M. D. McElvany, Department of Orthopaedics, The Permanente Medical Group, Santa Rosa, CA, USA P. H. Chan, H. A. Prentice, E. W. Paxton, Surgical Outcomes and Analysis, Kaiser Permanente, San Diego, CA, USA M. T. Dillon, Department of Orthopaedics, The Permanente Medical Group, Sacramento, CA, USA R. A. Navarro, Department of Orthopaedics, Southern California Permanente Medical Group, Harbor City, CA, USA.
Clin Orthop Relat Res. 2019 Jun;477(6):1358-1369. doi: 10.1097/CORR.0000000000000642.
Prior studies have identified diabetes and disease severity (defined using hemoglobin A1c [HbA1c]) as potential risk factors for complications after shoulder arthroplasty. Evaluations of diabetes status and risk of adverse outcomes beyond the 30-day window either are limited or have not accounted for disease severity. Further, measures of diabetes severity other than HbA1c have yet to be investigated in a shoulder arthroplasty population.
QUESTIONS/PURPOSES: (1) Are diabetes status and glycemic control associated with adverse events, including deep infection, all-cause revision, and 90-day readmission after shoulder arthroplasty? (2) Is postoperative HbA1c associated with revision risk? (3) Is there a threshold of preoperative HbA1c that best identifies patients with diabetes who are at higher risk of 3-year deep infection, 1-year all-cause revision, or 90-day readmission? (4) Can the Adapted Diabetes Complications Severity index (aDCSI) be used as an alternative measure of diabetes severity in evaluating the risk of deep infection, all-cause revision, and 90-day readmission and identification of patients with diabetes at higher risk for these events? (5) Is there a difference between elective and traumatic shoulder arthroplasty patients?
We conducted a retrospective registry-based cohort study using Kaiser Permanente's Shoulder Arthroplasty Registry (2005-2015). Primary shoulder arthroplasties were classified as patients with and without diabetes. Patients with diabetes were further evaluated using two disease severity measures (1) HbA1c, with good glycemic control classified as preoperative HbA1c < 7.0% and poor control defined as HbA1c ≥ 7.0%; and (2) aDCSI, classified as mild (score of 0-2) or severe (score ≥ 3) diabetes. Cox regression was used to evaluate the risk of deep infection and revision according to diabetes status and disease severity; conditional logistic regression was used for 90-day readmission. Time-dependent 1-year postoperative HbA1c was used to evaluate revision risk in Cox regression. All models were adjusted for covariates and stratified by elective versus trauma shoulder arthroplasty. Receiver operating characteristic curves were generated for HbA1c and aDCSI to determine whether a threshold exists to identify patients at higher risk of deep infection, all-cause revision, or 90-day readmission. The study sample consisted of 8819 patients; 7353 underwent elective shoulder arthroplasty and 1466 underwent shoulder arthroplasty due to trauma. For elective shoulder arthroplasty, 1430 patients (19%) had diabetes, and among the patients who underwent arthroplasty due to trauma, 444 (30%) had diabetes.
Patients with diabetes who underwent elective shoulder arthroplasty and had poor glycemic control had a higher likelihood of 90-day readmission compared with patients without diabetes (OR, 1.5; 95% CI, 1.0-2.1; p = 0.032). No association was found for patients with diabetes who underwent shoulder arthroplasty due to trauma. No association was found between postoperative HbA1c and revision risk. Receiver operating characteristic curve analysis suggested preoperative HbA1c performed poorly at differentiating adverse events. When using aDCSI, patients with severe diabetes who underwent both elective and traumatic shoulder arthroplasty had a higher likelihood of 90-day readmission compared with patients without diabetes (OR, 1.6; 95% CI, 1.2-2.2; p = 0.001 and OR, 1.8; 95% CI, 1.2-2.7; p = 0.005, respectively). Similar to HbA1c, the aDCSI was a poor classifier in differentiating adverse events.
Of the longer-term outcomes evaluated, more-severe diabetes was only found to be associated with an increase in 90-day readmissions after shoulder arthroplasty; a stronger association was found when using the aDCSI in identifying diabetes severity. Arbitrary cutoffs in HbA1c may not be the best method for determining risk of postoperative outcomes. Future work investigating perioperative diabetes management should work to identify and validate measures, such as the aDCSI, that better identify patients at higher risk for postoperative outcomes and, more importantly, whether outcomes can be improved by modifying these measures with targeted interventions.
