ORIGINAL RESEARCH

The Impact of Frequency and Tone of Parent–YouthCommunication on Type 1 Diabetes Management

Mark D. DeBoer . Rupa Valdez . Daniel R. Chernavvsky . Monica Grover .

Christine Burt Solorzano . Kirabo Herbert . Stephen Patek

Received: February 27, 2017 / Published online: April 12, 2017� The Author(s) 2017. This article is an open access publication

ABSTRACT

Background: The purpose of this study is toassess the impact of frequency and tone ofparent–youth communication on glycemiccontrol as measured by the Family Communi-cation Inventory (FCI). Adolescence provides aunique set of diabetes management challenges,

including suboptimal glycemic control. Con-tinued parental involvement in diabetes man-agement is associated with improved HbA1coutcomes; however, diabetes-related conflictwithin the family can have adverse effects.Although it is clear that communication playsan important role in diabetes outcomes, thespecific impact of frequency and tone of suchcommunication is largely understudied.Methods: A total of 110 youths with type 1diabetes and their parents completed question-naires assessing diabetes-related adherence,family conflict, and family communication (i.e.,frequency and tone) during a routine clinicvisit. Routine testing of HbA1c was performed.Results: Youth- and parent-reported frequencyof communication were unrelated to HbA1c.Instead, greater discrepancies between parentsand children on reported frequency of com-munication (most commonly parents reportingfrequent and youth reporting less frequentcommunication) corresponded with poorerglycemic control and increased family conflict.More positive tone of communication as ratedby youth was associated with lower HbA1c.Conclusions: Diabetes-related communicationis more complex than conveyed simply by howoften children and their parents communicate.Tone of communication and discrepancies in afamily’s perception of the frequency of com-munication were better than frequency as pre-dictors of glycemic control. The FCI appears tocapture the frequency and tone of

Enhanced Content To view enhanced content for thisarticle go to http://www.medengine.com/Redeem/8C08F0601556E12B.

Electronic supplementary material The onlineversion of this article (doi:10.1007/s13300-017-0259-2)contains supplementary material, which is available toauthorized users.

M. D. DeBoer (&) � M. Grover � C. Burt Solorzano �K. HerbertDepartment of Pediatrics, University of VirginiaHealth System, Charlottesville, VA, USAe-mail: deboer@virginia.edu

R. ValdezDepartment of Public Health Sciences, University ofVirginia School of Medicine, Charlottesville, VA22908, USA

D. R. ChernavvskyDepartment of Psychiatry and NeurobehavioralSciences, University of Virginia School of Medicine,Charlottesville, VA 22908, USA

S. PatekDepartment of Systems and InformationEngineering, University of Virginia School ofMedicine, Charlottesville, VA 22908, USA

Diabetes Ther (2017) 8:625–636

DOI 10.1007/s13300-017-0259-2

diabetes-related communication, though lar-ger-scale studies are warranted to inform futureuse of this scale.

Keywords: Adolescent; Communication;Diabetes mellitus; Family conflict; Parents;Surveys and questionnaires

INTRODUCTION

Adolescents with type 1 diabetes (T1D) face amyriad of obstacles to maintaining adequateblood glucose (BG) control, resulting in HbA1clevels that are the highest of any age group [1].Navigating the division of responsibility ofdiabetes management tasks between parentsand youth is a particularly salient challenge forthis population. Although it is recognized thatadolescents need to engage in some degree ofdevelopmentally appropriate diabetesself-management behaviors to prepare them forlife apart from their parents [2–4], it is wellestablished that they commonly have difficultycarrying out some of these tasks [5–9]. In addi-tion to the inherent challenges of adolescence,changes that occur within parent–youth rela-tionships and communication patterns duringthis developmental transition may also have animpact on diabetes management and glycemicoutcome. Continued parental involvement inadolescents’ diabetes management is recom-mended and has been associated with betteroutcomes (i.e., lower HbA1c) [10–14]. Thequality and degree of parental monitoring mayalso have an impact on a youth’s adherence andsubsequent glycemic control, with some evi-dence suggesting that parental monitoring isassociated with improved self-efficacy, adher-ence, and HbA1c in adolescents [15]. However,perceived parental over-involvement has beenassociated with poorer glycemic control andmore episodes of diabetic ketoacidosis [16].

Although family functioning [17] andco-regulation of care between parents and chil-dren [12, 13] appear to be important predictorsof diabetes control, the role of diabetes-relatedcommunication in diabetes management is notfully understood. Most research efforts havefocused on the quality of family

communication, demonstrating that increaseddiabetes-related family conflict [7, 18–20], par-ental negativity [17, 21], and parental criticism[17, 22] are related to suboptimal glycemiccontrol in youth, while perceived parentalwarmth, effective conflict resolution, and par-ental positive reinforcement are associated withbetter glycemic control [23]. Negative familycommunication, specifically diabetes-relatedconflict, also has a demonstrated negativeimpact on both general and diabetes-specificquality of life for youth with T1D [24, 25]. Infact, Laffel and colleagues found that dia-betes-related family conflict was the only sig-nificant predictor of quality of life aftercontrolling for demographic variables such asage, gender, and duration of diabetes, as well asHbA1c and parental involvement in insulinadministration [24].

