Effects of Moderate-to-VigorousIntensity Physical Activity onOvernight and Next-DayHypoglycemia in ActiveAdolescents With Type 1 DiabetesDiabetes Care 2014;37:1272–1278 | DOI: 10.2337/dc13-1973

OBJECTIVE

Physical activity (PA) provides many benefits to adolescents with type 1 diabetes;however, these individuals tend to have lower fitness and PA levels than theirdisease-free counterparts. The purpose of this study was to examine the acutetemporal associations between moderate-to-vigorous intensity PA (MVPA) andhypoglycemia (continuous glucose monitor [CGM] reading £70 mg/dL).

RESEARCH DESIGN AND METHODS

Nineteen participants (53% females) 14–20 years old with type 1 diabeteswere recruited. Participant fitness was evaluated via indirect calorimetryusing a maximal exercise test; body composition wasmeasured using air displace-ment plethysmography. An accelerometer was worn continuously (3–5 days) andacceleration data used to estimate MVPA (minutes per day). Blood glucose valueswere simultaneously tracked using CGM. Controlling for sex, percent body fat(%BF), fitness, and concurrent MVPA, the likelihood of nighttime and next-dayhypoglycemia due to MVPA was examined using logistic regression.

RESULTS

Participants were of average fitness (females: 43.9 mL/kg/min; males: 49.8mL/kg/min) and adiposity (females: 26.2%; males: 19.2%); 63.2% met the U.S.federal guideline of accumulating 60 min/day of MVPA. Hypoglycemia was 31%more likely in those who accumulated 30 min/day more MVPA in the previousafternoon than those with less (95% CI 1.05–1.63; P = 0.017).

CONCLUSIONS

The results suggest that participating in afternoon MVPA increases the risk ofovernight and next-day hypoglycemia, independent of sex, %BF, fitness, and con-current MVPA. While promoting PA as a healthy behavior, it is important toeducate adolescents with type 1 diabetes on prevention of hypoglycemia follow-ing PA.

There are many benefits of physical activity (PA) in adolescents, including improvedblood lipid profiles, cardiovascular fitness, bone health, and psychological well-being. PA is also inversely associated with levels of adiposity and stress (1). Notonly are these health outcomes immediately beneficial, but also health outcomes

1Department of Health and Human Physiology,University of Iowa, Iowa City, IA2Department of Pediatrics, University of Iowa,Iowa City, IA3Department of Biostatistics, University of Iowa,Iowa City, IA4School of Sport, Exercise & Health Sciences,Loughborough University, Leicestershire, U.K.

Corresponding author: Kristen M. Metcalf,kristen-metcalf@uiowa.edu.

Received 21 August 2013 and accepted 24January 2014.

© 2014 by the American Diabetes Association.See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

Kristen M. Metcalf,1 Ajay Singhvi,2

Eva Tsalikian,2 Michael J. Tansey,2

M. Bridget Zimmerman,3 Dale W. Esliger,4

and Kathleen F. Janz1

1272 Diabetes Care Volume 37, May 2014

CLINCARE/ED

UCATION/N

UTR

ITION/PSY

CHOSO

CIAL

and physically active behavior track intoadulthood, further increasing the im-portance of regular PA (1). The U.S. De-partment of Health and Human Services’2008 Physical Activity Guidelines forAmericans recommends a minimum of60 min of moderate-to-vigorous inten-sity PA (MVPA) daily for children andadolescents (2). Despite the benefits ofPA, the latest data from the 2011 Cen-ters for Disease Control and PreventionYouth Risk Behavior Surveillance Systemindicate that only 28.7% of high schoolstudents are meeting the federal guide-line of accumulating 60 min of MVPAdaily (3). Research shows that these pro-portions are even lower for individualswith chronic diseases, including type 1diabetes (4,5). Specifically, adolescentswith type 1 diabetes spend 17 fewerminin MVPA per day than healthy adoles-cents and have 3–10 mL/kg/min lowerVO2max values (5,6).While regular PA is an important as-

pect of disease management for adoles-cents with type 1 diabetes, fear ofhypoglycemia may lead to low levels ofPA (7,8). This fear is valid because of thecounterregulatory systems that controlinsulin levels not functioning properly,increasing the risk of exercise-inducedhypoglycemia. Episodes of hypoglyce-mia have been associated in childrenand adolescents participating in pro-longed moderate PA.60 min (9). Otherfactors that affect the glycemic re-sponse to exercise of an individualwith type 1 diabetes include exerciseduration, intensity, familiarity withtype of activity, metabolic control,blood glucose levels, type and timingof insulin injections and food intake, in-sulin absorption, muscle mass requiredfor activity, conditioning, degreeof stress,and timing of activity (10). Robertsonet al. (10) note that given consistenttiming of exercise, amount of insulin,and pre-exercise meal, the responseto 60 min of PA can be reproducible,indicating that it is safe for adolescentswith type 1 diabetes to be physicallyactive.The relationship between PA and

