=132; injections=128) made up the intention to treat set. Two hundred and forty eight participants had complete primary outcome data at 24 months.

Fig 1 CONSORT flowchart for REPOSE cluster randomised trial to compare the effectiveness of insulin pumps with multiple daily injections (MDI). DAFNE=dose adjustment for normal eating

Table 1 shows the baseline demographics and characteristics of the trial population stratified by treatment received. Baseline data were well balanced between treatment groups, with the exception of slightly higher baseline HbA1c in the pump group (9.3% v 9.0%). Just 9% had a baseline HbA1c <7.5%.

Table 1

Baseline demographics of trial population. Values are numbers (percentages) unless stated otherwise

Primary outcomes


At 24 months in participants whose baseline HbA1c was ≥7.5% (n=119 in pump group; n=116 in multiple daily injections group) the mean change in the pump group was a decrease of 0.85% (9.3 mmol/mol) compared with 0.42% (4.5 mmol/mol) in the multiple daily injections group. After adjusting for centre, course, and baseline HbA1c, the mean difference in HbA1c change from baseline was −0.24% (95% confidence interval −0.53% to 0.05%) (−2.7 mmol/mol, −5.8 to 0.5) in favour of pump treatment (P=0.10). The treatment difference was larger for the per protocol analysis; mean difference −0.36% (−0.64% to −0.07%) (−3.9 mmol/mol, −7.0 to −0.8) in favour of pump treatment (P=0.02), although this point estimate was still smaller than the prespecified minimal clinically important difference. Estimate of the intraclass correlation coefficient was approximately 0.5%.

At 24 months, for the treatment groups combined (n=248), in all participants with complete HbA1c data there was a decrease of 0.54% (95% confidence interval 0.38% to 0.69%) (5.9 mmol/mol, 4.2 to 7.6) and for participants with baseline HbA1c ≥7.5% (n=224) the decrease was slightly greater, at 0.64% (95% confidence interval 0.48% to 0.80%) (7 mmol/mol, 5.2 to 8.8).

Proportion of participants reaching HbA1c ≤7.5% (58 mmol/mol)

The proportions of participants reaching an HbA1c ≤7.5% (58 mmol/mol) after two years, regardless of baseline HbA1c value, were 25.0% for the pump group and 23.3% for the multiple daily injections group (odds ratio 1.22, 95% confidence interval 0.62 to 2.39, P=0.57, table 2). The results were similar at six and 12 months.

Table 2

Proportion of participants with HbA1c ≤7.5% (58 mmol/mol) at six, 12, and 24 months (including all participants regardless of baseline HbA1c)

The primary analysis at 24 months was repeated for six and 12 month follow-up visits (table 3). The results for these interim time points were consistent with the primary outcome analysis. The largest difference in HbA1c change from baseline was observed at six months, with an adjusted mean difference of −0.25% (95% confidence interval −0.52% to 0.02%) (−2.7 mmol/mol, −5.6 to 0.2), P=0.07. Figure 2 displays the change in HbA1c for participants with data at all four time points by treatment group.

Table 3

Mean difference in change in HbA1c (%) at six and 12 months in participants with baseline HbA1c ≥7.5%

Fig 2 Mean change (%) in glycated haemoglobin (HbA1c) over time in participants with baseline HbA1c ≥7.5% (58 mmol/mol) (including only participants with data at all four time points, n=208). MDI=multiple daily injections

Secondary outcomes


Relatively few severe hypoglycaemic episodes were observed post-baseline: 49 in 25 participants. The rate of severe hypoglycaemia during the 24 month follow-up did not differ between the treatment groups, adjusted for centre, course, baseline HbA1c, and presence of at least one severe hypoglycaemic episode in the 12 months before baseline (incidence rate ratio 1.13, 95% confidence interval 0.51 to 2.51, P=0.77).

