Abstract
Background:
Real-world data regarding indications for use of insulin pump remain sparse. We investigated characteristics among individuals with type 1 diabetes (T1D) in relation to indication for use of insulin pump (CSII). Comparison was made with T1D subjects using multiple daily injections (MDI).
Methods:
We included all individuals with T1D who had at least 1 registration in the National Diabetes Register during 2014-2015. Among 46 874 individuals, we excluded 2350 due to missing data. We examined 35 725 on MDI and 8799 on CSII regarding characteristics in relation to insulin delivery method, as well as association between insulin delivery and glycemic control (HbA1c) and presence of albuminuria.
Results:
Unadjusted mean (SD) HbA1c was 63.84 (15.07) mmol/mol (7.99 [1.38]%) and 63.75 (13.19) mmol/mol (7.99 [1.21]%) in the MDI and CSII group, respectively. MDI and CSII users were on average 48.8 and 41.5 years old, respectively. MDI users were on average 26 years old and CSII users 17 years old at the time of diabetes diagnosis. Overall, a higher proportion of CSII users were females (53.5%). As compared with MDI, use of CSII was associated with up to 7.84 mmol/mol (0.72%) lower HbA1c in a multivariable adjusted model. Use of CSII was, however, not associated with risk of having albuminuria.
Conclusions:
CSII was used more frequently in younger individuals, early-onset diabetes, and problematic glycemic control. The use of CSII was associated with lower HbA1c among CSII users except from those who started CSII due to high HbA1c.
Keywords: continuous insulin pump, type 1 diabetes, pump indication, HbA1c
Type 1 diabetes (T1D) requires daily administration of insulin with 4-5 multiple daily injections via pens or continuous subcutaneous insulin infusion (CSII) via an insulin pump. The aim of the treatment is to normalize glucose metabolism. Patients with T1D have a 3- to 5-fold increased risk of long-term micro- and macrovascular complications, comorbidities, and premature death,1 although the risk is reduced if diabetes is well controlled.1,2
The incidence of T1D has increased by 3% annually since the 1980s,3,4 and in 2015 the Swedish National Diabetes Register reported that there were 41 061 people with T1D.5 Despite a high T1D diabetes burden, Sweden has a high standard of care for the type 1 population.6 The Swedish National Diabetes Register (NDR) offers an excellent opportunity to study individuals with T1D; 95% of all individuals with T1D have been entered in the register, which includes detailed clinical data collected from each appointment at the local health care provider. In 2013, 1 in 4 women and 1 in 5 men used CSII. Starting in 2014, the NDR started collecting detailed data regarding CSII.
We have previously reported that CSII is associated with significantly lower risk of fatal coronary heart disease, fatal cardiovascular disease, and all cause mortality, as well as nonsignificant reduction in the risk of nonfatal or fatal cardiovascular disease.7
The use of CSII varies markedly between different countries.8,9 There are few reports on the different indications for CSII in various patient groups.9 We aimed to describe a nationwide sample of CSII users regarding indications and patient characteristics, as well as the association between indication for CSII and glycemic control.
Methods
The Central Ethical Review Board in Gothenburg, Sweden approved the study.
The NDR was launched in 1996 as a caregiver tool for quality assurance and as a feedback tool. Physicians and nurses, trained in register procedures, provide data obtained at visits in outpatient clinics of hospitals and primary care clinics. Clinical data regarding risk factors, complications of diabetes, and treatment and management are recorded. Informed consent for inclusion in the register was obtained from each patients (verbal or written). We assessed the clinician’s classification of diabetes to include persons with T1D aged 18 year and older.
We included all individuals with T1D with at least 1 registration in the NDR from June 1, 2014, to November 16, 2015. Use of CSII, as a dichotomous variable, has been registered since 2004. However, details regarding CSII treatment were not registered before June 2014. Starting in June 2014 information on main indication for use of CSII, manufacturer of CSII, issues with ketoacidosis, hypoglycemic events, skin reactions, infections, and reasons for terminating use of CSII were collected. These data have been reported for 70% the population using CSII.
Among 46 874 individuals, we excluded 2350 with missing data on method of insulin delivery. Of the remainder, 35 725 used MDI and 8799 used CSII. We present descriptive data, regarding risk factors and clinical characteristics according to (1) method of insulin delivery (MDI vs CSII), (2) indication for use of CSII, and (3) indication for use of CSII stratified by gender.