glycemic control is complex. Some studieshave shown no benefit of PA on glycemiccontrol, as measured by HbA1c (11).Others have shown that PA can reduceinsulin requirements by 6–15% (11).This relationship, however, may bemoderated by certain factors. Lukacs

et al. (5) have shown fitness to be in-versely associated with HbA1c, whileother studies have failed to show an as-sociation (12,13). Research on the rela-tionship between adiposity and HbA1chas shown mixed results. In one study,BMI and adiposity were positively asso-ciated with HbA1c (14); however, otherstudies have failed to show a significantassociation (5,9). Sex may or may not bean importantmodifier in the relationshipbetween PA and glycemic control. Someresearch shows a lack of association be-tween sex and HbA1c (9), while otherscite males as having more favorableHbA1c values, independent of PA, thanfemales (14).

Studies have looked at the complexrelationship between PA and glycemiccontrol in young patients with type 1diabetes, butmost have used self-reportmethods, such as questionnaires, to as-sess daily PA participation (9,14–16).While questionnaires are a cost- andtime-effective tool for assessing habit-ual PA, accelerometers are a more accu-rate measure of acute PA, as well astotal PA level and energy expenditure.Additionally, accelerometers allowviewing of daily activity patterns (17).

The aim of this study was to analyzethe relationship between MVPA and hy-poglycemia through the assessment oftheir acute, temporal associations. Weasked: 1) what are the acute, temporalassociations between MVPA and hypo-glycemia in adolescents with type 1 di-abetes and 2) are these associationsmoderated by sex, fitness, or adiposity?

RESEARCH DESIGN AND METHODS

Consent ProceduresThe Institutional Review Board at theUniversity of Iowa approved the studyprotocol, assent, and consent form. Aparent or guardian provided writtenconsent for all participants ,18 yearsold, and participants provided assent.Participants .18 years old providedtheir own consent. Participants werealso queried about their PA over thepast 7 days.

Eligibility Criteria and AssessmentTo be eligible for the study, the partici-pant had to: 1) be between 14 and 20years of age and at least Tanner stage IIIbreast development for girls and at leastTanner stage III genitalia developmentfor boys; 2) have a clinical diagnosis of

type 1 diabetes of $1 year duration; 3)have a stable insulin regimen using aninsulin pump or multiple daily injectionsfor at least 12 months prior; 4) have anHbA1c #10.0% in the past 3 monthsmeasured with the DCA 2000 (Bayer Di-agnostics, Tarrytown, NY); 5) have a BMIbetween the 5th and 95th percentile forage and sex; 6) have normal thyroid-stimulating hormone in the past12 months; and 7) be willing and able tocomplete all study requirements. Partici-pants were not eligible if they: 1) werehospitalized in the past month; 2) usedsystemic glucocorticoids in the pastmonth; 3) had amusculoskeletal problemor physical or mental illness that may af-fect exercise performance; or 4) had amedical condition or were using a medi-cation that, in the judgment of the inves-tigator, could affect completion of theexercise protocol.

Study ProceduresThe study consisted of two visits to theClinical Research Unit of the Institute forClinical and Translational Science (ICTS)at the University of Iowa separated by3–5 days. During the first visit, the con-tinuous glucose monitor (CGM) (SEVENPLUS; Dexcom, San Diego, CA) was cali-brated and inserted subcutaneously, an-thropometric measures and VO2maxtesting were performed, and an acceler-ometer (device model 1.1, GENEActiv;Activinsights Ltd., Kimbolton, U.K.) wasplaced on the left wrist. Participantswore the CGM and accelerometer untilthey returned for the second study visit.

Hypoglycemic EventsInterstitial glucose values were re-corded in milligrams per deciliter usingCGM. Capillary glucose was checkedduring the initial visit using the patient’shome glucose meter at 2:00, 3:00, and4:00 P.M., and these values were used tocalibrate the CGM. The CGM sensor wasplaced subcutaneously during the initialvisit to the ICTS and removed upon com-pletion of the second visit to the ICTS 3–5 days later. The CGM collected glucosereadings every 5 min, 24 h/day, and wascalibrated with the patient’s home glu-cose meter two times daily. Data fromthe CGM were downloaded during thesecond visit, and glucose values wereused to calculate the proportion of read-ings qualified as a hypoglycemic event,defined as any CGM reading#70mg/dL.Alternative analyses were completed for

care.diabetesjournals.org Metcalf and Associates 1273

hypoglycemic episodes, defined as suc-cessive CGM readings #70 mg/dL, at5-min intervals, with an episode endingwith at least two CGM readings .70mg/dL.