Level III, therapeutic study.
先前的研究已经确定糖尿病和疾病严重程度（通过血红蛋白 A1c [HbA1c] 定义）是肩部关节置换术后并发症的潜在危险因素。对糖尿病状态的评估以及 30 天窗口期之外的不良结果的风险评估有限，或者没有考虑到疾病严重程度。此外，在肩部关节置换术患者群体中，还没有研究其他衡量糖尿病严重程度的方法。
问题/目的：(1) 糖尿病状态和血糖控制是否与肩部关节置换术后的不良事件（包括深部感染、全因翻修和 90 天再入院）相关？(2) 术后 HbA1c 是否与翻修风险相关？(3) 是否存在术前 HbA1c 的阈值，能够最佳地识别出糖尿病患者，他们的深部感染、1 年全因翻修或 90 天再入院的风险更高？(4) 改良糖尿病并发症严重程度指数（aDCSI）能否作为一种替代的糖尿病严重程度衡量方法，用于评估深部感染、全因翻修和 90 天再入院的风险，并识别出这些事件风险较高的糖尿病患者？(5) 择期和创伤性肩部关节置换术患者之间是否存在差异？
我们进行了一项基于回顾性登记的队列研究，使用 Kaiser Permanente 的肩部关节置换术登记处（2005-2015 年）。将原发性肩部关节置换术分为糖尿病患者和非糖尿病患者。对糖尿病患者进一步使用两种疾病严重程度指标进行评估：（1）HbA1c，术前 HbA1c < 7.0%为良好血糖控制，术前 HbA1c ≥ 7.0%为不良控制；（2）aDCSI，分为轻度（得分 0-2）或重度（得分≥3）糖尿病。使用 Cox 回归评估根据糖尿病状态和疾病严重程度的深部感染和翻修风险；使用条件逻辑回归评估 90 天再入院率。使用时间依赖性术后 1 年 HbA1c 评估 Cox 回归中的翻修风险。所有模型均经过调整并按择期和创伤性肩部关节置换术进行分层。为了确定是否存在更高风险的深部感染、全因翻修或 90 天再入院的患者，绘制了 HbA1c 和 aDCSI 的受试者工作特征曲线。研究样本包括 8819 名患者；7353 名患者接受了择期肩部关节置换术，1466 名患者因创伤接受了肩部关节置换术。对于择期肩部关节置换术，1430 名患者（19%）患有糖尿病，在因创伤接受关节置换术的患者中，444 名（30%）患有糖尿病。
与非糖尿病患者相比，行择期肩部关节置换术且血糖控制不佳的糖尿病患者 90 天再入院的可能性更高（OR，1.5；95%CI，1.0-2.1；p = 0.032）。对于因创伤而接受肩部关节置换术的患者，未发现与糖尿病相关的关联。术后 HbA1c 与翻修风险之间没有关联。受试者工作特征曲线分析表明，术前 HbA1c 在区分不良事件方面表现不佳。当使用 aDCSI 时，接受择期和创伤性肩部关节置换术的重度糖尿病患者的 90 天再入院率与非糖尿病患者相比更高（OR，1.6；95%CI，1.2-2.2；p = 0.001 和 OR，1.8；95%CI，1.2-2.7；p = 0.005）。与 HbA1c 相似，aDCSI 在区分不良事件方面也是一种较差的分类器。
在所评估的长期结果中，只有更严重的糖尿病与肩部关节置换术后 90 天再入院率增加相关；使用 aDCSI 可以更好地识别出糖尿病严重程度，从而识别出更高的 90 天再入院风险。术前 HbA1c 的任意截断值可能不是确定术后结果风险的最佳方法。未来研究应致力于确定和验证衡量糖尿病严重程度的措施，如 aDCSI，这些措施能够更好地识别出术后结果风险更高的患者，更重要的是，是否可以通过有针对性的干预措施来改善这些措施，从而改善结果。
III 级，治疗性研究。
