Diabetic Ketoacidosis at Diagnosis of Type 1 Diabetes Predicts Poor

Diabetes Care Volume 40, September 2017 1249

Lindsey M. Duca,1,2 Bing Wang,1 Diabetic Ketoacidosis at Diagnosis Marian Rewers,1 and Arleta Rewers3 of Type 1 Diabetes Predicts Poor Long-term Glycemic Control Diabetes Care 2017;40:1249–1255 | https://doi.org/10.2337/dc17-0558

OBJECTIVE This study tested the hypothesis that diabetic ketoacidosis (DKA) at diagnosis of type 1 diabetes in children predicts poor long-term glycemic control independently of established risk factors.

RESEARCH DESIGN AND METHODS This was a prospective cohort study of 3,364 Colorado residents diagnosed with type 1 diabetes before 18 years of age, in 1998–2012, and monitored for up to 15 years. Of those, 1,297 (39%) had DKA at diagnosis (blood glucose >250 mg/dL, and venous pH <7.3 or bicarbonate <15 mEq/L). Severity of DKA was further classi- ﬁed as mild/moderate (pH 7.10–7.29 or bicarbonate 5–14 mEq/L) or severe (pH <7.10 or bicarbonate <5 mEq/L). HbA1c levels were measured an average of 2.8 times/year (median 20 HbA1c values/patient). A linear mixed model was used to examine the effect of DKA on long-term HbA1c levels, adjusting for age, race/ethnicity, sex, family history of diabetes, health insurance, and insulin pump use.

RESULTS

DKA at diagnosis predicted persistently elevated HbA1c levels. Compared with chil- RISK METABOLIC AND CARDIOVASCULAR dren without DKA, HbA1c tracked 1.4% (15.3 mmol/mol) higher in those with severe DKA (P < 0.0001) and 0.9% (9.8 mmol/mol) higher in those with mild/moderate DKA at diagnosis (P < 0.0001). These effects were independent of ethnic minority status or lack of health insurance at diagnosis that predicted higher HbA1c by 0.5% (5.5 mmol/mol; P < 0.0001) and 0.2% (2.2 mmol/mol; P < 0.0001), respectively. Insulin pump use or 1Barbara Davis Center for Diabetes, School of having a parent or sibling with type 1 diabetes predicted lower long-term HbA1c by, P < P Medicine, University of Colorado, Aurora, CO respectively, 0.4% (4.4 mmol/mol; 0.0001) and 0.2% (2.2 mmol/mol; = 0.01). 2Colorado School of Public Health, University of Colorado, Aurora, CO CONCLUSIONS 3Department of Pediatrics, School of Medicine, DKA at diagnosis of type 1 diabetes in children predicts poor long-term glycemic University of Colorado, Aurora, CO control, independent of demographic and socioeconomic factors. Corresponding author: Arleta Rewers, arleta. [email protected]. Received 18 March 2017 and accepted 5 June Diabetic ketoacidosis (DKA) is a life-threatening complication affecting 30–46% of chil- 2017. dren with newly diagnosed type 1 diabetes in the U.S. (1,2). Although mortality is This article contains Supplementary Data online currently ,1%, DKA is still associated with detrimental neurocognitive outcomes at http://care.diabetesjournals.org/lookup/ (3,4). DKA is largely preventable because it often results from delayed care caused suppl/doi:10.2337/dc17-0558/-/DC1. by low awareness of diabetic symptoms in the community and among health care © 2017 by the American Diabetes Association. providers (5–7). Readers may use this article as long as the work b is properly cited, the use is educational and not Type 1 diabetes results from the autoimmune destruction of -cells in the pancreas for proﬁt, and the work is not altered. More infor- and the subsequent lack of insulin. Severe hyperglycemia and systemic inﬂammation mation is available at http://www.diabetesjournals associated with DKA have been shown to further deplete functional pancreatic islets (8). .org/content/license. 1250 DKA at Diagnosis and Long-term Glycemic Control Diabetes Care Volume 40, September 2017