AdiposityBody fat and fat-free mass were mea-sured using air displacement plethys-mography via the BOD POD GoldStandard Body Composition TrackingSystem (model 2007A; COSMED USA,Inc., Concord, CA). The BOD POD usesbodymass and body volume to calculatebody density in grams per milliliter.Body mass is measured using a platformscale connected to the BOD POD, andbody volume is measured using air dis-placement. Body density is then used tocalculate percent body fat (%BF) usingthe Lohman density equation. The BODPOD has excellent validity when com-pared with dual energy X-ray absorpti-ometry in young females, with anintraclass correlation coefficient of0.92 (18). In adolescent males, theBOD POD has a correlation coefficientof 0.90when comparedwith hydrostaticweighing (19).

Cardiovascular FitnessCardiovascular fitness (VO2max) wasmeasured via maximal exercise testingto volitional fatigue by open-circuit spi-rometry using a metabolic cart (ParvoMedics, Sandy, UT). Participants werefitted with a heart rate chest strap (PolarUSA, Lake Success, NY) before com-pleting a modified Balke treadmill pro-tocol. Immediately before testing,participants were seated for 5 min, afterwhich resting heart rate and blood pres-sure were measured. Individuals thenpracticed walking on the treadmill whilebeing given instructions on propertreadmill walking technique. Partici-pants self-selected a brisk but comfort-able walking pace, and speeds rangedfrom 3.0–4.0 mph at 0% grade, for aninitial 4-min warm-up. While keepingthe speed constant, the grade wasthen increased to 5% for 4 min. Speedwas increased by 0.5 mph for 2 min,after which the grade was increased by2% each minute until exhaustion (20).During the test, measures of gas ex-change and heart rate were recorded.Ratingof perceived exertionwas assessedusing Borg’s 15-point scale during eachstage. Participants were considered tohave reached their maximal effort

when two of the following criteriawere met: 1) heart rate $200; 2) respi-ratory exchange ratio .1.0; or 3) #2mL/kg/min change in VO2 in final 60 sof test. VO2max in mL/kg/min was usedin analysis. Insulin levels were not ad-justed for exercise, as the exercise ses-sion was designed to mimic a typicalexercise session, and our participantsdid not typically adjust for exercise.If a participant experienced a bout ofhypoglycemia, he or she was treatedwith 15 g of quick-acting carbohydrateand then retested and retreated, if nec-essary, after 15 min.

MVPAPrior to fitness testing, a GENEActiv ac-celerometer (Activinsights Ltd.) wasplaced on the left wrist, to be worn be-tween the two clinic visits for 3–5 days.The GENEActiv has excellent criterionvalidity in both adults (r = 0.86) and chil-dren (r = 0.91) when worn at the leftwrist (21,22). Accelerometers were pro-grammed to collect data, starting on theinitial visit day, at a frequency of 75 Hz.Participants were instructed to wear themonitor 24h/day, includingwhile sleepingand during water activities. Accelerome-ters were removed at the completionof the participants’ second visit to theICTS, and data were downloaded. Theraw acceleration output was convertedto 60-s epochs using the GENEActivPost-Processing PC Software (version2.2, GENEActiv; Activinsights Ltd.). Next,the 60-s epoch data files were enteredinto an open source Excel macro (v2; Ac-tivinsights Ltd.) in order to classify activ-ity as sedentary, light, moderate, orvigorous intensity. MVPA (min/day)was calculated for each participant-day.Validated acceleration magnitude cutpoints from Esliger et al. (21) were usedto classify activity intensity. KineSoft soft-ware (version 3.3.75; KineSoft, Lough-borough, U.K.) was used to produce aseries of standardized accelerometryoutcome variables following proceduressimilar to those described by Esliger andTremblay (23) and Esliger et al. (24).KineSoft was also used to produce heatmaps, which show the amount of timespent in each activity intensity each day.In addition to all-day MVPA (MVPATOT),defined as 6:00 A.M. through bedtimein minutes per day, hours of the daywere split in two for analysis: daytimePA (MVPA06–15), defined as 6:00 A.M.

through 3:00 P.M.; and afternoon/evening PA (MVPA15–BT), defined asthe post–school day activity hours of3:00 P.M. through bedtime. Data werecollected between November 2011and April 2012.

