Diabetes Care Volume 41, December 2018 2471 Denice S. Feig,1,2,3 Rosa Corcoy,4,5 Pumps or Multiple Daily Injections Lois E. Donovan,6 Kellie E. Murphy,1,2,3 CLIN CARE/EDUCATION/NUTRITION/PSYCHOSOCIAL Jon F.R. Barrett,7 J. Johanna Sanchez,7 in Pregnancy Involving Type 1 Tim Wysocki,8 Katrina Ruedy,9 fi Craig Kollman,9 George Tomlinson,3,10 and Diabetes: A Prespeci ed Analysis Helen R. Murphy,11,12,13 on behalf of the of the CONCEPTT Randomized CONCEPTT Collaborative Group* Trial Diabetes Care 2018;41:2471–2479 | https://doi.org/10.2337/dc18-1437 OBJECTIVE 1Mount Sinai Hospital, Sinai Health System, To compare glycemic control, quality of life, and pregnancy outcomes of women Toronto, Ontario, Canada 2Lunenfeld-Tanenbaum Research Institute, Tor- using insulin pumps and multiple daily injection therapy (MDI) during the onto, Ontario, Canada Continuous Glucose Monitoring in Women With Type 1 Diabetes in Pregnancy 3Department of Medicine, University of Toronto, Trial (CONCEPTT). Toronto, Ontario, Canada 4Hospital de la Santa Creu i Sant Pau, Barcelona, RESEARCH DESIGN AND METHODS Spain 5 fi CIBER-BBN, Zaragoza, Spain This was a prespeci ed analysis of CONCEPTT involving 248 pregnant women from 6University of Calgary, Calgary, Alberta, Canada 7 31 centers. Randomization was stratified for pump versus MDI and HbA1c. The Sunnybrook Research Institute, Toronto, On- primary outcome was change in HbA1c from randomization to 34 weeks’ gestation. tario, Canada 8 ’ Key secondary outcomes were continuous glucose monitoring (CGM) measures, Nemours Children s Health System, Jacksonville, FL maternal-infant health, and patient-reported outcomes. 9Jaeb Center for Health Research, Tampa, FL 10University Health Network, Toronto, Ontario, RESULTS Canada At baseline, pump users were more often in stable relationships (P = 0.003), more 11Cambridge University Hospitals NHS Founda- P P tion Trust, Cambridge, U.K. likely to take preconception vitamins ( = 0.03), and less likely to smoke ( = 0.02). 12 ’ fi 6 Department of Women and Children s Health, Pump and MDI users had comparable rst-trimester glycemia: HbA1c 6.84 0.71 vs. King’s College London, London, U.K. 6.95 6 0.58% (51 6 7.8 vs. 52 6 6.3 mmol/mol) (P = 0.31) and CGM time in target 13Department of Medicine, University of East (51 6 14 vs. 50 6 13%) (P = 0.40). At 34 weeks, MDI users had a greater decrease in Anglia, Norwich, U.K. HbA1c (20.55 6 0.59 vs. 20.32 6 0.65%, P = 0.001). At 24 and 34 weeks, MDI users Corresponding author: Denice S. Feig, d.feig@ utoronto.ca. were more likely to achieve target HbA1c (P = 0.009 and P = 0.001, respectively). Pump users had more hypertensive disorders (P = 0.011), mainly driven by increased Received 5 July 2018 and accepted 18 September 2018. gestational hypertension (14.4 vs. 5.2%; P = 0.025), and more neonatal hypogly- P This article contains Supplementary Data online cemia (31.8 vs. 19.1%, = 0.05) and neonatal intensive care unit (NICU) at http://care.diabetesjournals.org/lookup/suppl/ admissions >24 h (44.5 vs. 29.6%; P = 0.02). Pump users had a larger reduction doi:10.2337/dc18-1437/-/DC1. in hypoglycemia-related anxiety (P = 0.05) but greater decline in health/well-being *A complete list of the members of the CONCEPTT (P = 0.02). Collaborative Group can be found in the Sup- plementary Data online. CONCLUSIONS © 2018 by the American Diabetes Association. In CONCEPTT, MDI users were more likely to have better glycemic outcomes and less Readers may use this article as long as the work likely to have gestational hypertension, neonatal hypoglycemia, and NICU admis- is properly cited, the use is educational and not for profit, and the work is not altered. More infor- sions than pump users. These data suggest that implementation of insulin pump mation is available at http://www.diabetesjournals therapy is potentially suboptimal during pregnancy. .org/content/license. 2472 Pumps or Pens in Pregnancy Involving T1D? Diabetes Care Volume 41, December 2018 Insulin pumps or continuous subcutane- of blood glucose [SMBG]) compared with In the current analyses, the glycemic ous insulin infusions (CSIIs) are increas- SMBG alone. The results demonstrated control, quality of life, and pregnancy ingly used to optimize glucose control that use of CGM during pregnancy was outcomes of pregnant women using before and during pregnancy involving associated with improved glycemic con- pump therapy were compared with type 1 diabetes (T1D). Pump therapy trol with lower rates of neonatal com- those of pregnant women using MDI. has been shown to improve glycemic plications in both insulin pump and MDI The primary outcome was glycemic con- control, reduce severe hypoglycemic epi- users (17). trol as measured by a change in HbA1c sodes, and improve quality of life in The aim of this prespecified secondary from baseline to 34 weeks’ gestation. people with T1D outside of pregnancy (1). analysis was to compare maternal gly- Severe hypoglycemia was defined as However, the effectiveness of insulin cemic control, obstetric