Painful Diabetic Peripheral Neuropathy in a Managed Care Setting: Patient Identification, Prevalence Estimates, and Pharmacy Utilization Patterns

Cynthia L. Hartsfield; Eli J. Korner; Jennifer L. EllisMarsha A. RaebelJohn MerenichNancy Brandenburg

摘要

The objectives of this study were to validate an algorithm for identifying patients with painful diabetic peripheral neuropathy (pDPN) and demonstrate its practical applications. Using the Kaiser Permanente Colorado Diabetes Registry, an algorithm was developed with selected ICD-9 diagnosis codes combined with automated pharmacy data for medications prescribed for pDPN symptoms. Medical records were reviewed to confirm pDPN presence and to inform algorithm refinement. Prevalence was estimated with a numerator of members with diabetes who had inclusion but no exclusion codes in 2003 (Method 1) and with a numerator of diabetes patients with inclusion codes between 1998 and 2003 who had no subsequent exclusion codes and who remained members in 2003 (Method 2); the denominator was all members with diabetes in 2003. Medication utilization was compared between patients with and without pDPN. A total of 19,577 members with diabetes were identified; 2612 met initial inclusion criteria. Medical record review (n = 298) demonstrated sensitivity of 94, specificity of 55, and positive predictive value (PPV) of 64. Inclusion criteria were modified and pharmacy data eliminated. The revised algorithm identified 1754 additional patients meeting inclusion criteria. Medical record review (n = 190) demonstrated sensitivity of 99, specificity of 49, and PPV of 79. Using the validated algorithm, pDPN prevalence was 113 (Method 1) and 208 (Method 2) per 1000 persons with diabetes. Significant differences were observed in medication prescriptions between patients with and without pDPN. Estimated pDPN prevalence among persons with diabetes was 11-21 and pDPN patients had greater utilization of selected medications than those without pDPN. Identifying patients with pDPN is a fundamental step for improving disease management and understanding the economic impact of pDPN.

机译：本研究的目的是验证一种用于识别疼痛性糖尿病周围神经病变（pDPN）患者的算法并展示其实际应用。使用 Kaiser Permanente Colorado Diabetes Registry，开发了一种算法，该算法将选定的 ICD-9 诊断代码与用于 pDPN 症状的药物的自动药房数据相结合。审查了医疗记录以确认 pDPN 的存在并为算法改进提供信息。用 2003 年有纳入但没有排除代码的糖尿病成员的分子（方法 1）和 1998 年至 2003 年间有纳入代码的糖尿病患者的分子来估计患病率，这些患者没有后续排除代码，但在 2003 年仍然是成员（方法 2）;分母是2003年所有患有糖尿病的成员。比较有和没有pDPN的患者之间的药物利用率。共确定了 19,577 名患有糖尿病的成员;2612 例符合初始纳入标准。病历审查（n = 298）显示敏感性为 94%，特异性为 55%，阳性预测值（PPV）为 64%。修改了纳入标准，并删除了药房数据。修订后的算法确定了另外 1754 名符合纳入标准的患者。病历审查（n = 190）显示敏感性为 99%，特异性为 49%，PPV 为 79%。使用经过验证的算法，pDPN 患病率为每 1000 名糖尿病患者 113 例（方法 1）和 208 例（方法 2）。在有和没有pDPN的患者之间观察到药物处方的显着差异。估计糖尿病患者的 pDPN 患病率为 11%-21%，pDPN 患者比没有 pDPN 的患者更多地使用所选药物。识别 pDPN 患者是改善疾病管理和了解 pDPN 经济影响的基本步骤。

Painful Diabetic Peripheral Neuropathy in a Managed Care Setting: Patient Identification, Prevalence Estimates, and Pharmacy Utilization Patterns

摘要

著录项

引文网络

相关主题

期刊订阅