Statistical AnalysisDescriptive statistics, including mean, SD,and range, were calculated for HbA1c,MVPA (min/day), VO2 max (mL/kg/min),adiposity (%BF), and number of hypo-glycemic events. Sex-specific differen-ces were evaluated using independentsample t tests. Hypoglycemia was rep-resented by the proportion of hypogly-cemic events occurring overnight andthe next day following activity. Statisti-cally significant associations betweenMVPA, using 30-min increments, andbouts of hypoglycemia were assessedusing logistic regression. Because wewere interested in acute effects, the lo-gistic regression used individual days ofdata collection rather than participants.Generalized linear mixed-model analy-sis was used to account for multiple out-come measures per participant. Modelswere constructed using previous-dayMVPA: 1) MVPATOT; 2) daytime MVPA(MVPA06–15); and 3) afternoon/eveningMVPA (MVPA15–BT) as exposure varia-bles. Bouts of hypoglycemia outcomevariables were: 1) overnight and nextday; 2) overnight (only); and 3) nextday (only). From the logistic regression,odds ratios (ORs) and 95% CIs were cal-culated. All models were adjusted forsex, VO2max, %BF, and concurrentMVPA (MVPA accumulated on thesame day as hypoglycemia risk was as-sessed). Analysis was completed on alldays and weekdays only; however, theresults were not affected by removingweekend days from analysis. Final anal-ysis includes all days of theweek. P valuesof ,0.05 were considered to be statisti-cally significant. All statistical analyseswere performed using SAS (SAS version9.3; SAS Institute, Cary, NC).

RESULTS

Twenty adolescents participated in thestudy; however, analyses are basedon 19 participants (82 days of data), as1 participant was lost to noncompliancewith study procedures. Participant char-acteristics and covariates (age, height,weight, %BF, VO2max, and MVPA),HbA1c, and events of hypoglycemia are

1274 Effects of MVPA on Hypoglycemia Diabetes Care Volume 37, May 2014

shown in Table 1. Average age of partic-ipants was 16.6 6 1.6 years; 53% werefemale. Males and females had meanHbA1c values of 8.6 and 8.2%, respec-tively. Participants were of average fit-ness with a mean VO2max of 43.8mL/kg/min for females (SD 6.8) and49.8 mL/kg/min for males (SD 6.8)(25). Participants also had average levelsof adiposity with a mean %BF in femalesof 26.1% (SD 4.8), and 19.1% in males(SD 9.3) (26). Sixty-three percent of par-ticipantsmet theU.S. federal guideline foradolescents of accumulating 60 min/dayof MVPA (2). Of the participants whomet the guideline, 75% (n = 9) reportedparticipating in organized sports duringthe 7 days prior to the study. The meanMVPA per day was 107.2 6 57.5 min inmales and 117.8 6 49.4 min in females.There were no statistically significant dif-ferences between male and femaleparticipants for MVPA and events of hy-poglycemia. A heat map of average timespent in sedentary, light, moderate, andvigorous physical activities throughoutthe week by all participants is shownin Fig. 1. The heat map represents all19 participants for each day that theywore the accelerometer and showsthe time spent in each intensity of PA(minutes per hour) throughout eachday. The majority of each hour is spentin sedentary and light-intensity activity,with the highest amounts of MVPA

occurring in the late afternoon and earlyevening hours. Due to small samplesize, our a priori analysis was betweenMVPA and hypoglycemia, as informedby previous research performed by theDirecnet Study Group (9). Therefore,we did not analyze sedentary andlight-intensity PA.

All 19 participants experienced atleast one bout of hypoglycemia overthe 82 participant-days analyzed. Four-teen participants had one or more hypo-glycemic events during the maximalexercise test. The median number of hy-poglycemic readings per day was 8.5(range 0–87). Specifically, the averagenumber of hypoglycemic episodes over-night was 0.7 per subject per night(range 0–4); the median duration perepisode was 20 min (range 5–145). Theaverage number of hypoglycemic epi-sodes the next day was 1.3 per subjectper day (range 0–7); the median dura-tion per episode was 20 min (range 5–135). The mean proportion of CGMreadings with a hypoglycemic eventwas 4.45% of readings taken duringsleep and 5.45% of readings taken dur-ing the day following activity.

Overnight and Next-DayHypoglycemiaRisk of hypoglycemia logistic regressionmodels described below were adjustedfor sex, VO2max, %BF, and concurrent

MVPA. Adjusted ORs for the full modelare presented in Table 2. Previous-dayMVPATOT (n = 50) was significantly asso-ciated with overnight and next-day hy-poglycemia (OR 1.26 [95% CI 1.08–1.46];P = 0.003). For every 30-min increase inprevious-day MVPATOT, overnight andnext-day hypoglycemia risk increasedby 26%. When we segmented the pre-vious day to more precisely examinetemporal responses, MVPA06–15 (n =50) was associated with a higher risk ofovernight and next-day hypoglycemia(OR 1.34 [95% CI 1.02–1.77]; P =0.036). Additionally, MVPA15–BT (n =67) was significant, with an OR of 1.36(95%CI 1.13–1.63; P = 0.001). (MVPA15–BTanalyses have a higher sample size dueto the inclusion of initial visit day MVPA,which was only available for afternoonhours.)

