Koziol James, Johnson Keith, Brenner Kathy, Fortmann Addie, Morrisey Robin, Philis-Tsimikas Athena
Scripps Research Institute, La Jolla, CA, USA.
Scripps Health, San Diego, CA, USA.
J Diabetes Sci Technol. 2015 Mar;9(2):246-56. doi: 10.1177/1932296814564992. Epub 2014 Dec 23.
Hyperglycemia and glucose variability in the hospital environment are associated with higher rates of complications, longer lengths of stay, and mortality. Standardized metrics are needed to assess the efficacy and safety of glucose management interventions. Glucometric data were collected from 2024 inpatients in a San Diego hospital between 2009 and 2011. As a complementary measure of glucose control, individual patient excursion rates were calculated using counts of distinct excursions from normal to critical glucose ranges >180 or <70 mg/dL. Prediction models for excursion rates were devised, based on patient demographic and clinical characteristics. Patients were predominantly male (51.2%), Caucasian (86.0%), and elderly (median age 72 years). Obesity was prevalent: 32% were overweight and 33% were obese. Median length of hospitalization was 5.0 days (range, 0.8-139.4 days). Unadjusted rate of excursions >180 mg/dL was 0.456 per 24 hours. The proportion of zero excursions decreased as severity of illness decreased, but was unrelated to age. Excursion rates were slightly smaller for major and extreme severity of illness compared to mild or moderate illness severity. Excursion rates did not vary in a monotone fashion with age, although the general pattern reflected a reduction in excursion rates from the first age quartile (19 to 59) through the last age quartile (83 to 100). Using the Akaike information criterion, zero-inflated negative binomial models were identified as appropriate for analyzing glucose excursion rates. Systematic approaches to glucose reporting and management in the hospital environment offer "windows of opportunity" to improve diabetes care.
医院环境中的高血糖和血糖变异性与更高的并发症发生率、更长的住院时间和死亡率相关。需要标准化指标来评估血糖管理干预措施的有效性和安全性。2009年至2011年期间，从圣地亚哥一家医院的2024名住院患者中收集了血糖仪数据。作为血糖控制的补充指标，通过计算从正常血糖范围到临界血糖范围（>180或<70mg/dL）的不同波动次数，得出个体患者的波动率。基于患者的人口统计学和临床特征，设计了波动率预测模型。患者主要为男性（51.2%）、白种人（86.0%）且年龄较大（中位年龄72岁）。肥胖很普遍：32%超重，33%肥胖。中位住院时间为5.0天（范围0.8 - 139.4天）。未调整的>180mg/dL波动率为每24小时0.456次。零波动的比例随着疾病严重程度的降低而减少，但与年龄无关。与轻度或中度疾病严重程度相比，重度和极重度疾病的波动率略小。波动率并不随年龄呈单调变化，尽管总体模式显示从第一年龄四分位数（19至59岁）到最后年龄四分位数（83至100岁）波动率有所降低。使用赤池信息准则，确定零膨胀负二项模型适合分析血糖波动率。医院环境中系统的血糖报告和管理方法为改善糖尿病护理提供了“机会窗口”。