Across both treatment groups, the number of severe hypoglycaemic episodes was reduced. The average number of episodes for each patient per year in the study reduced from 0.17 before baseline to 0.10 during follow-up. The incidence rate ratio for the number of severe hypoglycaemic episodes in the 24 month follow-up, compared with the year before baseline, was 0.46 (0.24 to 0.89, P=0.02).

Across both treatment arms, on average, three moderate hypoglycaemic episodes were recorded for each patient over a four week history at six months. By 24 months this number was slightly lower (2.6 for pump group, 2.3 for multiple daily injections group) but there was no statistically significant difference between the groups in rates of moderate hypoglycaemic episodes at any time point.

Other biomedical outcomes

Body weight remained roughly constant throughout the study period, and was not statistically significantly different between the treatment groups at any time point (table 4). A slight increase in high density lipoprotein cholesterol and a slight decrease in total cholesterol levels were observed in both groups, with no evidence of a difference between treatment groups in change from baseline (P values ranged from 0.22 to 0.86). Insulin dose decreased in both arms. At 12 months, participants in the pump group had a 0.07 IU/kg greater reduction in insulin dose than those in the multiple daily injections group (95% confidence interval 0.01 to 0.13 decrease, P=0.02). The difference was slightly smaller at six and 24 months and was not statistically significant. There was no statistically significant difference in the odds of proteinuria between the treatment groups at any time point (table 5).

Table 4

Secondary continuous outcomes: mean difference in change from baseline at six, 12, and 24 months

Table 5

Secondary outcomes: proportion of participants in each proteinuria category (as defined by albumin to creatinine ratio) at six, 12, and 24 months. Values are numbers (percentages) unless stated otherwise

Diabetic ketoacidosis

The number or type of serious adverse events did not differ between the groups, with the exception of diabetic ketoacidosis, which was greater in the pump group compared with multiple daily injections group (17 v 5). More patients using pumps than using multiple daily injections had several episodes (5 v 2) and the differences were confined to the first year, with four episodes in each group during the second year. Three episodes occurred in two participants who switched to pump treatment, and one in a participant who switched to multiple daily injections. Most episodes of ketoacidosis were caused by infections and 18% by set failure in those using pumps. Only five episodes occurred when participants implemented all sick day rules.

Ancillary outcomes

High levels of completion of the psychosocial questionnaires were observed across all questionnaires and time points (around 90% completed at each time point). The completion rate was slightly higher for participants allocated to the pump compared with multiple daily injections, which reflects the relative dropout rates in the two groups. No between group differences were found in the generic quality of life and health status instruments SF-12, WHOQOL-BREF, and EQ-5D, and the HADs score for depression and anxiety at six, 12, and 24 months (tables 6-8).

Table 6

Mean difference in change of psychosocial outcomes from baseline to six months

Table 7

Mean difference in change in quantitative psychosocial outcomes (from baseline to 12 months

Table 8

Mean difference in quantitative psychosocial outcomes from baseline to 24 months

The overall diabetes specific quality of life (DSQOL) (on a 100 point scale) showed that both groups improved at 24 months, by a mean of 8.2 (SD 13.1) points in the pump group and 4.2 (SD 13.2) points in the multiple daily injections group. Both groups showed improvements in the DSQOL subdomains, which were greater in the pump group although not always reaching statistical significance.

The improvement in DSQOL diet restrictions was larger for the pump group compared with multiple daily injections group at both 12 and 24 months (12 month adjusted mean difference in change from baseline −4.1 (95% confidence interval −7.2 to −1.0, P=0.01); 24 month adjusted mean difference in change from baseline −5.1 (−8.6 to −1.6, P=0.004); lower scores represent better outcomes. The pump group also had greater improvement in DSQOL daily hassle or functions at both 12 and 24 months; at 24 months the score had decreased by 10 points in the pump group, compared with 4 points in the multiple daily injections group (adjusted mean difference −6.3, −10.9 to −1.8, P=0.01).

Participants in the pump group had better improvement in treatment satisfaction at all time points (table 9) but the difference was statistically significant at 12 and 24 months only (P=0.07 at six months, P<0.001 at 12 and 24 months).