Physical activity (PA) is in the NDR database is self-reported. In the NDR there are 5 levels of PA, based on number of exercise sessions per week of at least a 30-minute duration: no PA (level 1), PA less than once a week (level 2), PA once or twice a week (level 3), PA 3-5 times per week (level 4), or PA daily (level 5).
Statistical Analysis
To increase power we imputed data by means of first observation carried backward for variables related to insulin delivery. First observation carried backward mean that the first available value is imputed backward in time, to decrease missing data on index observation (the first observation during the inclusion period).
We assessed adjusted differences in HbA1c, to compare whether there were any differences in relation to method of insulin delivery and various indications for CSII. This was done using linear regression, with HbA1c as response variable and age, gender, diabetes duration and insulin delivery as predictors. The latter predictor had 9 levels, 1 of which was MDI and the remaining 8 were the various indications for CSII. The MDI category served as reference group.
We also assessed the probability of having albuminuria (micro or macro), to compare differences in relation to method of insulin delivery and various indications for CSII. This was done using logistic regression, with albuminuria as dependent variable and the following independent variables: insulin delivery method, age, sex, HbA1c, duration of diabetes, and systolic blood pressure. The MDI category served as the reference group.
Due to the observational and exploratory nature of the study, P values are not intended for statistical inference; neither are P values corrected for multiple testing. Judgment of group differences should be made based on actual values, particularly adjusted estimates, which are provided for the main analysis (difference in HbA1c and albuminuria in relation to indication for CSII).10 P values for continuous variables are derived using student’s t-test (MDI vs CSII) and ANOVA (various CSII indication groups). P values for categorical variables are derived by means of chi square tests.
Results
Crude mean (SD) HbA1c was 63.84 (15.07) mmol/mol (7.99 [1.38]%) and 63.75 (13.19) mmol/mol (7.99 [1.21]%) in the MDI and CSII group, respectively. MDI users were on average 48.8 years old, as compared with CSII users who were 41.5 years old. MDI users were on average 26 years old at the time of diabetes diagnosis, whereas CSII users were on average 17 years old at onset of diabetes. Overall, a higher proportion of CSII users were females (53.5%), but within the CSII group, only one-third of those with the indication “physical activity” were women. Indications for CSII were glucose excursions 35.8%, elevated HbA1c 33.0%, patient preference 11.5%, frequent hypoglycemic events 7.9%, facilitation of glycemic control 6.4%, PA 2.1%, dawn phenomenon 1.5%, unawareness 1.8%. There were no differences in diastolic blood pressure, but MDI users were more frequently treated with antihypertensives and had higher systolic BP. Although albuminuria was slightly more common among MDI users, there were no differences in eGFR. Refer to Table 1 for details.
Table 1.
Characteristics of All Patients With Type 1 Diabetes in the Swedish National Diabetes Register According to Method of Insulin Delivery.
| Overall | MDI | CSII | P value | |
|---|---|---|---|---|
| n | 44 524 | 35 725 | 8799 | |
| Age (years), mean (SD) | 47.36 (17.43) | 48.80 (17.60) | 41.51 (15.39) | <.001 |
| Age at diagnosis (years), mean (SD) | 23.95 (16.27) | 25.69 (16.69) | 16.97 (12.15) | <.001 |
| Diabetes duration (years), mean (SD) | 23.41 (15.84) | 23.11 (16.39) | 24.60 (13.35) | <.001 |
| Females, n (%) | 19 761 (44.4) | 15 050 (42.1) | 4711 (53.5) | <.001 |
| HbA1c (IFCC, mmol/mol) | 63.82 (14.71) | 63.84 (15.07) | 63.75 (13.19) | .631 |
