Soares Rebecca R, Parikh Devayu, Shields Charlotte N, Peck Travis, Gopal Anand, Sharpe James, Yonekawa Yoshihiro
Wills Eye Hospital, Mid Atlantic Retina, Thomas Jefferson University, Philadelphia, Pennsylvania.
Biostatistics Consulting Core, Vickie and Jack Farber Vision Research Center, Wills Eye Hospital, Philadelphia, Pennsylvania.
Ophthalmol Retina. 2021 Sep;5(9):879-887. doi: 10.1016/j.oret.2020.12.006. Epub 2020 Dec 17.
To identify geographic and socioeconomic variables predictive of residential proximity to diabetic eye disease clinical trial locations.
Cross-sectional, retrospective study.
De-identified census tract-level data from public datasets and trial-level data from ClinicalTrials.gov.
Using public data from ClinicalTrials.gov, we identified all active interventional clinical trials in diabetic eye disease since 2017. After geolocating every trial site, we used an origin-destination cost-matrix to calculate the driving distance and travel time from the population-weighted United States census tract centroid to the nearest site. We then used public databases to identify census tract-level socioeconomic factors predictive of driving distance and time.
Driving distance > 60 miles and time traveled > 60 minutes to the nearest clinical trial site.
In a multivariate model, driving distance of more than 60 miles had a significant association with rural versus urban location (adjusted odds ratio, 5.22; 95% confidence interval [CI], 3.75-7.26; P < 0.001), percentage of population at less than 200% of federal poverty level compared with the fourth quartile (first quartile: adjusted odds ratio, 0.40 [95% CI, 0.29-0.55]; second quartile: adjusted odds ratio, 0.60 [95% CI, 0.47-0.77]; third quartile: adjusted odds ratio, 0.76 [95% CI, 0.63-0.91]; P < 0.001) and the Midwest (adjusted odds ratio, 2.15; 95% CI, 1.13-4.07; P = 0.02), South (adjusted odds ratio, 2.71; 95% CI, 1.23-5.99; P = 0.01), and West (adjusted odds ratio, 3.01; 95% CI, 1.21-7.54; P = 0.02) regions as compared with the Northeast. Driving distance was associated with county-level prevalence of diabetes in the univariate model (odds ratio, 1.12; 95% CI, 1.06-1.19; P < 0.001), although it was nonsignificant in the multivariate model. Similar predictors were found for time traveled in minutes.
Geographic maldistributions of clinical trial sites exist for diabetic eye disease in the United States. Those with higher travel burden are more likely to reside in a census tract that is rural, low income, and from areas outside the Northeast.
确定能够预测居住地与糖尿病眼病临床试验地点距离远近的地理和社会经济变量。
横断面回顾性研究。
来自公共数据集的匿名普查区层面数据以及来自ClinicalTrials.gov的试验层面数据。
利用ClinicalTrials.gov的公开数据，我们确定了自2017年以来所有正在进行的糖尿病眼病介入性临床试验。在对每个试验地点进行地理定位后，我们使用起讫点成本矩阵来计算从美国人口加权普查区中心点到最近试验地点的驾车距离和出行时间。然后，我们利用公共数据库确定能够预测驾车距离和时间的普查区层面社会经济因素。
到最近临床试验地点的驾车距离>60英里以及出行时间>60分钟。
在多变量模型中，驾车距离超过60英里与农村或城市地区显著相关（校正比值比，5.22；95%置信区间[CI]，3.75 - 7.26；P<0.001），与处于联邦贫困线200%以下人口比例处于第四四分位数相比，第一四分位数（校正比值比，0.40[95%CI，0.29 - 0.55]）、第二四分位数（校正比值比，0.60[95%CI，0.47 - 0.77]）、第三四分位数（校正比值比，0.76[95%CI，0.63 - 0.91]）；P<0.001），以及与东北部相比的中西部地区（校正比值比，2.15；95%CI，1.13 - 4.07；P = 0.02）、南部地区（校正比值比，2.71；95%CI，1.23 - 5.99；P = 0.01）和西部地区（校正比值比，3.01；95%CI，1.21 - 7.54；P = 0.02）。在单变量模型中，驾车距离与县级糖尿病患病率相关（比值比，1.12；95%CI，1.06 - 1.19；P<0.001），尽管在多变量模型中不显著。出行时间（分钟）也发现了类似的预测因素。
美国糖尿病眼病临床试验地点存在地理分布不均的情况。出行负担较重的人群更有可能居住在农村、低收入且位于东北部以外地区的普查区。