Table 9

Ancillary outcomes: diabetes treatment satisfaction questionnaire change from baseline at six and 24 months, questionnaire raw scores at 12 months

Retrospective analyses

Participants achieving the updated NICE recommendation of HbA1c ≤6.5%

A retrospective analysis showed that eight (3%) of all participants (4/128 pump group and 4/120 multiple daily injections group) reached an HbA1c of ≤6.5% (47 mmol/mol) after two years (table 10). Of these, two participants (both in the pump group) experienced one or more episodes of severe hypoglycaemia during follow-up.

Table 10

Number and proportion of participants achieving new National Institute for Health and Care Excellence HbA1c recommendation of ≤6.5% at six, 12, and 24 months

Participant contacts

A retrospective analysis showed that, on average, those allocated to pumps had around double the number of contacts with professionals for diabetes between baseline and the end of year 1, both face to face and by telephone. Between months 12 and 24, those using pumps had more face-to-face contacts, which were of longer duration (mean 1.6 v 1.3 contacts), but fewer telephone contacts (0.7 v 1.2).

Blood glucose testing

A retrospective analysis indicated that there was no difference in the mean frequency of blood glucose testing between treatment groups at 24 months after adjustment for baseline number of blood glucose tests, centre, and DAFNE course. The adjusted mean difference in blood glucose tests was 0.22 (95% confidence interval −0.24 to 0.68) per day or 3.1 (−3.4 to 9.6) over two weeks; P=0.35. Overall, the number of blood glucose tests increased from 3.6 per day at baseline to 4.1 per day at 24 months (95% confidence interval 0.33 to 0.82, P<0.001).

Subgroup analysis

A retrospective analysis found no reliable statistical evidence of any subgroup effects or interactions between group (figs 3 and 4). However, there was some indication that participants with qualifications up to A level or equivalent did better in the pump group than in the multiple daily injections group (mean difference in HbA1c change (%) at 24 months −0.7% (95% confidence interval −1.2% to −0.1%) (−7.4 mmol/mol, 95% confidence interval −13.2 to −1.5)), although the interaction test was not statistically significant (P=0.07).

Fig 3 Mean difference in HbA1c change (%) at 24 months for pump versus multiple daily injections by subgroup. MCID=minimal clinically important difference; IMD=index of multiple deprivation; ONS=Office for National Statistics occupational status from level 1, elementary trade, service, administration roles, to level 4, corporate managers or directors, research, teaching, business, and public service higher level professionals

Fig 4 Mean difference in HbA1c change (%) at 24 months by subgroup


In a group of adults with type 1 diabetes referred for structured training in flexible insulin treatment because of suboptimal diabetes control, participation in the REPOSE trial achieved a clinically worthwhile decrease in HbA1c of 0.6% (7 mmol/mol) at two years in those with a baseline HbA1c of >7.5%. However, in terms of the primary outcomes, there were no statistically significant differences in change from baseline to 24 months between those randomised to pump treatment or those using multiple daily injections, nor in the proportion of participants reaching an HbA1c of ≤7.5%, indicating that pump treatment provided no clear biomedical benefit over training in DAFNE skills.

Rates of severe hypoglycaemia were halved in both groups (despite lower HbA1c values), a benefit maintained to 24 months with no difference between the groups in this or in rates of moderate hypoglycaemia. There were no other differences in biomedical outcomes apart from slightly greater reductions in insulin doses in those randomised to pump treatment. Both groups showed improved satisfaction with treatment and diabetes specific quality of life. Treatment satisfaction and two subdomains of the diabetes specific quality of life scale improved to a greater extent at two years in those allocated to pump treatment.

Strengths and limitations of this study

Compared with previous trials of pump treatment our study had a robust, multisite design, involved much larger numbers, and had a clinically meaningful period of follow-up pump treatment. Participants in both groups used analogue insulins and bolus calculators. The study was conducted in experienced secondary care centres and involved attendance at a structured education intervention that is well established across the UK. It included a comprehensive psychological evaluation with high levels of data completeness. The pragmatic study design thus provides good external validity, particularly as participating in the educational course led to sustained improvements in glycaemic control and reduced rates of severe hypoglycaemia.