While previous studies have investigated therelationship between degree of conflict andT1D-related outcomes, they have not examinedthe role of family communication frequency orvalence (positive to negative tone). Valence hasthe potential to offer additional insights intothe impact of family communication on dia-betes management because negative tone canencompass more subtle interactions in par-ent–youth dyads beyond explicit conflict (e.g.,sarcasm). Moreover, discrepancies betweenparent and youth perceptions of the overallfrequency and valence of communication mayalso play an important role in promotingadaptive diabetes care. Previous research hasdemonstrated that discrepancies in perceptionsof family members’ responsibility for diabetescare are predictive of poorer metabolic out-comes [26].

The goal of this study was to investigate therole of parent- and youth-reported frequencyand valence of communication about T1D inoverall glycemic control. To achieve this goal,we developed and tested a T1D communicationquestionnaire, the Family CommunicationInventory (FCI). Our primary hypothesis wasthat higher frequency and more positive tone ofcommunication between youth and their par-ents would be associated with lower HbA1cvalues. Our secondary hypothesis was thatgreater discrepancies in parent–youth

626 Diabetes Ther (2017) 8:625–636

perceptions regarding frequency and tone ofcommunication would be associated with sub-optimal metabolic control as determined byhigher HbA1c values and poorer diabetes man-agement behaviors.

METHODS

This study was approved by the University ofVirginia (UVA) Social and Behavioral SciencesInstitutional Review Board. All procedures fol-lowed were in accordance with the ethical stan-dards of the responsible committee on humanexperimentation (institutional and national)and with the Helsinki Declaration of 1964, asrevised in 2013. Informed consent was obtainedfrom all patients for being included in the study.Eligible individuals were youth between the agesof 12 and 18 years and their accompanying par-ents who were at a clinic visit to receive care at theUVA Diabetes Clinic for T1D. Participants andtheir accompanying parent were invited to par-ticipate in the study by clinic nursing staff at theirroutine diabetes clinic visit. If two parents werepresent at the visit, the parent who assumes pri-mary responsibility of the youth’s diabetes carewas asked to participate. Participants wereinformed that their responses were completelyconfidential and that their participation or lackthereof would not affect their clinical care in anyway. Upon informed consent and assent, partic-ipants and their parents were provided withparent and youth self-report questionnaires tocomplete independently at the clinic visit. Allcompleted questionnaires were collected by thefront desk staff at the end of the clinic visit. Theadolescent received a US$20 gift card forparticipation.

Measures

Glycemic controlGlycated hemoglobin (HbA1c) was used toprovide an estimate of the youths’ glycemiccontrol for the preceding 2–3 months. Clinicalnursing staff collected blood samples via fingerstick during the clinic visit, which were ana-lyzed using a DCA Vantage AnalyzerTM (Sie-mens AG, Munich, Germany).

QuestionnairesDiabetes Family Conflict Scale-Revised(DFCS) The revised DCFS was used to assessdiabetes-specific family conflict over the pastmonth [19]. It consists of 19 items scored on a3-point Likert scale (1 = never argue,2 = sometimes argue, 3 = always argue), sum-med to yield a total score ranging between 19(no conflict) and 57 (high conflict). The revisedDCFS previously demonstrated adequate inter-nal consistency for both youth and caregiverreports (a = 0.85, a = 0.81 respectively) and hadhigh internal consistency with the presentsample (youth a = 0.91, parent a = 0.92) [19].

Child Self-Management Scale (CSM) TheCSM was developed for the present study (Sup-plementary Table 1) to measure youth adher-ence to diabetes management tasks over thepast 2 weeks. It is scored as the mean of sixitems that assess the frequency of missingspecific required self-management tasks (e.g.,‘‘Missed taking insulin for a meal’’), which areresponded to on a 5-point Likert scale(0 = Never, 4 = seven or more times). Higherscores indicate more difficulty with adherenceto diabetes management tasks.

Family Communication Inventory (FCI) TheFCI is a nine-item measure created for this studyto assess parent–youth communication,including frequency and tone, related to theyouth’s diabetes management (Table 1). Itemswere generated by a pediatric endocrinologistand human factors specialist on the basis ofclinical and research-related experience inter-acting with families. Parents and youth wereinstructed to rate how often they communi-cated about specific diabetes-related taskswithin the past week on a 5-point Likert scaleranging from 0 (never) to 4 (multiple timesdaily). For each task, they were also asked to ratethe tone of the communication on a 5-pointLikert scale ranging from 0 (very negative) to 4(very positive). If a participant indicated that adiabetes-related task was never discussed, therewas an option to indicate that the tone of thecommunication was ‘‘not applicable.’’ Higheritem mean scores on frequency of communica-tion (FCom) indicate more frequent

Diabetes Ther (2017) 8:625–636 627

Table 1 Family communication scales: frequency and tone of communication

Item Youth Parent Correlation(youth to parent)