Overnight HypoglycemiaWe also examined the risk of hypoglyce-mia during overnight (as opposed toovernight and next day). Previous-dayMVPATOT (n = 58) was significantly asso-ciated with bouts of overnight hypogly-cemia (OR 1.31 [95% CI 1.03–1.67]; P =0.026). A nonsignificant association wasseen for the previous day for MVPA06–15(n = 58) (OR 1.48 [95% CI 0.95–2.32]; P =0.082); however, MVPA15–BT (n = 70)was significant (OR 1.43 [95% CI 1.06–1.93]; P = 0.021). Additionally, maleshad a higher risk than females of ex-periencing a bout of hypoglycemiaovernight (OR 3.14 [95% CI 1.16–8.47];P = 0.026).

Next-Day HypoglycemiaFinally, we examined the risk of hypogly-cemia during the next day following ac-tivity (as opposed to overnight and nextday). Previous-dayMVPATOT (n = 51)wassignificantly associated with next-dayhypoglycemia (OR 1.28 [95% CI 1.07–1.53]; P = 0.008). Increasing MVPA by30 min/day significantly increased riskof next-day hypoglycemia by 28%.When the previous day’sMVPAwas seg-mented, MVPA06–15 (n = 51) was not sig-nificant (OR 1.11 [95% CI 0.83–1.49]; P =0.454); however, MVPA15–BT (n = 68)was (OR 1.31 [95% CI 1.05–1.63]; P =0.017). Increasing MVPA15-BT by 30min/day significantly increased risk ofnext-day hypoglycemia by 31%. In addi-tion, participants with a higher VO2maxhad a higher risk of experiencing next-day hypoglycemia than those with a

Table 1—Descriptive statistics for participant characteristics, MVPA, andhypoglycemic events

Males (n = 9)* Females (n = 10)*

Age (years) 16.7 6 1.9 16.5 6 1.3

Height (m) 1.7 6 0.1 1.6 6 0.1

Weight (kg) 65.2 6 12.7 60.5 6 7.9

%BF† 19.2 6 9.4 26.2 6 4.8

Percent fat-free mass† 80.8 6 8.9 73.8 6 4.6

VO2max‡ 49.8 6 6.8 43.9 6 6.4

HbA1c [% (mmol/mol)] 8.6 (70) 6 0.9 (9.8) 8.2 (66) 6 0.9 (9.8)

MVPA all day (min) 107.2 6 57.5 117.8 6 49.4

MVPA daytime, 6:00 A.M.–2:59 P.M. (min) 45.4 6 36.7 56.5 6 26.5

MVPA afternoon/evening, 3:00 P.M.

to bedtime (min)§ 60.7 6 31.6 68.6 6 44.6

Overnight/next-day hypoglycemia (%)| 7.2 (4.8–10.8) 3.8 (2.2–6.5)

Overnight hypoglycemia (%) 5.4 (3.2–9.1) 3.4 (1.6–6.9)

Next-day hypoglycemia (%) 7.0 (4.6–10.5) 4.0 (2.3–6.8)

Data are presented asmean6 SD unless otherwise indicated. *Data were collected for 43 days formales and 38 days for females. †%BF and percent fat-free mass determined by air-displacementplethysmography. ‡VO2max measured using indirect calorimetry in a maximal exercise test.§Bedtime varied by participant. Times ranged from 8:00 P.M. to 3:00 A.M., with an averagebedtime of 11:18 P.M. |Proportion of readings with hypoglycemic event (glucose #70 mg/dL).Data presented as proportion (95% CI).

care.diabetesjournals.org Metcalf and Associates 1275

lower VO2max. This relationship existsfor MVPA15–BT (OR 1.50 [95% CI 1.05–2.16]; P = 0.030).

CONCLUSIONS

While PA providesmany benefits to ado-lescents with type 1 diabetes, includingimprovements in cardiovascular riskprofile, cardiovascular fitness, bonehealth, and adiposity, there is a risk ofhypoglycemia after activity. In ourstudy, MVPA was assessed as totalminutes per day, daytime hours andafternoon/evening hours. This strategywas used to assess the effect of MVPAaccumulated during afterschool hours,which is likely the most common timeto participate in MVPA, and is also thetime of most organized school athleticspractices. We found the greatest risk forhypoglycemic events to be associatedwith afternoon/evening MVPA. Thiswas the time that our participants accu-mulated the greatest amount of MVPA;in general, this is when adolescents ac-cumulate the majority of their PA.Participants in our study were of av-

erage fitness; our sample differs fromthose in much of the previous literature,which indicates that adolescents withtype 1 diabetes have lower levels of fit-ness (5,6). The participants also accumu-lated more daily MVPA than has beenreported in previous studies (4,13);mean MVPA was 107.2 6 57.5 min/day

for males and 117.86 49.4 min/day forfemales. Sixty-eight percent of the ado-lescents participated in organized sports.Given the study required a fitness andbody composition test, more active andfit individuals may have volunteered. Inagreement with the previous literature(5,14), participants had average adipos-ity levels, and adiposity did not have aneffect on risk of hypoglycemia. Sex, how-ever, was significant in the relationshipbetween MVPA and overnight hypogly-cemia, with males having a greater risk.Male participants had higher percent-ages of fat-free mass, on average, thanthe female participants, which may con-tribute to this difference in risk, dueto increased skeletal muscle glucosetransport.