Lower residual b-cell function (9), 105) and those for whom fewer than compare categorical data. Individual aver- higher insulin requirements (10), and two HbA1c measurements were available age HbA1c levels were calculated annu- worse glycemic control (11) during the (n = 75). These exclusions left 3,364 chil- ally, starting 60 days after the diagnosis following 1–2 years have been shown in dren with type 1 diabetes for the study through the follow-up. Pearson correla- children with DKA at presentation com- population of interest. tion coefficients were used to examine pared with those diagnosed with milder Demographic, insurance status, and the correlation and covariance matrices symptoms. Early tight glycemic control, as clinical characteristic data were extracted for the repeated measures of HbA1c level measured by hemoglobin A1c (HbA1c), from medical records. Patients were over time. prevents diabetic complications (12–14), monitored at the BDC, which serves A linear mixed model was used to esti- while loss of residual b-cell function youth with diabetes in Colorado. The mate the mean change in the HbA1c levels increases the risk of retinopathy, ne- BDC is the default ambulatory service over time by DKA status (yes/no) to ac- phropathy, neuropathy, and severe hypo- for the Children’s Hospital of Colorado count for the inherent correlation of re- glycemia (15). DKA may worsen glycemic Network of Care (CHCO), which provides peated measures on the same individual control and prognosis of diabetes through more than 60% of general pediatric care over time. The first model was unad- exacerbation of b-cell loss at diagnosis and more than 90% of pediatric subspe- justed, and the fixed effect of the model independently of other factors such cialty care in Colorado. A population- consisted of time (yearly), DKA status as age and intensity of postdiagnosis based registry of childhood diabetes, (yes/no), and an interaction of DKA status management. Search for Diabetes in Youth (SEARCH), by time. A second unadjusted linear Despite advances in medical manage- has shown that 82% of all Colorado chil- mixed model was used to estimate ment, type 1 diabetes remains a burden- dren with diabetes and 84% of those the mean change in the HbA1c level over some disease with everyday challenges with type 1 diabetes diagnosed between time by DKA severity (no, mild/moderate, and substantial medical costs. In addition, 2002 and 2012 were seen at the BDC. This severe). A final linear mixed model was most youth with type 1 diabetes do not pattern has not changed during the past used to estimate the mean change meet the American Diabetes Association 15 years; thus, the data presented in this in the HbA1c level over time by DKA se- guidelines for glycemic control (16) and analysis are representative of the state verity category (no, mild/moderate, se- are likely to experience excess morbidity of Colorado. vere), adjusting for age at diagnosis, and premature mortality (17–19).The ob- Race/ethnicity was based on parental race/ethnicity, sex, family history of dia- jective of this study was to test the hy- report defined using the 2000 U.S. Census. betes, insurance status, and insulin pump pothesis that children who present in DKA Because only 16% of our participants were use. at the diagnosis of type 1 diabetes expe- Hispanic, 4% non-Hispanic Black, and 4% For all linear mixed models performed, rience worse glycemic control later in the other, these race categories were com- model assumptions, such as a normally course of the disease, independently of bined, and subsequently in the analyses distributed outcomemeasure,were demographic and socioeconomic factors race was dichotomized into non-Hispanic checked. All models were further as- or access to care that are associated with whiteversusother(nonwhiteorHis- sessed to verify there was in fact a linear both onset DKA and long-term diabetes panic). Insurance status before diagnosis trend over time and that a quadratic or care and glycemic control. If confirmed as of diabetes was categorized as 1)private spline term did not significantly improve an independent predictor of glycemic insurance, 2) government-provided insur- the model fit. Finally, transformed resid- control, DKA should be seen as a prevent- ance (Medicaid, TRICARE, Colorado Resi- uals, using the Cholesky decomposition able risk factor for long-term diabetic dent Discount Program, or Child Health technique, were examined to assess the complications. Plan Plus), or 3) none. HbA1c levels were adequacy of the fitted models and addi- measured at baseline and also at each tionally to look for an indication of any RESEARCH DESIGN AND METHODS visit (2.8 visits/year, on average; median outliers. Because the mixed model analy- Study Population 20 visits/patient) using the DCA 2000+ ses assume data are missing at random, The study cohort consists of 3,364 Colorado (Bayer). the missing data pattern was checked, residents aged 0–17 years diagnosed with Standard criteria (20) were used to de- and the assumption was verified by plot- type 1 diabetes between 1998 and fine DKA (blood glucose .250 mg/dL, and ting the means over time stratified by the 2012 and monitored at the Barbara Davis venous pH ,7.3 or bicarbonate ,15 time of the last measure completed. A Center for Diabetes (BDC) for at least mEq/L). Severity of DKA was further clas- P value of ,0.05 was considered statisti- 1 year and up to 17 years, until 8 April sified as mild/moderate (pH 7.10–7.29 or cally significant for all analyses. The sta- 2016. Type 1 diabetes was diagnosed bicarbonate 5–14 mEq/L) or severe tistical analysis was performed using by a physician based on the clinical phe- (pH ,7.10 or bicarbonate ,5mEq/L). SAS 9.3 software of the SAS System for notype and measurement of islet auto- The Colorado Institutional Review Board Windows. antibodies. Of the 3,926 potential study (COMIRB) approved the protocol and participants, excluded were children with granted a waiver of informed consent. RESULTS type 2 diabetes (n = 238), monogenic di- DKA at diagnosis was present in 1,297 abetes (n = 13), diabetes secondary to Statistical Analysis of the study participants (38.6%). The cystic fibrosis (n = 72) or other conditions A Student t test was used to compare baseline characteristics of the study pop- (n = 40), and indeterminate diabetes (n = continuous variables at baseline between ulation stratified by DKA status are reported 19) (2). Also excluded were children with the children who presented with DKA and in Table 1. Children who presented in unknown DKA status at diagnosis (n = those without DKA. A x2 test was used to DKA were younger, more often nonwhite care.diabetesjournals.org Duca and Associates 1251

