■自2000年代初以来,英国国家卫生服务(NHS)已扩大了在私立医院提供公共资助的医疗服务,以满足不断增长的选择性医疗需求。这项研究旨在比较患者的结果,私立和NHS医院提供选择性髋关节和膝关节置换术时的效率和不良事件。
■我们进行了一项基于人群的队列研究,包括≥18岁的患者,在2016年1月1日至2019年3月31日期间,在英格兰的私立和NHS医院接受公共资助的选择性髋关节或膝关节置换。估计了三种患者结局指标的比较概率(院内死亡率,紧急再入院28天,医院转移),两项有效措施(术前住院时间(LOS)>0天和术后LOS>2天),和四个不良事件(医院相关感染,药物不良反应,压疮,静脉血栓栓塞)。Probit回归用于调整可观察到的混杂因素,然后进行工具变量(IV)分析,以解决患者水平未观察到的混杂因素。然后使用倾向得分匹配作为稳健性检查。
■我们的研究样本包括私立医院的169,232名患者,NHS医院的262,659名患者。从probit回归估计表明,在私立医院接受治疗与降低住院死亡率的可能性相关(-0.0009,95%CI-0.0010,-0.0007),急诊再入院(-0.0181,95%CI-0.0191,-0.0172),医院转院(-0.0076,95%CI-0.0084,-0.0068),术后LOS延长(-0.1174,95%CI-0.1547,-0.0801),医院相关感染(-0.0115,95%CI-0.0123,-0.0107),药物不良反应(-0.0051,95%CI-0.0056,-0.0046),压疮(-0.0017,95%CI-0.0019,-0.0014),和静脉血栓栓塞(-0.0027,95%CI-0.0031,-0.0022)。IV分析在私立医院和NHS医院之间没有显着差异,除了私立医院发生医院相关感染的概率较低(-0.0057,95%CI-0.0081,-0.0032),在私立医院中,术后LOS延长的可能性更大(0.2653,95%CI0.1833,0.3472)。倾向得分匹配产生与probit回归相似的结果。
■我们的研究结果表明,在使用probit回归或倾向得分匹配时,私立医院和NHS医院之间在患者层面存在潜在的不可观察的混淆。
■这项研究没有获得任何专项资金。
UNASSIGNED: Since the early 2000s, the National Health Service (NHS) in England has expanded provision of publicly funded care in private hospitals as a strategy to meet growing demand for elective care. This
study aims to compare patient outcomes, efficiency and adverse events in private and NHS hospitals when providing elective hip and knee replacement.
UNASSIGNED: We conducted a population-based cohort
study including patients ≥18 years, undergoing a publicly funded elective hip or knee replacement in private and NHS hospitals in England between January 1st 2016 and March 31st 2019. Comparative probability was estimated for three patient outcome measures (in-hospital mortality, emergency readmissions with 28 days, hospital transfers), two efficiency measures (pre-operative length of stay (LOS) >0 day and post-operative LOS >2 days), and four adverse events (hospital-associated infection, adverse drug reactions, pressure ulcers, venous thromboembolism). Probit regression was used to adjust for observable confounding followed by instrumental variable (IV) analyses to also account for unobserved confounding at the patient-level. Propensity score matching was then used as a robustness check.
UNASSIGNED: Our
study sample included 169,232 patients in private hospitals, and 262,659 patients in NHS hospitals. Estimates from probit regression indicated that treatment in private hospital was associated with reduced probability of in-hospital mortality (-0.0009, 95% CI -0.0010, -0.0007), emergency readmissions (-0.0181, 95% CI -0.0191, -0.0172), hospital transfers (-0.0076, 95% CI -0.0084, -0.0068), prolonged post-operative LOS (-0.1174, 95% CI -0.1547, -0.0801), hospital-associated infection (-0.0115, 95% CI -0.0123, -0.0107), adverse drug reactions (-0.0051, 95% CI -0.0056, -0.0046), pressure ulcers (-0.0017, 95% CI -0.0019, -0.0014), and venous thromboembolism (-0.0027, 95% CI -0.0031, -0.0022). IV analyses produced no significant differences between private and NHS hospitals, except for lower probability in private hospitals of hospital-associated infection (-0.0057, 95% CI -0.0081, -0.0032), and greater probability in private hospitals of prolonged post-operative LOS (0.2653, 95% CI 0.1833, 0.3472). Propensity score matching produced similar results to probit regression.
UNASSIGNED: Our findings indicate there is potentially important unobservable confounding at the patient-level between private and NHS hospitals not adjusted for when using probit regression or propensity score matching.
UNASSIGNED: This research did not receive any dedicated funding.