Database of Electronic Health Records: Application in Clinical Research and Bias Control
-
摘要: 电子病历数据库(electronic health records, EHRs)的出现为大样本临床队列研究提供了低成本、高效率的愿景和机遇,但利用EHRs开展临床研究仍然存在误区和挑战。本文回归流行病学的本质,通过实例探讨EHRs的特征,EHRs中可获得的信息与临床研究中暴露变量、结局变量及协变量之间的对应关系,如何利用EHRs进行患病率及发病率估计、开展疗效和政策评价以及在临床研究中对选择偏倚、混杂偏倚尤其是未测量混杂偏倚的控制等,旨在促进EHRs与临床研究的整合,提高临床研究的效率及质量。Abstract: The emergence of electronic health records (EHRs) provides a good opportunity for clinical studies to be carried out on large samples with high efficiency and at low costs. However, misunderstanding and inappropriate application of EHRs in clinical research are common. This paper brings us back to the essentials of epidemiology. We use a number of examples to discuss: the characteristics of EHRs, the relationship of data domains of EHRs with their corresponding variables (exposure, outcomes, and covariates) in epidemiological research, and how to use EHRs to estimate prevalence and incidence and to evaluate the effectiveness of treatment and policy. We also focus on the selection bias and confounding controls, especially unmeasured confounding controls. Hopefully, this paper would contribute to the integration of EHRs and clinical research and to the improvement of efficiency and quality of clinical research.
-
Key words:
- electronic health records /
- clinical research /
- bias control
-
表 1 电子病历数据库中信息与临床研究中变量的对应关系[3]
电子病历数据库信息分类 电子病历数据库中的相关信息 可能在临床研究中扮演的角色 人口学 年龄、性别、民族、居住地 暴露因素、协变量(混杂因素、效应修饰因子) 生命指征 脉搏、血压 结局变量、暴露变量、协变量 诊断 ICD-9、ICD-10 结局变量、疾病严重程度、结合诊断时间可提示病程 实验室检查 血尿常规、肝肾功能、代谢相关指标 结局变量、协变量、同时可帮助评价诊断的准确性、疾病的严重程度等 用药 名称、剂量、频率、使用时间 干预指标、提示疾病的严重程度 影像学 超声、磁共振成像等 辅助诊断 文本描述 - 病史、其他暴露信息等 ICD:国际疾病分类 -
[1] ISO/TR 20514: 2005. Health informatics-electronic health record-definition, scope, and context[EB/OL].https://www.iso.org/standard/39525.html. [2] 王雯, 刘艳梅, 谭婧, 等.回顾性数据库研究的概念、策划与研究数据库构建[J].中国循证医学杂志, 2018, 18:230-237. http://www.cnki.com.cn/Article/CJFDTotal-ZZXZ201802016.htm [3] Casey JA, Schwartz BS, Stewart WF, et al. Using electronic health records for population health research: a review of methods and applications[J]. Annu Rev Publ Health, 2016, 37:61-81. doi: 10.1146/annurev-publhealth-032315-021353 [4] Kavakiotis I, Tsave O, Salifoglou A, et al. Machine learning and data mining methods in diabetes research[J]. Comput Struct Biotechnol J, 2017, 15:104-116. doi: 10.1016/j.csbj.2016.12.005 [5] Ng SC, Leung WK, Shi HY, et al. Epidemiology of inflammatory bowel disease from 1981 to 2014: results from a territory-wide population-based registry in HongKong[J]. Infl-amm Bowel Dis, 2016, 22:1954-1960. doi: 10.1097/MIB.0000000000000846 [6] Esteban-Vasallo MD, Dominguez-Berjon MF, Astray-Moch-ales J, et al. Epidemiological usefulness of population-based electronic clinical records in primary care: estimation of the prevalence of chronic diseases[J]. Fam Pract, 2009, 26:445-454. doi: 10.1093/fampra/cmp062 [7] Tomasallo CD, Hanrahan LP, Tandias A, et al. Estimating wisconsin asthma prevalence using clinical electronic health records and public health data[J]. Am J Public Health, 2014, 104:E65-E73. http://europepmc.org/abstract/med/24228643 [8] Bagley SC, Altman RB. Computing disease incidence, prevalence and comorbidity from electronic medical records[J]. J Biomed Inform, 2016, 63:108-111. doi: 10.1016/j.jbi.2016.08.005 [9] Mamtani R, Haynes K, Finkelman BS, et al. Distinguishing incident and prevalent diabetes in an electronic medical records database[J]. Pharmacoepidemiol Drug Saf, 2014, 23:111-118. doi: 10.1002/pds.3557 [10] Wu CY, Lin JT, Ho HJ, et al. Association of nucleos(t)-ide analogue therapy with reduced risk of hepatocellular carcinoma in patients with chronic hepatitis B: a nationwide cohort study[J]. Gastroenterology, 