| HbA1c (NGSP, %) | 7.99 (1.35) | 7.99 (1.38) | 7.99 (1.21) | .631 |
| Systolic BP | 127.32 (15.05) | 127.93 (15.26) | 124.74 (13.86) | <.001 |
| Diastolic BP | 73.49 (9.33) | 73.52 (9.46) | 73.36 (8.74) | .197 |
| Antihypertensives | 18 107 (47.0) | 15 211 (48.8) | 2896 (39.3) | <.001 |
| Total cholesterol | 4.64 (0.96) | 4.65 (0.97) | 4.61 (0.91) | .005 |
| Statins, n (%) | 18 317 (47.2) | 15 259 (48.8) | 3058 (40.7) | <.001 |
| eGFR (ml/min) | 91.96 (27.86) | 91.87 (28.57) | 92.33 (24.76) | .247 |
| Albuminuria, n (%) | <.001 | |||
| No albuminuria | 26 161 (82.5) | 20 766 (81.9) | 5395 (84.8) | |
| Microalbuminuria | 3808 (12.0) | 3156 (12.4) | 652 (10.3) | |
| Macroalbuminuria | 1743 (5.5) | 1431 (5.6) | 312 (4.9) | |
| Physical activity, n (%) | <.001 | |||
| No physical activity | 2601 (7.6) | 2201 (8.0) | 400 (5.9) | |
| <1 time per week | 4786 (14.0) | 3847 (14.0) | 939 (13.8) | |
| 1-3 times per week | 7836 (22.9) | 6160 (22.5) | 1676 (24.6) | |
| >3 times per week | 10 544 (30.8) | 8338 (30.4) | 2206 (32.4) | |
| Daily | 8479 (24.8) | 6892 (25.1) | 1587 (23.3) | |
Table 2 presents characteristics in relation to indication for CSII. There were substantial differences with respect to age; patients with frequent hypoglycemia as the indication were 49.7 years old on average, as compared with those with facilitation of glycemic control as the indication who were 22.4 years old on average. In terms of sex, the only distinguishing finding was that only one-third of those with the indication “physical activity” where women. PA varied substantially; about half of patients with “physical activity” as the indication exercised on a daily basis whereas only 18.7% of patients with facilitation of glycemic control as indication where exercising daily.
Table 2.
Characteristics of the Study Population, According to Indication for Insulin Pump Treatment (CSII).
| Glucose excursions | High HbA1c | Frequent hypoglycemias | Physical activity | Dawn phenomenon | Unawareness | Patients preference | Facilitation of glycemic control | P | |
|---|---|---|---|---|---|---|---|---|---|
| n | 1569 | 1445 | 348 | 94 | 64 | 78 | 506 | 281 | |
| Age (years), mean (SD) | 45.94 (14.67) | 40.15 (13.61) | 49.66 (14.78) | 39.93 (13.01) | 43.19 (12.00) | 53.21 (13.00) | 40.29 (15.00) | 22.35 (4.61) | <.001 |
| Age at diagnosis (years), mean (SD) | 19.27 (13.43) | 16.19 (11.08) | 19.81 (13.86) | 18.99 (12.04) | 18.80 (12.51) | 17.53 (12.03) | 18.90 (12.92) | 8.90 (4.98) | <.001 |
| Diabetes duration (years), mean (SD) | 26.72 (13.67) | 23.99 (11.78) | 29.95 (14.93) | 20.51 (12.39) | 24.39 (11.99) | 35.71 (13.85) | 21.43 (13.72) | 13.48 (5.83) | <.001 |
| Females, n (%) | 922 (58.8) | 779 (53.9) | 176 (50.6) | 30 (31.9) | 39 (60.9) | 37 (47.4) | 261 (51.6) | 130 (46.3) | <.001 |
| HbA1c (IFCC, mmol/mol) | 60.81 (10.79) | 69.18 (13.26) | 58.15 (11.77) | 56.30 (9.12) | 60.72 (9.13) | 59.08 (11.94) | 59.88 (11.49) | 64.36 (14.60) | <.001 |
| HbA1c (NGSP, %) | 7.72 (0.99) | 8.48 (1.21) | 7.47 (1.08) | 7.30 (0.83) | 7.71 (0.84) | 7.56 (1.09) | 7.63 (1.05) | 8.04 (1.34) | <.001 |
| Systolic BP (mmHg) | 124.96 (13.99) | 124.53 (14.18) | 125.06 (14.73) | 123.96 (13.66) | 123.58 (12.17) | 127.12 (13.34) | 123.69 (13.05) | 119.95 (11.57) | <.001 |
| Diastolic BP (mmHg) | 73.15 (8.93) | 74.04 (8.87) | 72.08 (9.00) | 74.53 (8.65) | 73.77 (8.10) | 71.83 (8.81) | 73.66 (8.32) | 72.15 (8.07) | .002 |
| Antihypertensives | 605 (47.0) | 527 (44.0) | 151 (52.6) | 22 (29.7) | 27 (51.9) | 43 (58.1) | 135 (35.6) | 9 (4.5) | <.001 |
| Total cholesterol | 4.65 (0.90) | 4.64 (0.91) | 4.65 (0.97) | 4.60 (0.84) | 4.74 (1.22) | 4.63 (0.85) | 4.65 (0.87) | 4.29 (.80) | .003 |
| Statins, n (%) | 673 (50.2) | 561 (46.2) | 159 (55.0) | 26 (34.2) | 27 (49.1) | 45 (60.8) | 143 (37.0) | 8 (3.8) | <.001 |