It is not possible to blind a trial where insulin delivery systems are fundamentally different and this imposes limitations on any randomised controlled trial involving pumps. However, for the primary outcomes, HbA1c was objectively assessed using a central laboratory.17 A trial studying people who have expressed a desire for pump treatment is likely to struggle to recruit participants if one arm continues to use multiple daily injections. Those randomised to the injections arm might also either drop out or exhibit poor outcomes because of “disappointment” and lack of motivation. We recruited individuals who had not specifically requested pump treatment but were awaiting a course in diabetes self management to help them improve their metabolic control. Thus, our aim was to determine any added benefit of pumps over multiple daily injections while controlling for the training itself.

A potential limitation is that those randomised to pump treatment might have been insufficiently motivated to make the most of any technological benefit since they had not expressed a particular wish to use a pump. A common reason given by patients for not wanting to participate in REPOSE was reluctance to use an insulin pump. However, educators encouraged participants to use additional technological pump features (eg using more sophisticated bolus delivery) and provided extra input when this training was requested. Overall, we reasoned that since participants had signed up for a course to improve their glucose control, any added benefits of pump treatment would emerge.

In this pragmatic trial we did not collect detailed information about pump basal rates, how often patients adjusted and tested these, and time spent with pump participants. Our study was not designed to establish which features of pumps determine success in lowering HbA1c values in those doing “well” with pump treatment. An explanatory trial would have required a different study design. We therefore cannot be sure why those allocated to pump treatment failed to show a greater decrease in HbA1c.

We recorded the average number of blood glucose tests each day in both groups over the previous two weeks, both at baseline and at 24 months. Both groups had increased the number of daily tests from 3.6 before training to more than four each day at 24 months, but this did not differ between the groups. Perhaps this frequency of testing reflects a level of engagement in self management that was insufficient for participants in the pump group to take full advantage of the technology. One interpretation was that the educators were unable to provide adequate training in use of the pump during the one week course. However, it is just as likely that provision of pump treatment in a group of patients who had not been previously trained in delivery of flexible intensive insulin treatment included many who subsequently found it too challenging to implement and maintain the intensity of self management that both pump treatment and multiple daily injections demand.

Comparison with other trials

Two appraisals of pumps by NICE have reviewed the evidence on clinical and cost effectiveness. One18 noted that there were no trials of pumps against “best multiple daily injections” with long acting and short acting analogue insulins; that some trials had unequal amounts of education in the arms (with more in the pump arms); and that the trials had focused on easily measurable outcomes such as HbA1c, rather than on benefits in terms of flexibility of lifestyle and quality of life. The report recommended trials of pumps against analogue based multiple daily injections. A more recent report19 found only three trials in adults, one a pilot and the other involving 39 adults with type 1 diabetes, already using pump treatment who were randomised to continue with the pump or to switch to glargine based multiple daily injections. Patients received four months treatment with each. A third trial recruited 57 adults randomised to pump or analogue injections in an equivalence study. None showed any difference in HbA1c. Thus, the evidence base from trials for comparing pumps and “best multiple daily injections” was weak in terms of numbers, with a total of only 103 patients and short follow-up.

The assessment for the second appraisal19 reviewed observational studies of adults switching to pumps for clinical indications. These have the advantage of measuring change in glycaemic control and hypoglycaemia in those who have most to gain, and these studies showed improved HbA1c of the order of around 0.5%. Bias in observational studies is more of a problem, and results must be treated with caution. Furthermore, of 48 observational studies, only nine reported quality of life. Study numbers were small, with at most 35 patients. Duration was usually short. The longest study noted that initial benefits from pump treatment might not be sustained. The present study has thus addressed several of these concerns with large numbers in an adequately powered trial and a virtually complete dataset for both biomedical and psychological outcomes.