, Table 1—Baseline characteristics of the study population 0.4% (4.4 mmol/mol; P 0.0001) and Missing DKA No DKA 0.2% (2.2 mmol/mol; P = 0.01), respec- Variable n (%) n = 1,297 n = 2,067 P value tively. Family history of diabetes other than type 1 diabetes in a parent or sibling Age at onset (years)* 0 (0) 8.8 6 4.6 9.4 6 4.3 ,0.0001 did not predict a change in long-term Sex 0 (0) 0.16 HbA levels. Female 585 (45.1) 956 (46.2) 1c Male 712 (54.9) 1,111 (53.8) The dose-response relationship be- Ethnicity 4 (0.1) ,0.0001 tween severity of DKA and average HbA1c Non-Hispanic white 897 (69.3) 1,638 (79.3) during the follow-up was confirmed. Spe- Nonwhite or Hispanic 398 (30.7) 427 (20.7) cifically, there was an increase in HbA1c Insurance status 22 (0.7) ,0.0001 levels over time of 0.9% (9.8 mmol/mol) Private 743 (57.9) 1,535 (74.6) for those diagnosed in mild/moderate Government 390 (30.4) 401 (19.5) DKA and a 1.4% (15.3 mmol/mol) in- None 151 (11.7) 122 (5.9) crease for those in severe DKA. The Location of residence 0 (0) 0.11 long-term mean HbA1c levels adjusted Rural 122 (9.4) 189 (9.2) for age, race/ethnicity, sex, insurance Urban 440 (33.9) 633 (30.6) Metropolitan 735 (56.7) 1,245 (60.2) status, family history of diabetes, and insulin pump use (Supplementary Fig. 1) Family history 43 (1.3) ,0.0001 First-degree relative with diabetes 58 (4.5) 287 (14.0) were similar to unadjusted patterns Family history of diabetes (Fig. 2). Yes 576 (45.3) 879 (42.9) No 638 (50.2) 883 (43.1) CONCLUSIONS Insulin pump use 0 (0) 471 (36.3) 844 (40.8) 0.009 The major novel finding of this study is At type 1 diabetes diagnosis* 410 (12) the sustained negative effect of DKA at 6 6 , HbA1c (%) 12.1 2.0 10.8 2.5 0.0001 diagnosis of type 1 diabetes in children 6 6 HbA1c (mmol/mol) 109 21.9 95 27.3 on glycemic control during the follow- At 60 days post type 1 diabetes ing 15 years. Importantly, this effect is diagnosis* 278 (8.3) independent of demographic and socio- HbA (%) 8.4 6 1.7 8.0 6 1.5 ,0.0001 1c economic factors (younger age, ethnic HbA1c (mmol/mol) 68 6 18.6 65 6 16.4 minority status, male sex) or access to DKA severity 256/1,297 (19.7) d Mild or moderate 693 (66.6) d care barriers that are associated with Severe 348 (33.4) d both DKA and long-term glycemic control. The dose-response effect of DKA *Student t test was used to compare continuous variables, which are presented as mean 6 SD; otherwise, a x2 test was used to compare categorical data, which are presented as n (%). severity on future poor glycemic control strengthens the evidence for causality. Therefore, DKA at diagnosis of type 1 diabetes is not just an acute complica- or Hispanic, more often uninsured or groups, consistent with gradual loss of tion but is also a harbinger of increased covered by a government-provided in- endogenous insulin secretion reported morbidity and mortality associated with surance plan, and less likely to have both in type 1 and type 2 diabetes. poor glycemic control. Consequently, first-degree relatives with type 1 diabe- effective prevention of DKA at diagnosis tes than children presenting without Multivariate Analysis may provide enduring benefits. Future DKA. As expected, HbA1c levels at diag- DKA at diagnosis predicted elevated lon- studies are warranted to assess the ef- nosis of type 1 diabetes, as well as 60 days gitudinal HbA1c levels independent of fectiveness of DKA prevention at type 1 after the diagnosis, were higher in chil- other factors in a multivariate mixed- diabetes onset on improving long-term dren diagnosed in DKA. Among children effects model (Table 2). Compared with glycemic control. who presented in DKA, one-third had se- children without DKA, HbA1c tracked 1.4% Our study confirmed previously known vere DKA. (15.3 mmol/mol) higher in those with risk factors for DKA at diagnosis of child- , Longitudinal HbA1c levels upto15 years severe DKA (P 0.0001) and 0.9% hood type 1 diabetes (Table 1). Children after type 1 diabetes diagnosis are shown (9.8 mmol/mol) higher in those with that presented in DKA were younger than by DKA status in Fig. 1 and by severity mild/moderate DKA (P , 0.0001). These those who presented with milder symp- of DKA in Fig. 2. Throughout the entire effects were independent of ethnic mi- toms, and younger age predicted higher study period, HbA1c levels in children nority status or lack of health insurance HbA1c over time (0.5% [5.5 mmol/mol] presenting in DKA remained 0.3–1.0% at diagnosis, which predicted higher per 10 years in age at diagnosis), similar (3.3–10.9 mmol/mol) higher than those HbA1c by 0.5% (5.5 mmol/mol; P , to previous studies (21,22). Consistent diagnosed with milder symptoms. There 0.0001) and 0.2% (2.2 mmol/mol; P , with other studies (1,23), we found a was an apparent dose-response relation- 0.0001), respectively. The effect of DKA high percentage of children diagnosed in ship between the severity of DKA and the was also independent of insulin pump use DKA were nonwhite or Hispanic and had average HbA1c during the entire follow-up or having a parent or sibling with type 1 government-provided or no medical in- period. HbA1c gradually increased in all diabetes, which predicted lower HbA1c by surance. These associations are likely 1252 DKA at Diagnosis and Long-term Glycemic Control Diabetes Care Volume 40, September 2017