We report that in adolescents withtype 1 diabetes, a 30-min increase inafternoon/evening MVPA increases therisk of hypoglycemia by 43% overnightand 31% on the following day. The asso-ciations are independent of sex, fitness,and adiposity, and the following-day re-lationship is stronger when adjusted forconcurrent MVPA. The associations areexponential; if an adolescent accumu-lates the recommended 60 min/day ofMVPA, they have a 104 and 72% higherrisk for hypoglycemia overnight andthe following day, respectively, whencompared with no MVPA. An increasedrisk of thismagnitudewarrants education

for not only the adolescent, but also forcoaches, parents, and school officials.

Organized sports often have practicesof 90–120 min, which would more thandouble their risk of having a hypoglyce-mic event. When adjusting for minutesof MVPA, however, participants in orga-nized sports did not differ in their hypo-glycemia risk from those not in organizedsports. This indicates that participationin organized sports does not increase hy-poglycemia risk over independent activ-ity of the same intensity and duration.The results also indicated that higher fit-ness is associated with an increasedrisk for hypoglycemia of 50% (for every5 mL/kg/min increase in VO2max). Theimplication is that being fit or habituallyactive does not protect against bouts ofhypoglycemia, but rather increases risk.Participants with higher fitness hadlower %BF; higher amounts of leanbody mass aid in transporting glucoseinto the muscles, leading to hypoglyce-mia. Additionally, in our study, the morefit adolescents had higher amounts ofvigorous intensity PA within their mi-nutes of MVPA, suggesting that absolutevolume of PA (for example metabolicequivalent of task minutes) may also in-fluence hypoglycemia risk.

In agreement with previous researchcompleted by the DirecNet Study Group(9), the current study found an in-creased risk of overnight hypoglycemiawith more MVPA. Additionally, wefound an extended period of time inwhich adolescents should be cognizantof their blood glucose levels. This findingindicates that there is an increased win-dow of time of 32 h post activity inwhich adolescents were at continuedrisk of hypoglycemia.

Strengths of the current study includethe use of valid measures to obtainstudy values, including the measure-ment of VO2max using indirect calorim-etry during a maximal exercise test, aswell as using air-displacement plethys-mography to assess %BF. Anotherstrength was the use of CGM and 24-haccelerometry to assess the temporalassociations between MVPA and hypo-glycemia round the clock. Much of theresearch in PA and adolescents withtype 1 diabetes has been completedin a controlled laboratory setting. Theuse of CGM and accelerometry allowedfor the analysis of the PA and hypogly-cemia relationship in a free-living

Figure 1—Heat map of average physical activity patterns of all participants. aTimes based onaverage activity of all participants (n = 19).

1276 Effects of MVPA on Hypoglycemia Diabetes Care Volume 37, May 2014

setting, permitting the measurement ofparticipants’ normal daily activity.Useof theGENEActiv is a strengthof this

study; it was used to decrease complianceissues seen with other accelerometers.The accelerometer was worn at the wristand is waterproof, ensuring participants

were able to comfortably wear the accel-erometer 24 h/day, resulting in full-daydata collection. In our study, compliancewas measured as the proportion of dayswith $600 min of wear time over thenumber of anticipated collection days;our compliance rate was 92%. Nineteen

of the 20 participants wore the monitorat least 2 days for $600 min/day andcontributed to the analysis. Anotherbenefit of the GENEActiv is that dataare collected in units of raw acceleration(21). The use of raw acceleration allowsfor more transparent data analyses thanproprietary movement counts.

Limitations of the current studyinclude a small convenience sample. Ad-ditionally, there were no dietary mea-sures collected. Diet is an importantaspect of glycemic control, and couldaffect hypoglycemic events. To partiallyaddress this limitation, analyses werecompleted adjusting for initial glucosevalues during both sleep and wakinghours; however, they were not signifi-cant. Use of CGM also presents limita-tions, including loss of sensitivity duringbouts of hypoglycemia and numericalerrors (27). To address this limitation,CGMs were calibrated twice daily withpatients’ home glucose meters. Futureanalyses should continue to assesspatterns of hypoglycemia to betterdetermine a time of highest risk ofhypoglycemia.

We report that increasing afternoon/evening MVPA results in an increasedrisk of hypoglycemia overnight and dur-ing the day following activity. Addition-ally, higher fitness levels are associatedwith an increased risk of hypoglycemia.This indicates the importance of educat-ing adolescents with type 1 diabetes, theirparents, and their coaches on preventionof hypoglycemia following activity.