Mean SD Correlationto tone

Mean SD Correlationto tone

Spearman’s rho

FCom 1: discussed insulin dosing of a

particular meal

2.67 1.36 0.228* 2.89 1.31 0.246* 0.599**

TCom 1 �/ 3.83 0.70 – 3.68 0.79 – 0.538**

FCom 2: discussed insulin dosing of a

particular high blood sugar

2.29 1.05 0.092 2.29 0.91 0.038 0.480**

TCom 2 �/ 3.68 0.79 – 3.54 0.86 – 0.625**

Fcom 3: discussed treatment of a

particular low blood sugar /(–)

2.06 0.95 0.150 2.00 0.87 0.183 0.528**

TCom 3 3.85 0.76 – 3.89 0.69 – 0.423**

FCom 4: discussed missed insulin boluses

�

1.50 0.84 -0.471** 1.47 0.82 -0.431** 0.353**

TCom 4 3.55 0.92 – 3.34 1.00 – 0.551**

FCom 5: discussed overall blood sugar

control

2.80 1.18 0.181 2.89 1.12 0.151 0.526**

TCom 5 �/ 3.61 0.87 – 3.53 0.83 – 0.616**

FCom 6: discussed frequency/number of

low blood sugars �(–) /(–)

1.96 0.98 0.178 1.93 0.98 0.068 0.555**

TCom 6 � 3.68 0.74 – 3.78 0.66 – 0.481**

FCom 7: discussed frequency/number of

high blood sugars �

2.51 1.05 -0.084 2.42 0.91 -0.002 0.393**

TCom 7 �/ 3.50 0.88 – 3.47 0.93 – 0.704**

FCom 8: discussed adjustment of

insulin/carb ratio

1.74 0.99 -0.105 1.75 1.00 0.124 0.578**

TCom 8 � 3.82 0.72 – 3.75 0.72 – 0.699**

Fcom 9: discussed adjustment of Lantus/

basal rates

1.57 0.83 0.079 1.51 0.73 0.066 0.502**

TCom 9 � 3.84 0.78 – 3.88 0.72 – 0.662**

628 Diabetes Ther (2017) 8:625–636

communication, and higher item mean scoreson tone of communication (TCom) indicatemore positive tone.

Discrepancy Data

Evidence suggests that discrepancies in per-spectives on diabetes care between children andparents can negatively impact overall diabetescontrol [26]. To test whether this association isalso the case for frequency and tone of com-munication, we developed a measure to calcu-late the mean difference in response to FCIitems between each youth–parent dyad. Themeasure is akin to a standard deviation and isgiven by Eq. 1 where x is each item (e.g., FCom4) and n is the total number of items on thescale:

Standard deviationAgreement

¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Pnx¼1 Parent scoreitem x � Youth scoreitem xð Þ

N items

2s

ð1Þ

Disagreement score is calculated by taking thesquare root of the sum of squared differences ofall scale items divided by the total number ofitems. Disagreement scores give the mean dif-ference of reported frequency and tone betweenparents and youth, or the magnitude of dis-crepancy. A disadvantage of this measure is thatit removes the direction of differences (i.e.,

whether youth believe there is more frequentcommunication than parents). Therefore, weseparately summed the total number of itemswhere parents or youth endorsed higher valueson the FCI scales to quantify the direction ofdiscrepancies in communication.

RESULTS

Participant Characteristics

We assessed a total of 110 participants with T1D(50% female), with a mean age of 14.5 years(range 12–18). Participants’ medical character-istics included a mean duration of diabetes of6.6 years (range 0.04–15) and mean HbA1c was8.7% (71.6 mmol/mol) (range 5.6–14.0%;37.7–129.5 mmol/mol)—above the level of7.5% (58.5 mmol/mol) recommended by theAmerican Diabetes Association for this agerange [27].

Frequency of Communication (FCom)

Scale CharacteristicsTable 1 shows the mean, standard deviation,and correlations between parents and youth foreach FCom item, while Supplementary Table 2provides the number or responses to each item.Averaging across items, parent(M = 2.13 ± 0.69) and youth (M = 2.14 ± 0.64)scores correlated highly on the frequency of

Table 1 continued

Item Youth Parent Correlation(youth to parent)

Mean SD Correlationto tone

Mean SD Correlationto tone

Spearman’s rho

Pearson’s R

FCom overall 2.13 0.69 0.077 2.14 0.64 0.077 0.604**

TCom overall 3.73 0.63 – 3.71 0.67 – 0.831**

* Correlation is significant at p\0.05** Correlation is significant at p\0.01�Youth response was significantly related to HbA1c (p\0.05), controlling for youth age, gender, and duration of diabetes;(–) indicates that the increased frequency was associated with lower HbA1c/ Parent response was significantly related to HbA1c (p\0.05), controlling for youth age, gender, and duration of diabetes;(–) indicates that increased frequency was associated with lower HbA1c

Diabetes Ther (2017) 8:625–636 629

diabetes communication (r = 0.60, n = 109,p\0.0005, r2 = 36.5%). Cronbach’s alphas foryouth (a = 0.84) and parents (a = 0.81) weregood for the nine-item scale. Exploratory factoranalysis (EFA) conducted with principal axisfactoring revealed that all FCom items loadedonto a single factor at Eigen values greater than0.30 for both parents and youth respondents(see Table 2), while a two-factor solution yieldedcross loading on multiple items.

