Rodgers Lauren R, Weedon Michael N, Henley William E, Hattersley Andrew T, Shields Beverley M
Institute of Health Research, University of Exeter Medical School, Exeter, UK.
Institute of Biomedical and Clinical Science, University of Exeter Medical School, Exeter, UK.
BMJ Open. 2017 Oct 12;7(10):e017989. doi: 10.1136/bmjopen-2017-017989.
This is a retrospective cohort study using observational data from anonymised primary care records. We identify and extract all patients with type 2 diabetes and associated clinical data from the Clinical Practice Research Datalink (CPRD) to inform models of disease progression and stratification of treatment.
Data were extracted from CPRD on 8 August 2016. The initial data set contained all patients (n=313 485) in the database who had received a type 2 diabetes medication. Criteria were applied to identify and exclude those with type 1 diabetes, polycystic ovarian syndrome or other forms of diabetes (n=40 204), and for data quality control (n=12). We identified 251 338 patients for inclusion in future analyses of diabetes progression and treatment response.
For 6-month response to treatment, measured by change in glycated haemoglobin (HbA1c), we have 91 765 patients with 119 785 treatment response episodes. The greatest impact on reduction of HbA1c occurs with first-line and second-line treatments, metformin and sulfonylurea. Patients moving to third-line treatments tend to have greater weights and higher body mass index. We have investigated the impact of non-adherence to commonly used glucose-lowering medications on HbA1c. For baseline-adjusted HbA1c change over 1 year, non-adherent patients had lower HbA1c reductions than adherent patients, with mean and 95% CI of -4.4 (-4.7 to -4.0) mmol/mol (-0.40 (-0.43 to -0.37) %).
Findings from studies using these data will help inform future treatment plans and guidelines. Additional data are added with updates from CPRD. This will increase the numbers of patients on newer medications and add more data on those already receiving treatment. There are several ongoing studies investigating different hypotheses regarding differential response to treatment and progression of diabetes. For side effects, links to Hospital Episode Statistics data, where severe events such as hypoglycaemia will be recorded, will also be explored.
这是一项回顾性队列研究，使用来自匿名初级医疗记录的观察数据。我们从临床实践研究数据链（CPRD）中识别并提取所有2型糖尿病患者及相关临床数据，以建立疾病进展模型和治疗分层模型。
数据于2016年8月8日从CPRD中提取。初始数据集包含数据库中所有接受过2型糖尿病药物治疗的患者（n = 313485）。应用标准来识别并排除1型糖尿病、多囊卵巢综合征或其他形式糖尿病患者（n = 40204），以及进行数据质量控制（n = 12）。我们确定了251338名患者纳入未来糖尿病进展和治疗反应分析。
对于通过糖化血红蛋白（HbA1c）变化衡量的6个月治疗反应，我们有91765名患者，共119785次治疗反应事件。对降低HbA1c影响最大的是一线和二线治疗药物二甲双胍和磺脲类药物。转向三线治疗的患者往往体重更大、体重指数更高。我们研究了不依从常用降糖药物对HbA1c的影响。对于1年期间经基线调整的HbA1c变化，不依从患者的HbA1c降低幅度低于依从患者，平均值及95%置信区间为-4.4（-4.7至-4.0）mmol/mol（-0.40（-0.43至-0.37）%）。
使用这些数据的研究结果将有助于为未来治疗计划和指南提供信息。随着CPRD的更新会添加更多数据。这将增加使用新药的患者数量，并增加已接受治疗患者的更多数据。有几项正在进行的研究正在调查关于糖尿病治疗反应差异和进展的不同假设。对于副作用，还将探索与医院事件统计数据的关联，其中将记录低血糖等严重事件。