Acknowledgments. The authors thank all ofthe study participants and the staff of theInstitute for Clinical and Translational Sciencefor the work completed.Funding. This work was supported by a grantfrom the Doris Duke Charitable Foundation, theNational Center for Advancing TranslationalSciences, and the National Institutes of Healththrough grant 2-UL1-TR000442-06 for the In-stitute for Clinical and Translational Science.Duality of Interest. M.J.T. is a consultant forDaiichi Sankyo. No other potential conflicts ofinterest relevant to this article were reported.Author Contributions. K.M.M. wrote themanuscript and researched data. A.S., E.T.,M.J.T., and D.W.E. researched data and re-viewed and edited the manuscript. M.B.Z.researched data and reviewed the manuscript.K.F.J. designed the study, researched data, andreviewed and edited the manuscript. K.M.M. isthe guarantor of this work and, as such, had fullaccess to all the data in the study and takesresponsibility for the integrity of the data andthe accuracy of the data analysis.

Table 2—ORs for temporal associations between daily MVPA and hypoglycemia

Variable OR (95% CI)*

Overnight and next day: MVPATOTNext-day MVPA 0.89 (0.73–1.09)Sex 1.70 (0.69–4.14)VO2max 1.28 (0.85–1.93)%BF 1.02 (0.94–1.10)MVPA† 1.26 (1.08–1.46)‡

Overnight and next day: MVPA06–15Next-day MVPA 0.89 (0.74–1.08)Sex 1.58 (0.59–4.23)VO2max 1.32 (0.84–2.07)%BF 1.01 (0.93–1.10)MVPA 1.34 (1.02–1.77)‡

Overnight and next day: MVPA15–BTNext-day MVPA 0.90 (0.77–1.04)Sex 1.93 (0.84–4.39)VO2max 1.40 (0.95–2.07)%BF 1.06 (0.98–1.13)MVPA 1.36 (1.13–1.63)‡

Overnight: MVPATOT§Sex 3.14 (1.16–8.47)‡VO2max 0.79 (0.48–1.31)%BF 1.00 (0.92–1.07)MVPA 1.31 (1.03–1.67)‡

Overnight: MVPA06–15Sex 2.83 (1.06–7.55)‡VO2max 0.83 (0.51–1.36)%BF 0.99 (0.92–1.06)MVPA 1.48 (0.95–2.32)

Overnight: MVPA15–BTSex 2.53 (0.94–6.84)VO2max 0.85 (0.52–1.38)%BF 1.01 (0.94–1.09)MVPA 1.43 (1.06–1.93)‡

Next day: MVPATOTNext-day MVPA 0.84 (0.64–1.10)Sex 1.50 (0.64–3.53)VO2max 1.35 (0.94–1.94)%BF 1.02 (0.95–1.10)MVPA 1.28 (1.07–1.53)‡

Next day: MVPA06–15Next-day MVPA 0.90 (0.71–1.13)Sex 1.38 (0.50–3.85)VO2max 1.43 (0.93–2.19)%BF 1.02 (0.94–1.12)MVPA 1.11 (0.83–1.49)

Next day: MVPA15–BTNext-day MVPA 0.89 (0.74–1.08)Sex 1.78 (0.80–3.98)VO2max 1.50 (1.05–2.16)‡%BF 1.06 (0.99–1.14)MVPA 1.31 (1.05–1.63)‡

*ORs are calculated based on days of data collection and adjusted for sex, VO2max, %BF, andconcurrent MVPA. †OR is for an additional 30 min of MVPA per day. ‡Results significant at P#0.05 level. §Overnight ORs are adjusted for sex, VO2max, and %BF.

care.diabetesjournals.org Metcalf and Associates 1277

Prior Presentation. Portions of this studywere presented in abstract form at the 73rdScientific Sessions of the American DiabetesAssociation, Chicago, IL, 21–25 June 2013.