On average, parents and youth disagreed byless than 1 point per FCom item(M = 0.85 ± 0.63, range = 0.0–2.67). Overall,these discrepancies were balanced, with youthand parents reporting more communication on

an average of 1.89 items and 1.86 items,respectively, t (100) = 0.094, p = 0.925.

Relationship to HbA1cIn separate linear regression models, neitheryouth nor parent mean FCom scores signifi-cantly predicted HbA1c after controlling foryouth age, duration of diabetes, and gender.However, FCom disagreement score was posi-tively associated with HbA1c (standardizedb = 0.22, t (99) = 2.27, p = 0.026), suggestingthat more youth–parent disagreement aboutfrequency of communication correspondedwith higher HbA1c scores.

Table 2 Factor loadings for parents and youth on the frequency of communication (FCom) items

Item Youth factor loading Parent factor loading

FCom 1: discussed insulin dosing of a particular meal 0.464 0.339

FCom 2: discussed insulin dosing of a particular high blood sugar 0.775 0.782

FCom 3: discussed treatment of a particular low blood sugar 0.828 0.839

FCom 4: discussed missed insulin boluses 0.802 0.750

FCom 5: discussed overall blood sugar control 0.451 0.373

FCom 6: discussed frequency/number of low blood sugars 0.844 0.878

FCom 7: discussed frequency/number of high blood sugars 0.688 0.777

FCom 8: discussed adjustment of insulin/carb ratio 0.900 0.842

FCom 9: discussed adjustment of Lantus/basal rates 0.904 0.855

Table 3 Pearson’s correlations between frequency/tone of communication, the family conflict scale, and the childself-management scale

Parent FCS Youth FCS Parent CSM Youth CSM CSM disagreement

Parent FCom -0.047 -0.063 -0.079 -0.177* 0.005

Youth FCom -0.061 0.076 0.013 0.010 0.128

Frequency disagreement 0.216* 0.209* 0.169* 0.222* 0.428**

Parent TCom -0.342** -0.319** -0.352** -0.257* -0.287*

Youth TCom -0.471** -0.521** -0.388** -0.388** -0.397**

Tone disagreement 0.267* 0.460 ** 0.604** 0.445** 0.677**

* Correlation is significant at p\0.05** Correlation is significant p\0.01

630 Diabetes Ther (2017) 8:625–636

Relationship to DFCSTable 3 displays Pearson’s correlations betweenyouth and parent FCom scores and scores onthe DFCS. No significant relationships emergedbetween mean FCom scores or total conflictscores for parents or youth. However, a positivecorrelation was found between FCom disagree-ment scores and parent and youth total conflictscores, suggesting that greater disagreementabout frequency of communication was associ-ated with more family conflict as reported byboth youth and their parents.

Tone of Communication (TCom)

Scale CharacteristicsTable 1 shows the mean, standard deviation,and correlation with tone for all respective FCIitems, and correlations between parents andyouth ratings for each TCom item, whileSupplementary Table 2 provides the numberor responses to each item. Parent mean TComscores (M = 3.71 ± 0.67) highly correlated withyouth mean scores (M = 3.73 ± 0.63) (r = 0.83,n = 51, p\0.001, r2 = 69.1%). Although asingle-factor TCom scale appears to have facevalidity for parents and youth, conductingEFA and Cronbach’s alpha analysis wasdeemed inappropriate because families whodid not report communicating about a par-ticular topic were unable to endorse whetherany communication on that topic was positiveor negative (i.e., there could not have been aparticular tone of communication if no com-munication occurred on that topic in thegiven time frame). In total 39 youths and 36parents responded to all nine TCom items,with individual items receiving response ratesbetween 50.0% and 92.7% (youth) and45.5–90.9% (parents).

To ensure that parent–youth dyad tonescores could be adequately compared, discrep-ancy scores were only calculated for dyads thatresponded to at least seven of the same TComitems. For the 48 dyads meeting this criterion,parents and youth disagreed by an average ofless than 1 point per TCom item(M = 0.49 ± 0.39, range = 0.0–1.27).

Relationship to HbA1cIn a linear regression model, more positiveyouth TCom scores were associated with lowerHbA1c after controlling for youth age, durationof diabetes, and gender (standardizedb = -0.45, t (64), p\0.001), but no associationswere found for TCom parent mean or discrep-ancy scores. Again, we performed separateregressions to test if individual TCom itemssignificantly predicted HbA1c, after controllingfor youth age, duration of diabetes, and gender.All significant youth and parent TCom itemsindicated that better metabolic control waslinked to more positive tone of communication(see Table 1).

Relationship to DFCSTable 3 also shows Pearson’s correlationsbetween parent/youth DFCS scores and parent/youth mean TCom scores; there were 107 fullycompleted parent/youth surveys. Parent TComscores correlated negatively with both parentand youth conflict scores, suggesting that morepositive communication was associated withless perceived family conflict. This relationshipwas also found between youth TCom scores andparent and youth conflict scores. Finally, TComdisagreement scores were positively related toboth parent and youth conflict scores, suggest-ing that more disagreement about the tone ofcommunication was related to greater perceivedfamily conflict.

Child Self-Management Scale (CSM)

Child self-management items, mean scores, andcorrelations between parents and youth areshown in Supplementary Table 1; there were109 fully completed parent/youth surveys.Cronbach’s alphas for youth (a = 0.66) andparents (a = 0.70) were lower than expected,but similar to previous adherence measures [28].In separate models controlling for youth age,duration of diabetes, and gender, mean CSMscore was positively associated with HbA1c forboth youth (standardized b = 0.44,t (108) = 4.87, p\0.001) and parents (stan-dardized b = 0.36, t (108) = 3.71, p\0.001).

Diabetes Ther (2017) 8:625–636 631

This suggests a relationship between more par-ent/youth-reported frequency of lapses in dia-betes management behaviors and suboptimalyouth diabetes control. Greater disagreementbetween parents and youth on the CSM alsocorresponded with higher HbA1c (standardizedb = 0.30, t (106) = 3.22, p = 0.002).

Table 3 shows Pearson’s correlations betweenthe CSM and FCom/TCom scores. There was asignificant negative association between parentmean FCom scores and youth mean CSM scores,suggesting that more frequent parent reportedcommunication was related to fewer misseddiabetes management behaviors. Greater dis-agreement on the FCom scale positively relatedto mean parent, mean youth, and disagreementscores on the CSM scale. This suggests that moredisagreement between youth and parents aboutfrequency of communication coincided withincreased frequency of lapses in diabetes man-agement behaviors, and greater discrepancybetween youth and parents about how oftenlapses occurred. Finally, higher mean parent,mean youth, and disagreement scores on theCSM were associated with more negative tone asreported by parents and youth.

Full Multivariate Model Predicting HbA1c

As a final exploratory analysis, hierarchicalmultiple regression was used to test adherence(measured through the CSM) as a potentialmediator for the relationship between youthmean TCom and FCom disagreement scoreswith HbA1c. There was evidence of collinearitybetween parent- and youth-reported CSMscores; therefore, only youth mean CSM scoreswere used in the analysis (results were similarusing parent mean scores). Entering age, dura-tion of diabetes, gender, and mean youth CSMscores yielded a significant regression model,F (4, 55) = 4.98, p = 0.002, which accounted for26.6% of the variance in HbA1c. Adding youthTCom and FCom disagreement scores explainedan additional 7.2% of the variance, with thischange trending toward significance, F (2,53) = 2.90, p = 0.064. In the full multivariatemodel, FCom disagreement scores were notsignificantly associated with HbA1c

(standardized b = 0.090, t (59) = 0.72,p = 0.477), while youth mean CSM (standard-ized b = 0.35, t (59) = 2.91, p = 0.005) andyouth mean TCom (standardized b = -0.28,t (59) = -2.21, p = 0.031) scores related signifi-cantly to adolescent glycemic control.

DISCUSSION

This is the first study to examine the frequencyand tone of communication between youth andtheir parents on various diabetes managementbehaviors. Contrary to the primary hypothesis,no associations were found between overallfrequency of communication and glycemiccontrol. Additionally, there were no significantassociations between HbA1c and parent-re-ported tone of communication or parent–youthdisagreement about tone of communication.Instead, results indicated that HbA1c was sig-nificantly related to the disagreement betweenyouth- and parent-reported frequency of com-munication, with greater disagreement beingassociated with poorer glycemic control. HbA1cwas also significantly associated with tone ofcommunication as reported by the adolescent,where more negative tone correlated withhigher HbA1c. These findings are consistentwith previous research on diabetes-relatedfamily discrepancies in perceptions aboutresponsibility and communication for aspectsof diabetes management [2, 16, 17, 26]. Afterincluding adherence in the regression model,FCom disagreement scores no longer con-tributed significantly to HbA1c. This findingsuggests that discrepancies about frequency ofcommunication may reflect the presence of lessthan ideal family communication that con-tribute to more lapses in diabetes managementbehaviors which in turn contributes to subop-timal glycemic control. This suggestion is simi-lar to the findings in the diabetes family conflictliterature, wherein greater family conflict indi-rectly contributed to poorer glycemic control bydecreasing self-care behaviors [7, 29]. From aclinical perspective, widely discordant views ondiabetes-related issues between parents andyouth may signal to the provider that inter-ventions promoting adaptive communication

632 Diabetes Ther (2017) 8:625–636

and conflict resolution skills may be of partic-ular benefit [10, 30]. Annual administration of adiabetes communication survey may also helpproviders screen for and track parent–youthcommunication discrepancies to help deter-mine if additional, more intensive interventionis needed.

This study also found no effect of parent-and youth-reported frequency of communica-tion on glycemic control and adherence. Onepossible explanation is that although raw fre-quency of communication may reflect oneaspect of parental involvement, it is necessaryto know whether that involvement tends to bepositive or negative [14]. For example, someyouth perceive their parents’ frequent commu-nication as ‘‘nagging’’ [31], while others per-ceive their parents, specifically fathers, to beunder-involved, despite efforts to include themin treatment discussions [32]. Consistent withthese findings, results from the present studyindicate that frequency of communicationgenerally did not correlate with the tone ofcommunication, with the exception of one item(discussed missed insulin boluses), wheregreater communication about missed boluseswas associated with more negative tone (seeTable 1). In cases of mismatched parent-youthcommunication, research shows increased riskof suboptimal diabetes management [16, 33].

