Lack of Association of A1C With Postoperative Complications in Children With Type 1 or Type 2 Diabetes

The Baylor College of Medicine Institutional Review Board approved the study (H-35613) with a waiver of informed consent because the study did not modify existing diagnostic or therapeutic strategies. The authors adhered to the Strobe guidelines for the study (31).

Data were collected retrospectively from the Epic electronic medical record system from the surgery and endocrinology databases. Children aged 1–18 years with a diagnosis of type 1 or type 2 diabetes undergoing an elective noncardiac surgery or diagnostic procedure at Texas Children’s Hospital from January 2011 to June 2021 were eligible for inclusion. Exclusion criteria included any urgent or emergent noncardiac surgery or diagnostic procedure, any cardiac surgery or cardiac catheterization procedure, and any diagnosis of diabetes other than type 1 or type 2 diabetes (e.g., cystic fibrosis–related diabetes, steroid-induced diabetes, and maturity-onset diabetes of the young).

Preoperative factors collected included age, sex, weight, height, BMI, race, ethnicity, A1C level within 3 months of the scheduled surgery or diagnostic procedure, glucose measurements within 2 hours of the procedure, type of diabetes, presence of moderate or large urine ketones, and type of diabetes treatment. Treatment types included insulin pump (with a traditional pump or automated insulin delivery system), multiple daily insulin injections (including intensive insulin management or fixed doses with or without a correction factor), long-acting insulin therapy only, metformin, and glucagon-like peptide 1 receptor agonist therapy. Intraoperative factors collected included surgery type (major or minor) and intraoperative glucose measurements. A major procedure was defined as either any intracavitary procedure or a procedure lasting >3 hours, and a minor procedure was defined as any noncavitary or peripheral procedure lasting <3 hours. Postoperative factors collected included postoperative glucose measurements for 7 days after the procedure.

The primary outcome was defined as a new-onset postoperative systemic infection, wound complication, or ketosis. Systemic infections were defined as either any new-onset bloodstream infection with a positive blood culture within 30 days of the procedure, pneumonia with a new infiltrate on chest radiograph within 30 days of the procedure, or urinary tract infection with a positive urine culture within 30 days of the procedure. Wound complications were defined as either any wound disruption or superficial, deep, or organ surgical site infection. Ketosis was defined as the presence of moderate or large urine ketones postoperatively.

Secondary outcomes were categorized as an unplanned outcome, defined as an unplanned admission, unplanned reoperation, or unplanned readmission within 30 days of the procedure. The time period of 30 days was used because it is the standard period to assess for postoperative complications in the Pediatric National Surgical Quality Improvement Program (NSQIP). Pediatric NSQIP data were further analyzed to determine the incidence of these complications nationally.

Patient demographics and characteristics were summarized by median with 25th and 75th percentiles and stratified between patients by A1C level (<7.0, ≥7.0 to <9, and ≥9%). These categories were determined by the Delphi method between anesthesiologists and endocrinologists at Texas Children’s Hospital. Only patients without preoperative ketosis were included in the postoperative ketosis outcome. To account for correlation between multiple procedures in the same patient, a generalized estimating equation (GEE) with unstructured correlation matrix were used to determine whether A1C strata were significantly associated with any infection, wound, ketosis, or unplanned outcome. A GEE was also used to test whether preoperative variables were significantly associated with outcomes. A significance level of 0.05 was used.

Of the 611 patients with a diagnosis of diabetes, 438 met the inclusion criteria (Figure 1). Summary statistics are shown in Table 1. Three-hundred and seventeen (72%) had a diagnosis of type 1 diabetes whereas 121 (28%) had a diagnosis of type 2 diabetes. One-hundred twenty children (28%) had an A1C <7.0%, 185 (42%) had an A1C ≥7.0% and <9%, and 133 (30%) had an A1C ≥9%. Figure 2 provides a scatterplot showing the relationship between A1C and preoperative glucose.

Outcome statistics are summarized in Table 2. The incidence of any postoperative systemic infections was 0.91% (n = 4). Of the three children with a postoperative urinary tract infection, none had an in-dwelling Foley catheter perioperatively. The incidence of any postoperative wound disruption was 3.33% (n = 19). The incidence of postoperative ketosis was 3.89% (n = 17). Univariable and multivariable analyses are shown in Tables 3 and 4, respectively. A1C levels were not associated with any postoperative systemic infections, wound complications, or ketosis. None of the other preoperative factors, including type of diabetes, BMI >95th percentile, and procedure type, were associated with these complications.

We found no association between A1C and postoperative infections, wounds, or ketosis complications in children with type 1 or type 2 diabetes presenting for elective noncardiac surgery or diagnostic procedures.

Studies in adults have not consistently demonstrated an association between A1C and postoperative complications. There is evidence that A1C levels may have no impact on 30-day mortality in adult surgery patients (32–36). Data demonstrating a relationship between A1C and postoperative infectious complications in adults, including surgical site infections and systemic infections, are mixed.