control in children and adolescents with type 1 diabetes (26). Several studies have demonstrated that early glycemic control is of para- mount importance to the prevention of diabetic complications (12,13). The Dia- betes Control and Complications Trial (DCCT) demonstrated that the mean HbA1c level during the trial was the over- riding predictor of retinopathy progression and that even what appeared to be minimal changes in HbA1c levels actually translated into large reductions in the risk of complications (27–29). More specifically, the DCCT study showed that a 10% reduction in HbA1c was associated with a 44% lower risk of retinopathy progression and that this relationship ap- plied over the range of HbA1c values (27). A 1.4% (15.3 mmol/mol) difference in HbA1c levels observed in the current study between the severe DKA group and no DKA group, at the HbA1c of 9.5% (80 mmol/mol), translates to a 13% reduction in HbA1c and an ;50% reduction of retinopathy. Exacerbated loss of b-cells is the most plausible mediator of the powerful effect Figure 1—DKA at diagnosis of children with type 1 diabetes and long-term glycemic control. The of DKA at diagnosis on long-term glycemic numbers reported below the figure represent the number of participants contributing the overall control. Severe hyperglycemia and sys- 6 HbA1c level during that point in time. Data presented are mean SE from unadjusted linear mixed temic inflammation associated with DKA model. augments damage from prolonged autoimmune destruction and further depletes multifactorial and inherently incorporate adecreaseinHbA1c level of 0.6% functional pancreatic islets (8). This leads language barriers, parental education, so- (6.6 mmol/mol) from 1998 to 2005 (24) to lower residual b-cell function (9) and cioeconomic status, and cultural inequal- and an even larger reduction of 1.5% worse glycemic control despite higher ities. Although adjusting for insurance (16.4 mmol/mol) from 2000 to 2011 insulin requirements (10). Children who status and insulin pump treatment may (25). The rate of DKA was found to be experience DKA can exhibit long-term not fully control for disparities in access the lowest among children or siblings of cognitive complications, which in turn to care, the BDC provides the same level people with type 1 diabetes, which illus- adversely influence their ability to engage of diabetes management to all children, trates the importance of parental aware- in self-care. Ghetti et al. (3) assessed regardless of the family’s ability to pay. ness of signs and symptoms of diabetes. memory deficits in 33 children with The Center provides, as needed, free di- However, having a first-degree relative type 1 diabetes who suffered at least abetes care supplies (Helping Hands by was not an independent predictor of one episode of DKA and found that chil- the Children’s Diabetes Foundation) as HbA1c in children diagnosed without dren with DKA history had a significantly well as free transportation and parking. DKA in this study (P = 0.22) and was found lower ability to recall events in associa- Since 2002, the BDC Hispanic/Latino to increase post diagnosis HbA1c in Danish tion with specific details, as tested by Childhood Diabetes Program has pro- children who have a low prevalence of event-color and event-spatial position vided bilingual culturally sensitive diabe- DKA at diagnosis (11). We have shown associations. Furthermore, when cogni- tes care service for Spanish-speaking and predictors of higher HbA1c over time, in- tive problems persist into late adoles- bilingual families. Therefore, it is unlikely dependent of DKA status at onset, include cence, there is evidence indicating that disparate access to care could explain increased age, female sex, minority race/ greater risk for poor diabetes manage- the observed sustained differences in ethnicity, and being on government in- ment during early adulthood (30,31), HbA1c over time. surance or having no insurance compared which is in turn associated with chronic In support of the observed reduction with private insurance. These results hyperglycemia (32). in HbA1c levels from insulin pump use highlight lifestyle factors other than DKA Development of islet autoantibodies (continuous subcutaneous insulin infu- thatcanbetargetedtoreducepoorgly- occurs before clinical diagnosis of type 1 sion) shown in this study, previous studies cemic control over time. In addition, all of diabetes, making type 1 diabetes a pre- performed in similarly aged cohorts these factors have been shown to affect dictable disease in an individual with two with comparable follow-up time found social contributors of poor glycemic or more autoantibodies (33). Screening care.diabetesjournals.org Duca and Associates 1253

be effective and feasible in preventing DKA (36), highlighting that DKA is a mod- ifiable risk factor that can be targeted to improve glycemic control over time in children with type 1 diabetes. The current study has inherent strengths and limitations. The main strength of this study is that it was performed in a large cohort of children with type 1 diabetes that have been monitored for a long time using standardized methods. In addition, this study is externally valid in that it is generaliz- able to the entire state of Colorado. How- ever, there are some limitations because we do not have reliable data on insulin analogs or insulin dose over time. It would be of interest to examine changes in insulin dose– adjusted HbA1c over time. To ultimately prove the postulated effect of DKA on residual endogenous insulin secretion, se- rial C-peptide measurements will be needed in future studies. It is possible there are metabolic or physiologic differences in the rate or severity of the b-cell damage that then lead to the development of DKA and consequently also ac- count for the poorer long-term glycemic control. For example, differences have been observed in complement levels; Figure 2—DKA severity at diagnosis of children with type 1 diabetes and long-term glycemic control. specifically, higher plasma complement fi The numbers reported below the gure represent the number of participants contributing the C4A protein concentrations were associ- overall HbA1c level during that point in time. Data presented are mean 6 SE from unadjusted linear mixed model. ated with higher levels of stimulated C-peptide at 1 month post diabetes diag- children for the presence of islet autoan- family history of diabetes (34,35). General nosis, whereas those with higher C4B lev- tibodies and close monitoring of those population screening for islet autoanti- els were less likely to retain the capability who are positive can prevent 80–90% of bodies, combined with a community tosecreteendogenousinsulinat9months DKA, especially among children with no awareness campaign, has been shown to post diagnosis (37). Finally, we have not