It is also plausible that frequency of com-munication is differentially related to parame-ters of glycemic control depending onindividual-specific diabetes managementbehaviors. As an additional exploratory itemsanalysis to test this hypothesis, we performedseparate regressions to test if individual FComitems significantly predicted HbA1c, after con-trolling for youth age, duration of diabetes, andgender (see Table 1). When examining FComitems, results indicated that more parent- andyouth-reported communication about hypo-glycemia (e.g., discussed particular low bloodsugar; discussed frequency/number of lowblood sugars) corresponded to lower HbA1cvalues. In contrast, increased communicationabout hyperglycemia as reported by youth (e.g.,discussed missed insulin boluses; discussed fre-quency/number of high blood sugars) wasassociated with higher HbA1c. These findings

suggest that parents and youth tend to com-municate most frequently about issues mostrelevant to the way in which youth managetheir diabetes. That is, when youth maintaintighter glycemic control (and thus are at ahigher risk for hypoglycemic events) par-ent–youth dyads communicate more frequentlyabout low BG values. However, when youth arein suboptimal control, high BG values andinsulin dosing are discussed more frequently. Ifthis is the case, increasing the overall frequencyof communication is likely inadequate to fullyaddress problem areas in diabetes managementwithout examining all of these self-manage-ment factors.

Finally, this study assessed the tone of com-munication about diabetes management. Con-sistent with our hypothesis, lower HbA1c valueswere associated with more youth-perceivedpositive communication beyond the effects ofadherence. However, there was no associationbetween glycemic control and parent-perceivedtone or discrepancy between parents and youthon tone of communication. Past research gen-erally supports these findings for youth-re-ported family communication behaviors, withmost studies demonstrating improved glycemiccontrol with increased positive communicationbehaviors (i.e., support, love, acceptance,warmth) [15, 17, 34, 35], and poorer glycemiccontrol with more youth-reported negativecommunication styles (i.e., anger, interruption,unsupportiveness) [17, 22, 36]. There is someevidence that mother-reported positive com-munication behaviors are associated with betterglycemic control [35], and negative behaviorsare related to worsened control [36], but thesefindings are inconsistent [34, 35]. No identifiedresearch has examined the relationship betweendiscrepancies in tone of communication andglycemic control. The non-significant finding inthis study may be due to the limited power toexplore discrepancies in parent–youth dyads fortone, but future research will need to establish ifthis is the case.

Although the results of this study are infor-mative, there are several methodological limi-tations. As a result of the anonymity of thesurvey, we were unable to identify whether themother or father participated, limiting our

Diabetes Ther (2017) 8:625–636 633

ability to draw more specific conclusions aboutdifferences in communication. We also lackinformation regarding other demographiccharacteristics of the participants and howmany declined the survey, limiting generaliz-ability of the findings. Additionally, the ques-tionnaires selected for this study were limited topromote efficiency and to reduce participantburden. Consequently, additional relevantcommunication concepts, such as parentalmonitoring [37, 38], and family responsibility[26] were not assessed. While the CSM and FCIhad thus not previously been validated, we tookmultiple steps to maintain rigor in their for-mulation and assessment, described below. TheCSM assessed diabetes care behaviors and, asseen in other validated behavior scales, exhib-ited acceptable reliability and correlated signif-icantly with HbA1c levels [17]. The FCI assessedconcepts for the most part not covered by othertools but was formulated by consulting otherfamily diabetes scales to inform the manner ofassessment items. We noted a high correlationbetween TCom and a previously validated tool,the Family Conflict Scale [11]. We further eval-uated Cronbach alpha scores for the FComscale, which appeared to measure a single factorwith high internal reliability. Nevertheless, thepsychometric properties of the TCom scalecould not be fully evaluated given that parentsand youth did not report communicating aboutcertain diabetes management behaviors withinthe designated time frame (i.e., 1 week), becauseof questions regarding tone of communicationwere not relevant to families not reportingrecent communication, only 33% of partici-pants having data from all questions regardingthe tone of communication—rendering inade-quate power for Cronbach alpha analyses.Future iterations of the inventory should con-sider larger sample sizes or assessing tone/fre-quency over a longer time period (i.e., 1 month)or implementing a more general measure oftone for each item (i.e., ‘‘when you talk aboutthis topic, is communication positive or nega-tive?’’) and performing further validationagainst other external outcomes or—in theabsence of similar tools—with direct observa-tional data.

CONCLUSION

In spite of the limitations, this preliminarystudy highlights the complexity of communi-cation in diabetes management and challengesthe overly broad notion that more communi-cation leads to improved diabetes control. Infact, in the present study, valence of commu-nication as measured by tone and discrepanciesin parent and youth perceptions about howoften they are communicating about diabetesmanagement tasks had higher correlations withoverall glycemic control than frequency ofcommunication, suggesting that these areaswarrant further investigation. In addition,future larger-scale studies are needed to furtherexamine the association between parent–youthdiscrepancies in tone of communication andglycemic control.

ACKNOWLEDGEMENTS

The U.S. National Library of Medicine1R01LM012090 (MDD, RV, DRC, SV) fundedthe study and contributed toward article pro-cessing charges. The remainder of article pro-cessing charges were paid for by the authors. Allnamed authors meet the International Com-mittee of Medical Journal Editors (ICMJE) cri-teria for authorship for this manuscript, takeresponsibility for the integrity of the work as awhole, and have given final approval to theversion to be published. We would also like toacknowledge the support of Jackie Shepard,Jesse Grabman, and Linda Gonder-Frederick fortheir invaluable assistance in the assessment ofthese tools.

Disclosures. Mark D. DeBoer, Rupa Valdez,Daniel R. Chernavvsky, Monica Grover, Chris-tine Burt Solorzano, Kirabo Herbert and StephenPatel have nothing to disclose.

Compliance with Ethics Guidelines. Allprocedures followed were in accordance withthe ethical standards of the responsible com-mittee on human experimentation (institu-tional and national) and with the Helsinki

634 Diabetes Ther (2017) 8:625–636

Declaration of 1964, as revised in 2013.Informed consent was obtained from allpatients for being included in the study.

Data Availability. The datasets during and/or analyzed during the current study are avail-able from the corresponding author on reason-able request.

Open Access. This article is distributedunder the terms of the Creative CommonsAttribution-NonCommercial 4.0 InternationalLicense (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommer-cial use, distribution, and reproduction in anymedium, provided you give appropriate creditto the original author(s) and the source, providea link to the Creative Commons license, andindicate if changes were made.

REFERENCES

1. Gerstl EM, Rabl W, Rosenbauer J, et al. Metaboliccontrol as reflectet by HbA1c in children, adoles-cents and young adults with type-1 diabetes melli-tus: combined longitudinal analysis including27,035 patients from 207 centers in Germany andAustria during the last decade. Eur J Pediatr.2008;167:447–53.

2. Hanna KM, Juarez B, Lenss SS, Guthrie D. Par-ent-adolescent communication and support fordiabetes management as reported by adolescentswith type 1 diabetes. Issues Compr Pediatr Nurs.2003;26:145–58.

3. Wysocki T, Meinhold P, Cox DJ, Clarke WL. Surveyof diabetes professionals regarding developmentalchanges in diabetes self-care. Diabetes Care.1990;13:65–8.

4. Sawyer SM, Aroni RA. Self-management in adoles-cents with chronic illness. What does it mean andhow can it be achieved? Med J Aust.2005;183:405–9.

5. Guilfoyle SM, Crimmins NA, Hood KK. Blood glu-cose monitoring and glycemic control in adoles-cents with type 1 diabetes: meter downloads versusself-report. Pediatr Diabetes. 2011;12:560–6.

6. Campbell M, Schatz D, Chen V, et al. A contrastbetween children and adolescents with excellent

and poor control: the T1D exchange clinic registryexperience. Pediatr Diabetes. 2014;15:110–7.

7. Hilliard ME, Wu YP, Rausch J, Dolan LM, Hood KK.Predictors of deteriorations in diabetes manage-ment and control in adolescents with type 1 dia-betes. J Adolesc Health. 2013;52:28–34.

8. Olinder AL, Kernell A, Smide B. Missed bolus doses:devastating for metabolic control in CSII-treatedadolescents with type 1 diabetes. Pediatr Diabetes.2009;10:142–8.

9. VanderWel BW, Messer LH, Horton LA, et al. Mis-sed insulin boluses for snacks in youth with type 1diabetes. Diabetes Care. 2010;33:507–8.

10. Laffel LMB, Vangsness L, Connell A, Goebel-FabbriA, Butler D, Anderson BJ. Impact of ambulatory,family-focused teamwork intervention on glycemiccontrol in youth with type 1 diabetes. J Pediatr.2003;142:409–16.

11. Anderson BJ, Vangsness L, Connell A, Butler D,Goebel-Fabbri A, Laffel LM. Family conflict, adher-ence, and glycaemic control in youth with shortduration type 1 diabetes. Diabet Med.2002;19:635–42.

12. Vesco AT, Anderson BJ, Laffel LM, Dolan LM,Ingerski LM, Hood KK. Responsibility sharingbetween adolescents with type 1 diabetes and theircaregivers: importance of adolescent perceptions ondiabetes management and control. J Pediatr Psy-chol. 2010;35:1168–77.

13. Ellis DA, Podolski CL, Frey M, Naar-King S, Wang B,Moltz K. The role of parental monitoring in ado-lescent health outcomes: impact on regimenadherence in youth with type 1 diabetes. J PediatrPsychol. 2007;32:907–17.

14. Wysocki T, Taylor A, Hough BS, Linscheid TR,Yeates KO, Naglieri JA. Deviation from develop-mentally appropriate self-care autonomy. Associa-tion with diabetes outcomes. Diabetes Care.1996;19:119–25.

15. Berg CA, King PS, Butler JM, Pham P, Palmer D,Wiebe DJ. Parental involvement and adolescents’diabetes management: the mediating role ofself-efficacy and externalizing and internalizingbehaviors. J Pediatr Psychol. 2011;36:329–39.

16. Cameron FJ, Skinner TC, de Beaufort CE, et al. Arefamily factors universally related to metabolic out-comes in adolescents with type 1 diabetes? DiabetMed. 2008;25:463–8.

17. Lewin AB, Heidgerken AD, Geffken GR, et al. Therelation between family factors and metabolic

Diabetes Ther (2017) 8:625–636 635

control: the role of diabetes adherence. J PediatrPsychol. 2006;31:174–83.

18. Drotar D, Ittenbach R, Rohan JM, Gupta R, PendleyJS, Delamater A. Diabetes management and gly-cemic control in youth with type 1 diabetes: test ofa predictive model. J Behav Med. 2013;36:234–45.

19. Hood KK, Butler DA, Anderson BJ, Laffel LM.Updated and revised diabetes family conflict scale.Diabetes Care. 2007;30:1764–9.

20. Miller-Johnson S, Emery RE, Marvin RS, Clarke W,Lovinger R, Martin M. Parent-child relationshipsand the management of insulin-dependent diabetesmellitus. J Consult Clin Psychol. 1994;62:603–10.

21. Geffken GR, Lehmkuhl H, Walker KN, et al. Familyfunctioning processes and diabetic ketoacidosis inyouths with type I diabetes. Rehabil Psychol.2008;53:231–7.

22. Duke DC, Geffken GR, Lewin AB, Williams LB,Storch EA, Silverstein JH. Glycemic control inyouth with type 1 diabetes: family predictors andmediators. J Pediatr Psychol. 2008;33:719–27.

23. Martin MT, Miller-Jonson S, Kitzmann KM, EmeryRE. Parent-child relationships and insulin-depen-dent diabetes mellitus: observational ratings ofclinically relevant dimensions. J Fam Psychol.1998;12:102–11.

24. Laffel LM, Connell A, Vangsness L, Goebel-Fabbri A,Mansfield A, Anderson BJ. General quality of life inyouth with type 1 diabetes: relationship to patientmanagement and diabetes-specific family conflict.Diabetes Care. 2003;26:3067–73.

25. Weissberg-Benchell J, Nansel T, Holmbeck G, et al.Study SCotFMoD: generic and diabetes-specificparent-child behaviors and quality of life amongyouth with type 1 diabetes. J Pediatr Psychol.2009;34:977–88.

26. Anderson BJ, Holmbeck G, Iannotti RJ, et al. Dyadicmeasures of the parent-child relationship duringthe transition to adolescence and glycemic controlin children with type 1 diabetes. Fam Syst Health.2009;27:141–52.

27. American Diabetes Association. (11) Children andadolescents. Diabetes Care. 2015;38(Suppl):S70–6.

28. Lewin AB, LaGreca AM, Geffken GR, et al. Validityand reliability of an adolescent and parent ratingscale of type 1 diabetes adherence behaviors: theSelf-Care Inventory (SCI). J Pediatr Psychol.2009;34:999–1007.

29. Hilliard ME, Guilfoyle SM, Dolan LM, Hood KK.Prediction of adolescents’ glycemic control 1 yearafter diabetes-specific family conflict: the mediatingrole of blood glucose monitoring adherence. ArchPediatr Adolesc Med. 2011;165:624–9.

30. Wysocki T, Harris MA, Buckloh LM, et al. Ran-domized trial of behavioral family systems therapyfor diabetes: maintenance of effects on diabetesoutcomes in adolescents. Diabetes Care.2007;30:555–60.

31. Weinger K, O’Donnell KA, Ritholz MD. Adolescentviews of diabetes-related parent conflict and sup-port: a focus group analysis. J Adolesc Health.2001;29:330–6.

32. Seiffge-Krenke I. ‘‘Come on, say something, dad!’’:communication and coping in fathers of diabeticadolescents. J Pediatr Psychol. 2002;27:439–50.

33. Wiebe DJ, Berg CA, Korbel C, et al. Children’sappraisals of maternal involvement in coping withdiabetes: enhancing our understanding of adher-ence, metabolic control, and quality of life acrossadolescence. J Pediatr Psychol. 2005;30:167–78.

34. Berg CA, Butler JM, Osborn P, et al. Role of parentalmonitoring in understanding the benefits of par-ental acceptance on adolescent adherence andmetabolic control of type 1 diabetes. Diabetes Care.2008;31:678–83.

35. Wysocki T, Harris MA, Buckloh LM, et al. Ran-domized, controlled trial of behavioral family sys-tems therapy for diabetes: maintenance andgeneralization of effects on parent-adolescentcommunication. Behav Ther. 2008;39:33–46.

36. Iskander JM, Rohan JM, Pendley JS, Delamater A,Drotar D. A 3-year prospective study of parent-childcommunication in early adolescents with type 1diabetes: relationship to adherence and glycemiccontrol. J Pediatr Psychol. 2015;40:109–20.

37. Ellis DA, Templin TN, Podolski CL, Frey MA,Naar-King S, Moltz K. The parental monitoring ofdiabetes care scale: development, reliability andvalidity of a scale to evaluate parental supervisionof adolescent illness management. J AdolescHealth. 2008;42:146–53.

38. Palmer DL, Osborn P, King PS, et al. The structure ofparental involvement and relations to diseasemanagement for youth with type 1 diabetes. J Pe-diatr Psychol. 2011;36:596–605.

636 Diabetes Ther (2017) 8:625–636