Notably, there is significant heterogeneity in the A1C level cutoffs used in these studies. In a recent systematic review, Lopez et al. (37) attempted to determine whether there is a quantitative relationship between preoperative A1C and postoperative complications and whether there is a critical A1C level beyond which postoperative complications become significant. The authors found an inconsistent relationship between A1C and postoperative complications regardless of cohort size. Furthermore, attempting to categorize the threshold level for risk was challenging because most of the studies that found an association assessed A1C as a dichotomous variable (37).

A number of factors limits extrapolation to children. A1C levels tend to be higher in children and adolescents than in adults, and children undergo less invasive procedures and have fewer comorbidities than adults. Finally, the length of disease and distribution of diabetes type varies in children (38). Type 2 diabetes is additionally more common than type 1 diabetes in adults. Children have a higher incidence of type 1 diabetes than of type 2 diabetes, although the incidence of type 2 diabetes in children is increasing secondary to increasing obesity. Type 2 diabetes in children is associated with decreased insulin sensitivity, increased insulin resistance, and faster β-cell deterioration than in adults (39,40). There are no data in children with type 1 or type 2 diabetes demonstrating an association between elevated A1C levels and postoperative complications; however, hyperglycemia has been associated with increased mortality and rate of postoperative complications in children without diabetes after cardiac surgery, neurological injury, burn injury, and necrotizing enterocolitis (15–18).

Published guidelines for the management of children with diabetes presenting for noncardiac surgery provide recommendations for perioperative glucose monitoring but do not identify high-risk A1C levels for elective procedures (28,29). We initially selected an A1C cutoff of 7.0% based on the ADA’s recommended glycemic target for children (26). We subsequently stratified A1C ≥7.0% into two levels: ≥7.0 to <9% and ≥9%. Although we arbitrarily defined these two A1C categories, consensus was based on the Delphi method between anesthesiologists and endocrinologists at our institution because there is no published consensus on cutoffs for elevated and extremely elevated A1C levels.

Our study was limited by the low postoperative complication rate. A larger sample size may have detected an association. The majority of our cohort underwent noninvasive, minor procedures resulting in few wound complications and systemic infections, even though procedure type was also not associated with postoperative complications. However, the relationship between the incidence of minor and major procedures reflects our institutional ratio for children undergoing elective noncardiac surgery or diagnostic procedures.

Additionally, in our cohort, ∼28% of children had type 2 diabetes, which is similar to the 31% reported by the National Health and Nutrition Examination Survey III (1988–1994) (41). We chose to investigate only elective procedures because children undergoing such procedures are more likely to be medically optimized before their procedure occurs. Investigating urgent or emergent surgery or diagnostic procedures may have demonstrated an association between A1C and postoperative complications.

The intention of this study was to investigate elective procedures because perioperative glycemic management guidelines provide recommendations for preoperative A1C level assessment but no criteria for when to delay these procedures (38,39). Although the incidence of postoperative infection and wound complications in our cohort was low, other metrics commonly used in adults (e.g., readmission rates, reoperation rates, postoperative renal dysfunction, and mortality) may not be as applicable in children given their low incidence.

Finally, our incidence of these complications was similar to the national incidence from the Pediatric NSQIP database, suggesting that the cohort size or institutional practice did not affect these complications (Table 2). However, the pediatric NSQIP database does not collect type 1 or type 2 diabetes as comorbidities or A1C as a preoperative factor.

Although we did not find a relationship between acute perioperative glucose levels and infectious and wound complications, we did find a relationship between preoperative glucose and postoperative ketosis. The odds of postoperative ketosis slightly increased as the mean preoperative glucose increased, possibly suggesting that the focus of intervention needs to be on this acute phase of glycemia rather than the more challenging long-term outpatient setting. Not surprisingly, there was no correlation between A1C and preoperative glucose, given that one assesses long-term glycemic status and another provides just a snapshot of glycemia.

Future directions could include assessing the effects of immediate preoperative glycemic status, particularly on postoperative ketosis, given this association found in our cohort. Also, although unplanned admissions and readmission rates are almost twice that in children with elevated A1C, we did not find a significant association between A1C and these two outcomes. A multi-institutional study with a larger cohort may reveal an association between A1C and these postoperative outcomes that was not evident in our cohort even though our incidence of postoperative outcomes matched that from a national database.

This study found no association of elevated preoperative A1C levels with postoperative infection, wound, or ketosis complications in children with type 1 or type 2 diabetes presenting for elective noncardiac surgery or diagnostic procedures. Although improving glycemic status is essential for children’s long-term health, delaying elective surgeries until A1C is consistently normalized may not be warranted given that it is often difficult to improve A1C levels rapidly.

No potential conflicts of interest relevant to this article were reported.

G.K. participated in conceptualization, methodology, data collection, writing and editing the initial draft, and reviewing the final manuscript. M.C.R., A.B.S., and K.M.S. participated in data collection, writing and editing the initial draft, and reviewing the final manuscript. X.H. and K.S. performed data analysis. S.A.S. participated in conceptualization, methodology, writing and editing the initial draft, and reviewing the final manuscript. R.G.B. supervised the work and participated in conceptualization, methodology, writing and editing the initial draft, and reviewing the final manuscript. R.G.B. is the guarantor of this work and, as such, had full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.