Clinical and policy implications

Our study suggests that extending the availability of pumps to adults with type 1diabetes with suboptimal glycaemic control and no desire to use this form of insulin delivery is unlikely to result in either lower levels of glycaemia as measured by HbA1c or lower rates of hypoglycaemia. The absence of a control arm in which structured training was not provided, means we cannot be certain that participation in training explains the considerable decreases in HbA1c and severe hypoglycaemia in both groups. Yet, it seems unlikely that mere trial participation led to sustained decreases in HbA1c and severe hypoglycaemia for up to two years, particularly as allocation to such a control arm, in previous trials involving the DAFNE intervention, showed no effect on HbA1c.2

The results would appear to support the current clinical pathway as proposed by NICE, in which people desiring improved diabetes control undergo structured training in flexible insulin treatment with multiple daily injections alone. The trial outcomes, together with the review for NICE of observational pump studies, suggest that pumps might usefully be reserved to support those who actively engage in self management after structured education. Those who find that despite their best efforts, injections fail to deliver the expected benefits could then be offered the additional technological advantages of an insulin pump.

Clearly some patients improved more than others in terms of glucose control or hypoglycaemia and we explored whether there were any demographic differences in those who did particularly well. There was no reliable evidence of any plausible subgroup effects or interactions between the pump and multiple daily injections group, and the baseline characteristics of those whose glycaemic control changed to <7.5% during the trial were no different from the pump population as a whole. We observed modest centre effects but no systematic differences according to greater experience in pump treatment.

Those using insulin pumps did show some quality of life benefits, reporting less restriction in diet and daily hassles in the DSQOL scale and greater treatment satisfaction. Nevertheless, the differences were modest and observed in comparison with a group given no novel technology. Since they were not associated with other positive treatment outcomes, those observations are probably insufficient to justify a major alteration in guidelines for the use of pumps.

One of the more striking results of this trial was the generally high level of HbA1c among adults in the UK enrolling for self management training in flexible insulin treatment. Participation in the courses produced important and sustained decreases, but for most participants still fell well short of the target recommended by NICE, recently reduced from 7.5% to 6.5%.20 The high levels of HbA1c among people with type 1 diabetes in UK centres compared with most other European countries have also been noted in a recent study.21 There is an urgent need to explore the barriers to successful self management in adults with type 1 diabetes in the UK and to understand why referral for appropriate training is often left so long. This was also the conclusion of our recently completed research programme.15 Our results suggest that these problems cannot be overcome merely by providing additional technology in the form of pumps.


People with type 1 diabetes might be better served by ensuring far greater availability of high quality, structured self management training, which is currently only accessed by around 10% of adults in the UK.22 Participants might only recognise the limitations of insulin delivery by multiple daily injections if they start actively managing their diabetes after training. Those individuals could then be offered pump treatment to help them reach the stringent glucose targets necessary to achieve an HbA1c of 6.5% or to overcome problematic hypoglycaemia.

What is already known on this topic

  • Assessments of insulin pump treatment in type 1 diabetes concluded that it was worthwhile for those who were otherwise unable to achieve good glycaemic control without disabling hypoglycaemia

  • The case for wider use is uncertain, given the small size and short duration of trials of pumps versus modern multiple daily injections, and the need to distinguish the effects of the pump and the extra education provided

What this study adds

  • The REPOSE trial randomised patients to pump or multiple daily injections, with both groups receiving similar structured education; HbA1c levels and rates of severe hypoglycaemia decreased in both groups, slightly more in the pump group, but without statistical significance

  • Both groups showed improved quality of life benefits, but those using pumps showed additional albeit modest benefits in quality of life, reported fewer restrictions in diet and daily hassles in the diabetes specific quality of life scale, and greater treatment satisfaction

  • Although participation in the courses produced sustained improvement, levels of glucose control remained far short of those currently recommended by NICE


This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See:


Source: NEJM/BMJ

Like this:

%d bloggers like this: