Platen Moritz, Buchholz Maresa, Rädke Anika, Gläser Eva, Iskandar Audrey, van den Berg Neeltje, Hoffmann Wolfgang, Michalowsky Bernhard
German Center for Neurodegenerative Diseases (DZNE), Site Greifswald, Ellernholzstrasse 1-2, 17489, Greifswald, Germany.
Section Epidemiology of Health Care and Community Health, Institute for Community Medicine, University Medicine Greifswald (UMG), Ellernholzstrasse 1-2, 17489, Greifswald, Germany.
Neurol Ther. 2025 Jun 12. doi: 10.1007/s40120-025-00778-y.
Claims data typically lack clinical parameters such as dementia severity, limiting insights into disease progression and related healthcare utilization and costs. Although diagnoses, prescriptions, and utilization patterns may serve as proxies, their validity is unclear. This study aimed to identify and validate these parameters to distinguish early from severe dementia stages.
Baseline data from 737 patients with dementia were analyzed. Dementia severity was assessed using the Mini-Mental State Examination and classified as early (≥ 27), mild (20-26), and moderate to severe (0-19). Healthcare utilization was recorded via structured interviews. Diagnoses, long-term care levels, and prescribed medications were extracted from physicians' files. Ordinal logistic regression evaluated associations between predictors and severity, with average marginal effects (AME) quantifying impact. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were computed for key predictors.
Among the sample (56% female patients, mean age 80), 18% were in the early stages, 43% mild, and 39% moderate to severe. Antipsychotic prescriptions (odds ratio (OR) 3.40, 95% confidence interval (CI) 1.94-5.95), antidementia drugs (OR 2.31, 95% CI 1.56-3.40), and higher long-term care levels (OR 5.59, 95% CI 2.23-13.99 for level ≥ 4) were associated with advanced severity. AME analysis revealed that antipsychotic use reduced early-stage probability by 14% and increased severe-stage probability by 21%. Similarly, antidementia drugs lowered early-stage probability by 9% and raised severe-stage probability by 13%. Increasing care levels were associated with a 2-16% decline in early-stage probability and a 3-34% rise in severe-stage probability. The combined model showed high specificity (99.6%) and PPV (84.6%) for severe dementia, but sensitivity and NPV for early stage were low.
Antidementia drugs, antipsychotics, and long-term care level serve as robust predictors of moderate to severe dementia, whereas early-stage detection remains challenging. Future studies should validate these markers and explore additional predictors to improve early detection in claims data.
索赔数据通常缺乏痴呆严重程度等临床参数，这限制了对疾病进展以及相关医疗保健利用情况和成本的深入了解。尽管诊断、处方和利用模式可能可作为替代指标，但其有效性尚不清楚。本研究旨在识别并验证这些参数，以区分痴呆的早期和重度阶段。
对737例痴呆患者的基线数据进行分析。使用简易精神状态检查表评估痴呆严重程度，并分为早期（≥27分）、轻度（20 - 26分）和中度至重度（0 - 19分）。通过结构化访谈记录医疗保健利用情况。从医生档案中提取诊断、长期护理水平和处方药物。有序逻辑回归评估预测因素与严重程度之间的关联，平均边际效应（AME）量化影响。计算关键预测因素的敏感性、特异性、阳性预测值（PPV）和阴性预测值（NPV）。
在样本中（56%为女性患者，平均年龄80岁），18%处于早期阶段，43%为轻度，39%为中度至重度。抗精神病药物处方（比值比（OR）3.40，95%置信区间（CI）1.94 - 5.95）、抗痴呆药物（OR 2.31，95% CI 1.56 - 3.40）以及更高的长期护理水平（对于≥4级水平，OR 5.59，95% CI 2.23 - 13.99）与病情严重程度增加相关。AME分析显示，使用抗精神病药物使早期概率降低14%，重度阶段概率增加21%。同样，抗痴呆药物使早期概率降低9%，重度阶段概率增加13%。护理水平提高与早期概率下降2% - 16%以及重度阶段概率上升3% - 34%相关。联合模型对重度痴呆显示出高特异性（99.6%）和PPV（84.6%），但早期阶段的敏感性和NPV较低。
抗痴呆药物、抗精神病药物和长期护理水平可作为中度至重度痴呆的有力预测指标，而早期检测仍然具有挑战性。未来的研究应验证这些标志物，并探索其他预测因素，以改善索赔数据中的早期检测。