| eGFR (ml/min) | 88.19 (22.90) | 94.07 (25.46) | 83.92 (22.80) | 92.11 (15.17) | 87.47 (17.99) | 83.69 (22.44) | 92.01 (22.23) | 11.14 (23.11) | <.001 |
| Albuminuria, n (%) | <.001 | ||||||||
| No albuminuria | 931 (85.5) | 777 (77.7) | 187 (82.0) | 58 (93.5) | 33 (80.5) | 46 (79.3) | 282 (87.0) | 167 (93.3) | |
| Microalbuminuria | 113 (10.4) | 144 (14.4) | 30 (13.2) | 3 (4.8) | 5 (12.2) | 7 (12.1) | 30 (9.3) | 9 (5.0) | |
| Macroalbuminuria | 45 (4.1) | 79 (7.9) | 11 (4.8) | 1 (1.6) | 3 (7.3) | 5 (8.6) | 12 (3.7) | 3 (1.7) | |
| Physical activity, n (%) | .013 | ||||||||
| No physical activity | 63 (4.9) | 67 (5.9) | 11 (4.0) | 1 (1.2) | 1 (1.9) | 5 (7.8) | 18 (4.5) | 11 (5.0) | |
| <1 time per week | 155 (12.1) | 155 (13.6) | 28 (10.1) | 3 (3.8) | 4 (7.5) | 7 (10.9) | 47 (11.8) | 32 (14.6) | |
| 1-3 times per week | 314 (24.5) | 284 (24.8) | 65 (23.4) | 12 (15.0) | 13 (24.5) | 13 (20.3) | 100 (25.1) | 56 (25.6) | |
| >3 times per week | 430 (33.6) | 358 (31.3) | 87 (31.3) | 28 (35.0) | 19 (35.8) | 21 (32.8) | 147 (36.8) | 79 (36.1) | |
| Daily | 318 (24.8) | 279 (24.4) | 87 (31.3) | 36 (45.0) | 16 (30.2) | 18 (28.1) | 87 (21.8) | 41 (18.7) | |
Figure 1 shows the association between method of insulin delivery and HbA1c. MDI users served as the reference category, with whom each CSII indication group was compared. It was evident that only patients with high HbA1c as the indication for CSII had a higher HbA1c than MDI users; remaining CSII indication groups had lower HbA1c than MDI users. Patients with PA as indication for CSII had 7.84 mmol/mol (0.72%) lower HbA1c than MDI users. Those with glucose excursions, frequent hypoglycemia, unawareness and patient’s preference as CSII indication also had a significantly lower HbA1c, compared with MDI users.
Figure 1.
Association between indication for CSII and HbA1c. Adjusted differences in HbA1c, according to insulin delivery MDI or CSII. Predictors used in the model: insulin delivery (MDI or respective CSII indication group), age, sex, diabetes duration. HbA1c was the response variable. Estimate shows adjusted differences in HbA1c (mmol/mol).
Figure 2 shows the probability of having albuminuria (micro or macro) in relation to indication for CSII, using the MDI group as reference. As evident from the figure, there were almost no differences in the risk of having albuminuria, with the exception of those participants with high HbA1c as indication for CSII had an odds ratio of 1.43 (P < .001), as compared with MDI users. However, when excluding HbA1c as a covariate in the model, CSII users with glucose excursions or PA as indication had statistically significant lower risk of having albuminuria (Supplementary Figure 1).
Figure 2.
Association between indication for CSII and probability of having albuminuria. Adjusted odds ratios for having albuminuria (micro or macro), according to insulin delivery. Predictors used in the model: method of insulin delivery (MDI or respective CSII indication group), age, HbA1c, systolic BP, sex, diabetes duration. Albuminuria was the response variable. Estimate shows odds ratios for having albuminuria in each CSII indication group, as compared with MDI users.
Discussion
This nationwide Swedish study of 46 874 individuals with T1D presents descriptive data regarding clinical characteristics among Swedish T1D patients in relation to method of insulin delivery. We compared MDI users with CSII users and delineated differences in relation to indication for CSII. This study documents that almost about 9 out of 10 CSII users have started the treatment due to various problems with glycemic control. The remainder had started CSII due to PA or personal preference. In the adjusted analysis, all CSII users displayed markedly lower HbA1c than MDI users, with the exception of those with high HbA1c as the indication for CSII. However, we show that use of CSII was not associated with the risk of having albuminuria.