Table 2—Predictors of longitudinal HbA1c levels during follow-up for up to 15 years in children with type 1 diabetes*

Predictor of HbA1c HbA1c % (mmol/mol) estimate HbA1c % (mmol/mol) 95% CI P value DKA at diagnosis None Ref. dd Mild/moderate 0.87 (9.5) 0.71–1.03 (7.8–11.3) ,0.0001 Severe 1.35 (14.8) 1.23–1.47 (13.4–16.1) ,0.0001 Age at diagnosis (years) 0.05 (0.5) 0.04–0.06 (0.4–0.7) ,0.0001 Male sex 20.17 (21.9) 20.25 to 20.08 (22.7 to 20.9) 0.0001 Ethnicity Non-Hispanic white Ref. dd Other 0.48 (5.2) 0.38–0.59 (4.2–6.4) ,0.0001 Insurance status Private Ref. dd Government 0.20 (2.2) 0.03–0.37 (0.3–4.0) 0.02 None 0.22 (2.4) 0.11–0.33 (1.2–3.6) ,0.0001 Insulin pump use (yes) 20.41 (24.5) 20.50 to 20.32 (25.5 to 23.5) ,0.0001 Family history No family history of diabetes Ref. dd Parent or sibling with type 1 diabetes 20.19 (22.1) 20.34 to 20.05 (23.7 to 20.5) 0.01 Other family history of diabetes 20.04 (20.4) 20.13 to 0.05 (21.4 to 0.5) 0.39

*Results show either an increase or decrease in mean HbA1c levels associated with the predictor, from a linear mixed-effects model, with all predictors analyzed simultaneously. 1254 DKA at Diagnosis and Long-term Glycemic Control Diabetes Care Volume 40, September 2017

adjusted the analyses for recurrent DKA 7. Baldelli L, Flitter B, Pyle L, et al. A survey of 21. Helgeson VS, Siminerio L, Escobar O, Becker that occurs more often in patients initially youth with new onset type 1 diabetes: opportu- D. Predictors of metabolic control among adoles- diagnosed with DKA and may further de- nities to reduce diabetic ketoacidosis. Pediatr Di- cents with diabetes: a 4-year longitudinal study. abetes 11 October 2016 [Epub ahead of print]. J Pediatr Psychol 2009;34:254–270 teriorate residual insulin secretion. https://doi.org/10.1111/pedi.12455 22. Urbach SL, LaFranchi S, Lambert L, Lapidus JA, In summary, this study demonstrated 8. Atkinson MA, Bluestone JA, Eisenbarth GS, Daneman D, Becker TM. Predictors of glucose that DKA (mild/moderate or severe) in chil- et al. How does type 1 diabetes develop? The control in children and adolescents with type 1 dren diagnosed with type 1 diabetes was notion of homicide or b-cell suicide revisited. Di- diabetes mellitus. Pediatr Diabetes 2005;6: abetes 2011;60:1370–1379 69–74 associated with higher long-term HbA1c 9. Fernandez Castaner~ M, Montana~ E, Camps I, 23. Mallare JT, Cordice CC, Ryan BA, Carey DE, levels during follow-up compared with et al. Ketoacidosis at diagnosis is predictive of Kreitzer PM, Frank GR. Identifying risk factors the children who were not in DKA when lower residual beta-cell function and poor meta- for the development of diabetic ketoacidosis in diagnosed with type 1 diabetes. Lower bolic control in type 1 diabetes. Diabetes Metab new onset type 1 diabetes mellitus. Clin Pediatr – – residual insulin secretion in children 1996;22:349 355 (Phila) 2003;42:591 597 10. Bowden SA, Duck MM, Hoffman RP. Young 24. de Beaufort CE, Swift PG, Skinner CT, et al.; presenting with DKA may be at fault children (,5 yr) and adolescents (.12 yr) with Hvidoere Study Group on Childhood Diabetes and could be prevented with earlier type 1 diabetes mellitus have low rate of partial 2005. Continuing stability of center differences diagnosis. remission: diabetic ketoacidosis is an important in pediatric diabetes care: do advances in diabetes risk factor. Pediatr Diabetes 2008;9:197–201 treatment improve outcome? The Hvidoere Study 11. Fredheim S, Johannesen J, Johansen A, et al.; Group on Childhood Diabetes. Diabetes Care Danish Society for Diabetes in Childhood and Ad- 2007;30:2245–2250 Acknowledgments. The study was performed at olescence. Diabetic ketoacidosis at the onset of 25. Dovc K, Telic SS, Lusa L, et al. Improved met- the Barbara Davis Center for Diabetes in Denver, type 1 diabetes is associated with future HbA1c abolic control in pediatric patients with type 1 di- ’ CO, and at the Children s Hospital Colorado. levels. Diabetologia 2013;56:995–1003 abetes: a nationwide prospective 12-year time Funding. Support for the study was provided by 12. Rudberg S, Ullman E, Dahlquist G. Relation- trends analysis. Diabetes Technol Ther 2014;16: ’ the Children s Diabetes Foundation and the Ri- ship between early metabolic control and the de- 33–40 chard S. Abrams Endowment for Clinical Research. velopment of microalbuminuriada longitudinal 26. Delamater AM, de Wit M, McDarby V, Malik J, fl Duality of Interest. No potential con icts of in- study in children with type 1 (insulin-dependent) Acerini CL; International Society for Pediatric and terest relevant to this article were reported. diabetes mellitus. Diabetologia 1993;36:1309– Adolescent Diabetes. ISPAD Clinical Practice Con- AuthorContributions.L.M.D. researched data, 1314 sensus Guidelines 2014. Psychological care of chil- analyzed data, and wrote the manuscript. B.W. 13. Svensson M, Eriksson JW, Dahlquist G. Early dren and adolescents with type 1 diabetes. was involved in data management. M.R. and glycemic control, age at onset, and develop- Pediatr Diabetes 2014;15(Suppl. 20):232–244 A.R. designed the study, researched data, con- ment of microvascular complications in child- 27. The Diabetes Control and Complications Trial tributed to the discussion, and reviewed the hood-onset type 1 diabetes: a population-based Research Group. The relationship of glycemic ex- manuscript. A.R. is the guarantor of this work study in northern Sweden. Diabetes Care 2004; posure (HbA1c) to the risk of development and and, as such, had full access to all the data in 27:955–962 progression of retinopathy in the Diabetes Control the study and takes responsibility for the in- 14. Diabetes Control and Complications Trial Re- and Complications Trial. Diabetes 1995;44:968– tegrity of the data and the accuracy of the data search Group. Effect of intensive diabetes treat- 983 analysis. ment on the development and progression of 28. Diabetes Control and Complications Trial Prior Presentation. Parts of this study were long-term complications in adolescents with (DCCT)/Epidemiology of Diabetes Interventions fi presented at the 76th Scienti c Sessions of the insulin-dependent diabetes mellitus: Diabetes and Complications (EDIC) Research Group; Lachin American Diabetes Association, New Orleans, Control and Complications Trial. J Pediatr 1994; JM, White NH, Hainsworth DP, et al. Effect of in- – LA, 10 14 June 2016. 125:177–188 tensive diabetes therapy on the progression of 15. Steffes MW, Sibley S, Jackson M, Thomas W. diabetic retinopathy in patients with type 1 dia- References b-Cell function and the development of diabetes- betes: 18 years of follow-up in the DCCT/EDIC. 1. Rewers A, Klingensmith G, Davis C, et al. Pres- related complications in the Diabetes Control Diabetes 2015;64:631–642 ence of diabetic ketoacidosis at diagnosis of and Complications Trial. Diabetes Care 2003;26: 29. Lachin JM, Genuth S, Nathan DM, Zinman B, diabetes mellitus in youth: the Search for Diabe- 832–836 Rutledge BN; DCCT/EDIC Research Group. Effect tes in Youth Study. Pediatrics 2008;121:e1258– 16. Wood JR, Miller KM, Maahs DM, et al.; T1D of glycemic exposure on the risk of microvascular e1266 Exchange Clinic Network. Most youth with type 1 complications in the diabetes control and compli- 2. Rewers A, Dong F, Slover RH, Klingensmith GJ, diabetes in the T1D Exchange Clinic Registry do cations trialdrevisited. Diabetes 2008;57:995– Rewers M. Incidence of diabetic ketoacidosis at not meet American Diabetes Association or Inter- 1001 diagnosis of type 1 diabetes in Colorado youth, national Society for Pediatric and Adolescent Di- 30. Bryden KS, Peveler RC, Stein A, Neil A, Mayou 1998-2012. JAMA 2015;313:1570–1572 abetes clinical guidelines. Diabetes Care 2013;36: RA, Dunger DB. Clinical and psychological course 3. Ghetti S, Lee JK, Sims CE, Demaster DM, Glaser 2035–2037 of diabetes from adolescence to young adult- NS. Diabetic ketoacidosis and memory dysfunc- 17. Orchard TJ, Nathan DM, Zinman B, et al.; hood: a longitudinal cohort study. Diabetes Care tion in children with type 1 diabetes. J Pediatr Writing Group for the DCCT/EDIC Research 2001;24:1536–1540 2010;156:109–114 Group. Association between 7 years of intensive 31. Wysocki T, Hough BS, Ward KM, Green LB. 4. Cameron FJ, Scratch SE, Nadebaum C, et al.; treatment of type 1 diabetes and long-term mor- Diabetes mellitus in the transition to adulthood: DKA Brain Injury Study Group. Neurological con- tality. JAMA 2015;313:45–53 adjustment, self-care, and health status. J Dev sequences of diabetic ketoacidosis at initial pre- 18. Livingstone SJ, Levin D, Looker HC, et al.; Scot- Behav Pediatr 1992;13:194–201 sentation of type 1 diabetes in a prospective tish Diabetes Research Network epidemiology 32. Northam EA, Anderson PJ, Werther GA, cohort study of children. Diabetes Care 2014;37: group; Scottish Renal Registry. Estimated life ex- Warne GL, Andrewes D. Predictors of change in 1554–1562 pectancy in a Scottish cohort with type 1 diabetes, the neuropsychological profiles of children with 5. Usher-Smith JA, Thompson MJ, Sharp SJ, 2008-2010. JAMA 2015;313:37–44 type 1 diabetes 2 years after disease onset. Di- Walter FM. Factors associated with the presence 19. Lind M, Svensson AM, Kosiborod M, et al. abetes Care 1999;22:1438–1444 of diabetic ketoacidosis at diagnosis of diabetes in Glycemic control and excess mortality in type 1 33. Ziegler AG, Rewers M, Simell O, et al. Sero- children and young adults: a systematic review. diabetes. N Engl J Med 2014;371:1972–1982 conversion to multiple islet autoantibodies and BMJ 2011;343:d4092 20. Wolfsdorf J, Craig ME, Daneman D, et al. Di- risk of progression to diabetes in children. JAMA 6. Bui H, To T, Stein R, Fung K, Daneman D. Is abetic ketoacidosis in children and adolescents with 2013;309:2473–2479 diabetic ketoacidosis at disease onset a result of diabetes. Pediatr Diabetes 2009;10(Suppl. 12): 34. Barker JM, Goehrig SH, Barriga K, et al.; DAISY missed diagnosis? J Pediatr 2010;156:472–477 118–133 study. Clinical characteristics of children diagnosed care.diabetesjournals.org Duca and Associates 1255

with type 1 diabetes through intensive screen- prevalence of diabetic ketoacidosis at diagnosis general population: design and initial results of ing and follow-up. Diabetes Care 2004;27:1399– of type 1 diabetes in young children participating the Fr1da study. BMJ Open 2016;6:e011144 1404 in longitudinal follow-up. Diabetes Care 2011;34: 37. Kingery SE, Wu YL, Zhou B, Hoffman RP, Yu 35. Elding Larsson H, Vehik K, Bell R, et al.; 2347–2352 CY. Gene CNVs and protein levels of complement TEDDY Study Group; SEARCH Study Group; Swe- 36. Raab J, Haupt F, Scholz M, et al.; Fr1da Study C4A and C4B as novel biomarkers for partial dis- diabkids Study Group; DPV Study Group; Finn- Group. Capillary blood islet autoantibody screen- ease remissions in new-onset type 1 diabetes pa- ish Diabetes Registry Study Group. Reduced ing for identifying pre-type 1 diabetes in the tients. Pediatr Diabetes 2012;13:408–418