2014, 147:143-151 doi: 10.1053/j.gastro.2014.03.048 [11] Qiu Q, Duan XW, Li Y, et al. Impact of partial reimbursement on hepatitis B antiviral utilization and adherence[J]. World J Gastroenterol, 2015, 21:9588-9597. doi: 10.3748/wjg.v21.i32.9588 [12] Qiu Q, Li Y, Cheng K, et al. Cost-effectiveness of partial reimbursement for hepatitis B anti-viral drugs in Beijing, China: an analysis based on a retrospective cohort study[J]. Lancet, 2015, 386:S23-S23. http://www.onacademic.com/detail/journal_1000038924218110_f114.html [13] Lucas R, Ponsonby AL, McMichael A, et al. Observational analytic studies in multiple sclerosis: controlling bias through study design and conduct. The Australian Multicentre Study of Environment and Immune Function[J]. Mult Scler, 2007, 13:827-839. doi: 10.1177/1352458507077174 [14] Brookhart MA, Sturmer T, Glynn RJ, et al. Confounding control in healthcare database research: challenges and potential approaches[J]. Med Care, 2010, 48:S114-S120. http://drc.bmj.com/lookup/external-ref?access_num=20473199&link_type=MED&atom=%2Fbmjdrc%2F5%2F1%2Fe000435.atom [15] Pearce N, Checkoway H, Kriebel D. Bias in occupational epidemiology studies[J]. Occup Environ Med, 2007, 64:562-568. doi: 10.1136/oem.2006.026690 [16] Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects[J]. Biometrika, 1983, 70:41-55. doi: 10.1093/biomet/70.1.41 [17] Brookhart MA, Wang PS, Solomon DH, et al. Evaluating short-term drug effects using a physician-specific prescribing preference as an instrumental variable[J]. Epidemiology, 2006, 17:268-275. doi: 10.1097/01.ede.0000193606.58671.c5 [18] Ashenfelter O, Card D. Using the longitudinal structure of earnings to estimate the effect of training programs[J]. Rev Econ Stat, 1985, 67:648-660. doi: 10.2307/1924810 [19] Austin PC, Grootendorst P, Anderson GM. A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study[J]. Stat Med, 2007, 26:734-753. doi: 10.1002/sim.2580 [20] Mccandless LC, Gustafson P, Austin PC. Bayesian propensity score analysis for observational data[J]. Stat Med, 2009, 28:94-112. doi: 10.1002/sim.3460 [21] Li L, Shen C, Wu AC, et al. Propensity score-based sensitivity analysis method for uncontrolled confounding[J]. Am J Epidemiol, 2011, 174:345. doi: 10.1093/aje/kwr096 [22] Schlesselman JJ. Assessing effects of confounding variables[J]. Am J Epidemiol, 1978, 108:3-8. http://academic.oup.com/aje/article/108/1/3/195354 [23] Greenland S. The Impact of Prior Distributions for uncont-rolled confounding and response bias[J]. JASA, 2003, 98:47-54. doi: 10.1198/01621450338861905 [24] Mccandless LC, Gustafson P, Austin PC, et al. Covariate balance in a Bayesian propensity score analysis of beta blocker therapy in heart failure patients[J]. Epidemiol Perspect Innov, 2009, 6:5. doi: 10.1186/1742-5573-6-5 [25] An W. Bayesian propensity score estimators: incorporating uncertainties in propensity scores into causal inference[J]. Sociol Methodol, 2015, 40:151-189. http://biomet.oxfordjournals.org/cgi/ijlink?linkType=ABST&journalCode=spsmx&resid=40/1/151 [26] Greenland S. An introduction to instrumental variables for epidemiologists[J]. Int J Epidemiol 2000;29:1102. http://ije.oxfordjournals.org/content/29/4/722 [27] Stukel TA, Fisher ES, Wennberg DE, et al. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods[J]. JAMA, 2007, 297:278-285. doi: 10.1001/jama.297.3.278 [28] Ashenfelter O. Estimating the effect of training programs on earnings[J]. Rev Econ Stat, 1978, 60:47-57. doi: 10.2307/1924332 [29] 黄远飞, 张家业.城乡医保整合对农村居民医疗服务利用的影响-以广州市为例[J].中国公共政策评论, 2017, 1:34-52. http://search.cnki.net/down/default.aspx?filename=ZGZP201701004&dbcode=CJFD&year=2017&dflag=pdfdown
点击查看大图
表(1)
计量
- 文章访问数: 242
- HTML全文浏览量: 64
- PDF下载量: 251
- 被引次数: 0