and neonatal an episode requiring third-party as- pump therapy during pregnancy is un- health outcomes, and patient-related sistance. Other secondary glycemic certain (2). Several randomized trials outcome measures in CONCEPTT partic- outcomes included CGM time in tar- showed no differences in glycemic con- ipants using pumps versus MDI. get range (63–140 mg/dL), time spent trol or pregnancy outcomes (3–6). These above and below the target range, studies were limited by small sample episodes of hypoglycemia (defined as sizes, performed in the 1980s using out- RESEARCH DESIGN AND METHODS CGM glucose values ,63 mg/dL for at dated equipment, and lacked statistical Full details of the clinical study protocol least 20 min), and glucose coefficient power for evaluating obstetric and neo- have previously been published (16). of variation (CV) to reflect glycemic natal outcomes. More recent cohort stud- Women with T1D were eligible if they variability. ies have found conflicting results, with were aged 18–40 years, had .12 months’ Patient-reported outcome measures some but not all finding differences in duration of diabetes, and were on an in- included the following questionnaires glycemic control (7–14). tensive insulin regimen using either a pump completed at baseline and 34 weeks’ The largest observational study, in- or multiple daily injections. Women plan- gestation. These assessed satisfaction cluding 113 pregnancies (.20 weeks’ ning pregnancy had to have an HbA1c level with their glucose monitoring system gestation), among insulin pump users between 7.0 and 10% (53–86 mmol/mol). (Blood Glucose Monitoring System Rat- reported clinically relevant differences Pregnant women had to have a live sin- ing Questionnaire) (18), fear of hypogly- fi in maternal HbA1c levels, favoring pump gleton fetus con rmed by ultrasound cemia (Hypoglycemia Fear Survey [HFS]-II]) users throughout pregnancy (10). De- before 14 weeks’ gestational age and an (19), diabetes-related distress (Prob- – spite lower HbA1c levels (0.7 and 0.3% HbA1c level between 6.5 and 10% (48 86 lem Areas in Diabetes scale) (20), and in the first and third trimesters, respec- mmol/mol). After a run-in period where qualityof life(12-ItemShort-FormSurvey tively), pregnancy outcomes were not eligible women wore a masked CGM for [SF-12]) (21). All questionnaires had ac- improved. Similar to an earlier Cochrane at least 96 h (neither they nor their clin- ceptable estimates of internal reliability review reporting higher birth weight in ical teams could see the glucose values), calculated using Cronbach’s a coefficient infants of mothers who used insulin women were randomized to the inter- or the Spearman-Brown split-half reliability pumps in pregnancy (15), this Canadian vention, where they received a CGM, coefficient. study also described a higher propor- or to the control group, where they Obstetric outcomes included hyper- tion of infants who were large for gesta- were instructed to continue with their tensive disorders (worsening chronic tional age (LGA): 55 vs. 39% for pump vs. usual SMBG at least seven times per hypertension, gestational hypertension multiple daily injection therapy (MDI). day. The CGM group were asked to with and without preeclampsia, and As with previous retrospective studies, perform a glucose test prior to insulin preeclampsia together and individu- these data were confounded by base- dose adjustment to verify the accuracy ally), cesarean section delivery, and line differences, with pump users being of the CGM, as required by the regula- gestational weight gain (22). Neonatal older, of higher parity, better educated, tory labeling instructions of the CGM outcomes included pregnancy loss (mis- and better prepared for pregnancy. How- manufacturer. carriage, stillbirth, neonatal death #28 ever, pump users also had a longer dura- In the main CONCEPTT trial, random- days of life), preterm birth (,37 and tion of diabetes and more retinopathy, ization was stratified by insulin delivery sys- ,34 weeks’ gestation), birth injury, suggesting that they may have had a tem (pump or MDI) and by baseline HbA1c shoulder dystocia, birth weight percen- more severe glycemic disturbance. level (,7.5 vs. $7.5% or 58 mmol/mol tile, rates of LGA or small for gestational The Continuous Glucose Monitoring in during pregnancy and ,8.0 vs. $8.0% or age (.90th or ,10th percentile us- Women With Type 1 Diabetes in Preg- 64 mmol/mol while planning pregnancy). ing Gestation Related Optimal Weight nancy Trial (CONCEPTT) was a multicen- Women in the SMBG planning pregnancy [GROW] software [23], which adjusts for ter, open-label, randomized trial in 325 and pregnant groups were asked to wear infant sex and gestational age, maternal women with T1D, who were either plan- another masked CGM at 12 and 24 weeks height, weight, parity, and ethnicity), ning pregnancy (n =110)orinearlypreg- and 24 and 34 weeks, respectively, for a neonatal hypoglycemia requiring intra- nancy (n = 215), from 31 centers in minimum of 96 h.