Background

The imperative to identify patients at risk of medication-related harm has never been greater. Hospital clinicians cannot easily predict risk of readmission or harm. Candidate variables associated with medication-related harm derived from the literature or significantly represented in a complex patient cohort have been previously described by PHarmacie-4. With a focus on polypharmacy and high-risk medicines in vulnerable patient cohorts, PHarmacie-4 was easy to use and highlighted risks. However it over-estimated risk, reducing its usefulness in stratifying risk of readmission.

Objective

Develop a risk prediction tool built into a smart phone app, enabling clinicians to identify and refer high-risk patients for an early post-discharge medicines review. Demonstrate usability, real world application and validity in an independent dataset.

Methods

A retrospective, observational study was conducted with 1201 randomly selected patients admitted to Sir Charles Gairdner Hospital between June 1, 2016 to December 31, 2016. Patient characteristics and outcomes of interest were reported, including unplanned hospital utilisation at 30, 60 and 90 days post-discharge. Using multivariable logistic regression modelling, an algorithm was developed, built into a smart phone app and used and validated in an independent dataset.

Results

738 patients (61%) were included in the derivation sample. The best predictive performance was achieved by PHarmacie-R (C-statistic 0.72, 95% CI 0.68–0.75) which included PHarmacie-4 risk variables, a non-linear effect of age, unplanned hospital utilisation in the preceding six months and gender. The independent validation dataset had a C-statistic of 0.64 (95% CI 0.56–0.72).

Conclusion

PHarmacie-R is the first readmission risk prediction tool, built into a smart phone app, focussing on polypharmacy and high-risk medicines in vulnerable patients. It can assist clinical pharmacists to identify medical inpatients who may benefit from early post-discharge medication management services. External validation is needed to enable application in other clinical settings.

中文翻译：

---

开发 PHarmacie-R：一种以药物管理为重点的床边风险预测工具，用于识别再入院风险

背景

识别有药物相关伤害风险的患者的必要性从未如此强烈。医院临床医生无法轻易预测再入院或伤害的风险。PHarmacie-4 之前已经描述了与来自文献的药物相关伤害相关的候选变量或在复杂患者队列中显着代表的候选变量。Pharmacie-4 专注于弱势患者群体中的多种药物和高风险药物，易于使用并突出风险。然而，它高估了风险，降低了其在再入院风险分层方面的有用性。

客观的

开发内置于智能手机应用程序的风险预测工具，使临床医生能够识别和转诊高风险患者，以便在出院后进行早期药物审查。在独立的数据集中展示可用性、现实世界的应用和有效性。

方法

对 2016 年 6 月 1 日至 2016 年 12 月 31 日期间在查尔斯盖尔德纳爵士医院收治的 1201 名随机选择的患者进行了一项回顾性观察性研究。报告了患者特征和感兴趣的结果，包括术后 30、60 和 90 天的计划外医院使用情况-释放。使用多变量逻辑回归建模，开发了一种算法，将其内置到智能手机应用程序中，并在独立数据集中使用和验证。

结果

推导样本中包括 738 名患者 (61%)。PHarmacie-R (C-statistic 0.72, 95% CI 0.68–0.75) 实现了最佳预测性能，其中包括 PHarmacie-4 风险变量、年龄的非线性效应、前六个月的计划外医院利用率和性别。独立验证数据集的 C 统计量为 0.64（95% CI 0.56–0.72）。

结论

Pharmacie-R 是第一个再入院风险预测工具，内置于智能手机应用程序中，专注于弱势患者的多药治疗和高风险药物。它可以帮助临床药剂师识别可能受益于早期出院后药物管理服务的内科住院患者。需要外部验证才能在其他临床环境中应用。