References1. Loprinzi PD, Cardinal BJ, Loprinzi KL, Lee H.Benefits and environmental determinants ofphysical activity in children and adolescents.Obes Facts 2012;5:597–6102. U.S. Department of Health and Human Ser-vices. 2008 Physical Activity Guidelines forAmericans. Washington, D.C., U.S. Govt PrintingOffice, 20083. Centers for Disease Control and Prevention.Youth risk behavior surveillance -United States,2011. MMWR Surveill Summ 2012;61:1–1624. Maggio ABR, HoferMF,Martin XE,MarchandLM, Beghetti M, Farpour-Lambert NJ. Reducedphysical activity level and cardiorespiratory fit-ness in children with chronic diseases. Eur J Pe-diatr 2010;169:1187–11935. Lukacs A,Mayer K, Juhasz E, Varga B, Fodor B,Barkai L. Reduced physical fitness in childrenand adolescents with type 1 diabetes. PediatrDiabetes 2012;13:432–4376. Komatsu WR, Gabbay MA, Castro ML, et al.Aerobic exercise capacity in normal adolescentsand those with type 1 diabetes mellitus. PediatrDiabetes 2005;6:145–1497. Bernardini AL, Vanelli M, Chiari G, et al. Ad-herence to physical activity in young peoplewith type 1 diabetes. Acta Biomed 2004;75:153–1578. Brazeau AS, Rabasa-Lhoret R, Strychar I,Mircescu H. Barriers to physical activity amongpatients with type 1 diabetes. Diabetes Care2008;31:2108–21099. Tsalikian E, Mauras N, Beck RW, et al.; Dia-betes Research In Children Network DirecnetStudy Group. Impact of exercise on overnight

glycemic control in childrenwith type 1 diabetesmellitus. J Pediatr 2005;147:528–53410. Robertson K, Adolfsson P, Riddell MC,Scheiner G, Hanas R. Exercise in children andadolescents with diabetes. Pediatr Diabetes2008;9:65–7711. Chimen M, Kennedy A, Nirantharakumar K,Pang TT, Andrews R, Narendran P. What are thehealth benefits of physical activity in type 1 di-abetes mellitus? A literature review. Diabetolo-gia 2012;55:542–55112. Nadeau KJ, Regensteiner JG, Bauer TA, et al.Insulin resistance in adolescents with type 1 di-abetes and its relationship to cardiovascularfunction. J Clin Endocrinol Metab 2010;95:513–52113. Cuenca-Garcıa M, Jago R, Shield JP, BurrenCP. How does physical activity and fitness influ-ence glycaemic control in young people withType 1 diabetes? Diabet Med 2012;29:e369–e37614. Herbst A, Bachran R, Kapellen T, Holl RW.Effects of regular physical activity on control ofglycemia in pediatric patients with type 1 dia-betes mellitus. Arch Pediatr Adolesc Med 2006;160:573–57715. Aman J, Skinner TC, de Beaufort CE, SwiftPG, Aanstoot HJ, Cameron F; Hvidoere StudyGroup on Childhood Diabetes. Associations be-tween physical activity, sedentary behavior, andglycemic control in a large cohort of adolescentswith type 1 diabetes: the Hvidoere Study Groupon Childhood Diabetes. Pediatr Diabetes 2009;10:234–23916. Schweiger B, Klingensmith G, Snell-BergeonJK. Physical activity in adolescent females withtype 1 diabetes. Int J Pediatr 2010;2010:32831817. Westerterp KR. Assessment of physical ac-tivity: a critical appraisal. Eur J Appl Physiol2009;105:823–828

18. Maddalozzo GF, Cardinal BJ, Snow CA. Con-current validity of the BOD POD and dual energyx-ray absorptiometry techniques for assessingbody composition in young women. J Am DietAssoc 2002;102:1677–167919. Moon JR, Tobkin SE, Costa PB, et al. Validityof the BOD POD for assessing body compositionin athletic high school boys. J Strength Cond Res2008;22:263–26820. Nemeth BA, Carrel AL, Eickhoff J, Clark RR,Peterson SE, Allen DB. Submaximal treadmilltest predicts VO2max in overweight children. JPediatr 2009;154:677–68121. Esliger DW, Rowlands AV, Hurst TL, Catt M,Murray P, Eston RG. Validation of the GENEAAccelerometer. Med Sci Sports Exerc 2011;43:1085–109322. Phillips LR, Parfitt G, Rowlands AV. Calibra-tion of the GENEA accelerometer for assess-ment of physical activity intensity in children. JSci Med Sport 2013;16:124–12823. Esliger DW, Tremblay MS. Physical activityand inactivity profiling: the next generation. CanJ Public Health 2007;98(Suppl. 2):S195–S20724. Esliger DW, Tremblay MS, Copeland JL,Barnes JD, Huntington GE, Bassett DR Jr.Physical activity profile of Old Order Amish,Mennonite, and contemporary children. MedSci Sports Exerc 2010;42:296–30325. Shvartz E, Reibold RC. Aerobic fitnessnorms for males and females aged 6 to 75 years:a review. Aviat Space Environ Med 1990;61:3–1126. Ogden C, Li Y, Freedman D, Borrud L, FlegalK. Smoothed percentage body fat percentilesfor US children and adolescents, 1999-2004.Natl Health Stat Report 2011;43:1–727. Kovatchev B, Anderson S, Heinemann L,ClarkeW. Comparison of the numerical and clin-ical accuracy of four continuous glucose moni-tors. Diabetes Care 2008;31:1160–1164

1278 Effects of MVPA on Hypoglycemia Diabetes Care Volume 37, May 2014