Given that CSII has been associated with lower risk of adverse outcomes and mortality,6,10 and there are health-economic benefits of CSII,9 one could argue that there is room for treating more patients, provided that it suits their preferences. Indeed, this study shows that the vast majority of CSII users have started treatment due to problems with glycemic control and patients who are older (and consequently at higher risk of cardiovascular and adverse outcomes) are less frequently treated with CSII. Clinicians and policy makers should note the sum of the evidence and offer CSII to a broader range of patients.
We cannot rule out that the marked difference in HbA1c between MDI and CSII due to PA might be a case of confounding by indication. However, we argue that the overall pattern—that is, CSII is associated with a lower HbA1c among all indications except those who had high HbA1c as their indication—underscores that CSII users are at greater chance of reducing their HbA1c. Although we did not assess the longitudinal progress of HbA1c, given our cross-sectional design, other studies have shown that patients lower their HbA1c on average by approximately 5 mmol/mol, when switching form MDI to CSII.11
However, CSII does not appear to be associated with the risk of having albuminuria. We believe there could be rather straightforward explanations for this: (1) CSII users have presumably had (on average) higher HbA1c in the preceding years (indeed 9 out of 10 had problems with glycemia as their indication for CSII); (2) the cross-sectional design of the study could have underestimated the long-term beneficial effects of CSII on albuminuria. In the future we will address the effect of CSII on risk of albuminuria in a propensity score based study to test this hypothesis.
Whether the pump itself brings about the observed benefits of CSII, or via the management that accompanies the pump, is difficult to determine. Intensified management through educations and additional visits to their clinic may privilege CSII users. The ensuing increase in motivation and knowledge could also explain the favorable trajectories in glycemic control. Deciphering which components of the CSII (ie, the steady infusion of insulin, associated use of continuous glucose monitoring devices, education, increased motivation, etc) that brings about the beneficial effects,7,11 is a difficult task and one could—at least meanwhile—be satisfied knowing that CSII is beneficial. As noted by Redberg, disentangling causal relationships for medical devices is not a trivial task.12
There are numerous long-term studies that have consistently indicated that the use of ambulatory insulin infusion pumps systems is associated with improved overall glycemic control as evidenced by lower HbA1c values without any increased risk of hypoglycemia or hyperglycemia.7,13,14 The landmark Diabetes Control and Complications Trial demonstrated that lower HbA1c values were associated with significant delays in nonfatal myocardial infarction, stroke or death from cardiovascular disease.15 Clinical trials have shown that HbA1c levels are on average 5.5 mmol/mol (1.5%, in absolute figures) lower in individuals receiving insulin pump therapy vs injection therapy.16
Currently there are no strict guidelines for switching from multiple daily injections to insulin pump therapy in Sweden and regional variations occur. Among indications for a physician to recommend that an individual with T1D to switch treatment regimen from MDI to CSII are not meeting the individual glycemic goal regarding HbA1c, large glucose excursions, frequent hypoglycemic events, dawn phenomenon, or the need to improve quality of life. In Sweden both treatment by insulin pump and/or multiple daily injections are covered by the health care system without any additional costs for the patient. The indication for insulin pump is broad with no out-of-pocket expenses for the patient. Therefore there should be no economical hindrance for the patient to be treated with insulin pump.
CSII has been shown in observational studies to reduce mortality, to what extent should be consider this when choosing patients to be started on insulin pump therapy?16
Study limitations deserve mentioning. Despite the relatively large study cohort the data is of observational and cross-sectional nature. The level of PA is self-reported with the risk for recall bias. There were no significant differences in the level of self-reported PA between the MDI and the CSII group despite PA as an indication for CSII. Regarding the indication for CSII there are limitations regarding the precision of the definitions of indicators, possible over-lap and the fact that one individual may have multiple indications, how CSII treatment is chosen by the providers. Furthermore, we used first observation carried backward to impute data and such means may introduce uncertainty about the results that should be recognized when interpreting the data.
Conclusion
This study shows that in subjects with T1D CSII was used more predominantly younger individuals, individuals with early-onset diabetes, and those with problematic glycemic control. Despite the fact that about 9 out of 10 CSII users started CSII due to problematic glycemic control, use of CSII was associated with lower HbA1c among CSII users except from those who started CSII due to high HbA1c.
