PDF(Pubmed)
Abstract：
BACKGROUND: The English National Health Service has multiple waiting time standards relating to cancer diagnosis and treatment. Targets can have unintended effects, such as prioritisation based on targets instead of clinical need. In this case, a `threshold effect\' will appear as a spike in hospitals just meeting the target.
METHODS: We conducted a retrospective study of publicly available cancer waiting time data, including a 2-week wait for a specialist appointment, a 31-day decision to first treatment and a 62-day referral to treatment standard that attracted a financial penalty. We examined the performance of hospital trusts against these targets by financial year to look for threshold effects, using Cattaneo et al. manipulation density test.
RESULTS: Trust performance against cancer waiting targets declined over time, and this trend accelerated since the start of the Covid-19 pandemic. Statistical evidence of a threshold effect for the 2-week and 31-day standard was only present in a few years. However, there was strong statistical evidence of a threshold effect for the 62-day standard across all financial years (p < 0.01).
CONCLUSIONS: The data suggests that the effect of threshold targets alters hospital behaviour at target levels but does not do so equally for all standards. Evidence of threshold effects for the 62-day standard was particularly strong, possibly due to some combination of a smaller volume of eligible patients, a larger penalty, multiple waypoints where hospitals can intervene, baseline performance against the target and where the target is set (i.e. how much headroom is available). RCTs of the use of threshold targets and of different designs for such targets in the future would be extremely informative.
METHODS: We conducted a retrospective study of publicly available cancer waiting time data, including a 2-week wait for a specialist appointment, a 31-day decision to first treatment and a 62-day referral to treatment standard that attracted a financial penalty. We examined the performance of hospital trusts against these targets by financial year to look for threshold effects, using Cattaneo et al. manipulation density test.
RESULTS: Trust performance against cancer waiting targets declined over time, and this trend accelerated since the start of the Covid-19 pandemic. Statistical evidence of a threshold effect for the 2-week and 31-day standard was only present in a few years. However, there was strong statistical evidence of a threshold effect for the 62-day standard across all financial years (p < 0.01).
CONCLUSIONS: The data suggests that the effect of threshold targets alters hospital behaviour at target levels but does not do so equally for all standards. Evidence of threshold effects for the 62-day standard was particularly strong, possibly due to some combination of a smaller volume of eligible patients, a larger penalty, multiple waypoints where hospitals can intervene, baseline performance against the target and where the target is set (i.e. how much headroom is available). RCTs of the use of threshold targets and of different designs for such targets in the future would be extremely informative.
摘要：
背景：英国国家卫生服务有多个与癌症诊断和治疗相关的等待时间标准。目标可能会产生意想不到的影响，例如基于目标而不是临床需要的优先级。在这种情况下,在刚刚达到目标的医院中，“阈值效应”将出现峰值。
方法：我们对公开的癌症等待时间数据进行了回顾性研究，包括2周等待专家预约,31天的首次治疗决定和62天的治疗标准转诊，导致经济处罚。我们按财政年度检查了医院信托对这些目标的表现，以寻找门槛效应，使用Cattaneo等人。操纵密度试验。
结果：对癌症等待目标的信任表现随着时间的推移而下降，自新冠肺炎大流行开始以来，这一趋势加速了。2周和31天标准的阈值效应的统计证据仅在几年内出现。然而,有强有力的统计证据表明,所有财政年度的62天标准都有阈值效应(p<0.01).
结论：数据表明，阈值目标的影响改变了目标水平的医院行为，但并非所有标准都一样。62天标准的阈值效应证据特别强，可能是由于少量合格患者的某种组合，更大的惩罚，医院可以介入的多个航路点，相对于目标和目标设置的基准性能(即有多少净空可用)。将来使用阈值目标和针对此类目标的不同设计的RCT将提供极大的信息。
方法：我们对公开的癌症等待时间数据进行了回顾性研究，包括2周等待专家预约,31天的首次治疗决定和62天的治疗标准转诊，导致经济处罚。我们按财政年度检查了医院信托对这些目标的表现，以寻找门槛效应，使用Cattaneo等人。操纵密度试验。
结果：对癌症等待目标的信任表现随着时间的推移而下降，自新冠肺炎大流行开始以来，这一趋势加速了。2周和31天标准的阈值效应的统计证据仅在几年内出现。然而,有强有力的统计证据表明,所有财政年度的62天标准都有阈值效应(p<0.01).
结论：数据表明，阈值目标的影响改变了目标水平的医院行为，但并非所有标准都一样。62天标准的阈值效应证据特别强，可能是由于少量合格患者的某种组合，更大的惩罚，医院可以介入的多个航路点，相对于目标和目标设置的基准性能(即有多少净空可用)。将来使用阈值目标和针对此类目标的不同设计的RCT将提供极大的信息。