Supplementary Material
Acknowledgments
We thank all regional Swedish National Diabetes Register coordinators, contributing nurses, physicians, and patients who contributed to collection of the data. We also thank the Swedish Society of Diabetology and the Swedish Diabetes Association for supporting the Swedish National Diabetes Register.
Footnotes
Abbreviations: CSII, continuous subcutaneous insulin infusion; MDI, multiple daily insulin injections; NDR, National Diabetes Register; PA, physical activity; T1D, type 1 diabetes.
Declaration of Conflicting Interests: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study received funding from the European Association for the Study of Diabetes. The Swedish National Diabetes Register is funded by the Swedish Association of Local Authorities and Regions (SKL).
Supplementary Material: The supplementary material is available at http://dst.sagepub.com/supplemental
References
- 1. Nathan DM, Cleary PA, Backlund J-YC, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med. 2005;353(25):2643-2653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Lind M, Svensson A-M, Kosiborod M, et al. Glycemic control and excess mortality in type 1 diabetes. N Engl J Med. 2014;371(21):1972-1982. [DOI] [PubMed] [Google Scholar]
- 3. Patterson CC, Dahlquist GG, Gyürüs E, Green A, Soltész G, EURODIAB Study Group. Incidence trends for childhood type 1 diabetes in Europe during 1989-2003 and predicted new cases 2005-20: a multicentre prospective registration study. Lancet. 2009;373(9680):2027-2033. [DOI] [PubMed] [Google Scholar]
- 4. Rawshani A, Landin-Olsson M, Svensson A-M, et al. The incidence of diabetes among 0-34 year olds in Sweden: new data and better methods. Diabetologia. 2014;57(7):1375-1381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Gudbjörnsdottir S. Annual report—The Swedish National Diabetes Register. 2014. [Google Scholar]
- 6. McKnight JA, Wild SH, Lamb MJE, et al. Glycaemic control of type 1 diabetes in clinical practice early in the 21st century: an international comparison. Diabet Med. 2015;32(8):1036-1050. [DOI] [PubMed] [Google Scholar]
- 7. Steineck I, Cederholm J, Eliasson B, et al. Insulin pump therapy, multiple daily injections, and cardiovascular mortality in 18,168 people with type 1 diabetes: observational study. BMJ. 2015;350:h3234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Sherr JL, Hermann JM, Campbell F, et al. Use of insulin pump therapy in children and adolescents with type 1 diabetes and its impact on metabolic control: comparison of results from three large, transatlantic paediatric registries. Diabetologia. 2016;59(1):87-91. [DOI] [PubMed] [Google Scholar]
- 9. Roze S, Saunders R, Brandt A-S, de Portu S, Papo NL, Jendle J. Health-economic analysis of real-time continuous glucose monitoring in people with type 1 diabetes. Diabet Med. 2015;32(5):618-626. [DOI] [PubMed] [Google Scholar]
- 10. Wasserstein RL, Lazar NA. The ASA’s statement on p-values: context, process, and purpose [published online ahead of print March 7, 2016]. Am Stat. 2016;70(2):129-133. [Google Scholar]
- 11. Pickup JC. Insulin-pump therapy for type 1 diabetes mellitus. N Engl J Med. 2012;366(17):1616-1624. [DOI] [PubMed] [Google Scholar]
- 12. Redberg RF. Sham controls in medical device trials. N Engl J Med. 2014;371(10):892-893. [DOI] [PubMed] [Google Scholar]
- 13. Yeh H-C, Brown TT, Maruthur N, et al. Comparative effectiveness and safety of methods of insulin delivery and glucose monitoring for diabetes mellitus: a systematic review and meta-analysis. Ann Intern Med. 2012;157(5):336-347. [DOI] [PubMed] [Google Scholar]
- 14. Hanaire H. External insulin pump treatment in the day-to-day management of diabetes: benefits and future prospectives. Diabetes Metab. 2011;37(suppl 4):S40-S47. [DOI] [PubMed] [Google Scholar]
- 15. Nathan DM, Cleary PA, Backlund J-YC, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med. 2005;353(25):2643-2653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. NICE. Continuous subcutaneous insulin infusion for the treatment of diabetes mellitus. Available at: https://www.nice.org.uk/guidance/ta151. Accessed January 4, 2016. [DOI] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.