학술논문
Estimation of treatment effects in observational stroke care data: comparison of statistical approaches
Document Type
Report
Author
Amini, Marzyeh; van Leeuwen, Nikki; Eijkenaar, Frank; van de Graaf, Rob; Samuels, Noor; van Oostenbrugge, Robert; van den Wijngaard, Ido R.; van Doormaal, Pieter Jan; Roos, Yvo B. W. E. M.; Majoie, Charles; Roozenbeek, Bob; Dippel, Diederik; Burke, James; Lingsma, Hester F.; Dippel, Diederik W. J.; van der Lugt, Aad; Majoie, Charles B. L. M.; van Oostenbrugge, Robert J.; van Zwam, Wim H.; Boiten, Jelis; Vos, Jan Albert; Brouwer, Josje; den Hartog, Sanne J.; Hinsenveld, Wouter H.; Kappelhof, Manon; Compagne, Kars C. J.; Goldhoorn, Robert-Jan B.; Mulder, Maxim J. H. L.; Jansen, Ivo G. H.; van Es, Adriaan C. G. M.; Emmer, Bart J.; Coutinho, Jonathan M.; Schonewille, Wouter J.; Wermer, Marieke J. H.; van Walderveen, Marianne A. A.; Staals, Julie; Hofmeijer, Jeannette; Martens, Jasper M.; Lycklama à Nijeholt, Geert J.; de Bruijn, Sebastiaan F.; van Dijk, Lukas C.; van der Worp, H. Bart; Lo, Rob H.; van Dijk, Ewoud J.; Boogaarts, Hieronymus D.; de Vries, J.; de Kort, Paul L. M.; van Tuijl, Julia; Peluso, Jo Jo P.; Fransen, Puck; van den Berg, Jan S. P.; van Hasselt, Boudewijn A. A. M.; Aerden, Leo A. M.; Dallinga, René J.; Uyttenboogaart, Maarten; Eschgi, Omid; Bokkers, Reinoud P. H.; Schreuder, Tobien H. C. M. L.; Heijboer, Roel J. J.; Keizer, Koos; Yo, Lonneke S. F.; den Hertog, Heleen M.; Sturm, Emiel J. C.; Brouwers, Paul; Sprengers, Marieke E. S.; Jenniskens, Sjoerd F. M.; van den Berg, René; Yoo, Albert J.; Beenen, Ludo F. M.; Postma, Alida A.; Roosendaal, Stefan D.; van der Kallen, Bas F. W.; Bot, Joost; van Doormaal, Pieter-Jan; Meijer, Anton; Ghariq, Elyas; van Proosdij, Marc P.; Krietemeijer, G. Menno; Peluso, Jo P.; Lo, Rob; Gerrits, Dick; Dinkelaar, Wouter; Appelman, Auke P. A.; Hammer, Bas; Pegge, Sjoert; van der Hoorn, Anouk; Vinke, Saman; Flach, H. Zwenneke; el Ghannouti, Naziha; Sterrenberg, Martin; Puppels, Corina; Pellikaan, Wilma; Sprengers, Rita; Elfrink, Marjan; Simons, Michelle; Vossers, Marjolein; de Meris, Joke; Vermeulen, Tamara; Geerlings, Annet; van Vemde, Gina; Simons, Tiny; van Rijswijk, Cathelijn; Messchendorp, Gert; Nicolaij, Nynke; Bongenaar, Hester; Bodde, Karin; Kleijn, Sandra; Lodico, Jasmijn; Droste, Hanneke; Wollaert, Maureen; Verheesen, Sabrina; Jeurrissen, D.; Bos, Erna; Drabbe, Yvonne; Sandiman, Michelle; Aaldering, Nicoline; Zweedijk, Berber; Khalilzada, Mostafa; Vervoort, Jocova; Ponjee, Eva; Romviel, Sharon; Kanselaar, Karin; Barning, Denn; Venema, Esmee; Chalos, Vicky; Geuskens, Ralph R.; van Straaten, Tim; Ergezen, Saliha; Harmsma, Roger R. M.; Muijres, Daan; de Jong, Anouk; Berkhemer, Olvert A.; Boers, Anna M. M.; Huguet, J.; Groot, P. F. C.; Mens, Marieke A.; van Kranendonk, Katinka R.; Treurniet, Kilian M.; Tolhuisen, Manon L.; Alves, Heitor; Weterings, Annick J.; Kirkels, Eleonora L. F.; Voogd, Eva J. H. F.; Schupp, Lieve M.; Collette, Sabine; Groot, Adrien E. D.; LeCouffe, Natalie E.; Konduri, Praneeta R.; Prasetya, Haryadi; Arrarte-Terreros, Nerea; Ramos, Lucas A.
Source
BMC Medical Research Methodology. April 10, 2022, Vol. 22 Issue 1
Subject
Language
English
ISSN
1471-2288
Abstract
Author(s): Marzyeh Amini[sup.1] , Nikki van Leeuwen[sup.1] , Frank Eijkenaar[sup.2] , Rob van de Graaf[sup.3,4] , Noor Samuels[sup.1,3,4] , Robert van Oostenbrugge[sup.5] , Ido R. van den Wijngaard[sup.6] , Pieter [...]
Introduction Various statistical approaches can be used to deal with unmeasured confounding when estimating treatment effects in observational studies, each with its own pros and cons. This study aimed to compare treatment effects as estimated by different statistical approaches for two interventions in observational stroke care data. Patients and methods We used prospectively collected data from the MR CLEAN registry including all patients (n = 3279) with ischemic stroke who underwent endovascular treatment (EVT) from 2014 to 2017 in 17 Dutch hospitals. Treatment effects of two interventions - i.e., receiving an intravenous thrombolytic (IVT) and undergoing general anesthesia (GA) before EVT - on good functional outcome (modified Rankin Scale [less than or equai to]2) were estimated. We used three statistical regression-based approaches that vary in assumptions regarding the source of unmeasured confounding: individual-level (two subtypes), ecological, and instrumental variable analyses. In the latter, the preference for using the interventions in each hospital was used as an instrument. Results Use of IVT (range 66-87%) and GA (range 0-93%) varied substantially between hospitals. For IVT, the individual-level (OR ~ 1.33) resulted in significant positive effect estimates whereas in instrumental variable analysis no significant treatment effect was found (OR 1.11; 95% CI 0.58-1.56). The ecological analysis indicated no statistically significant different likelihood ([beta] = - 0.002%; P = 0.99) of good functional outcome at hospitals using IVT 1% more frequently. For GA, we found non-significant opposite directions of points estimates the treatment effect in the individual-level (ORs ~ 0.60) versus the instrumental variable approach (OR = 1.04). The ecological analysis also resulted in a non-significant negative association (0.03% lower probability). Discussion and conclusion Both magnitude and direction of the estimated treatment effects for both interventions depend strongly on the statistical approach and thus on the source of (unmeasured) confounding. These issues should be understood concerning the specific characteristics of data, before applying an approach and interpreting the results. Instrumental variable analysis might be considered when unobserved confounding and practice variation is expected in observational multicenter studies. Keywords: Unmeasured confounding, Confounding by indication, Statistical approaches, Ecological-analysis, Instrumental variable, Acute ischemic stroke, Intravenous thrombolysis, General anesthesia
Introduction Various statistical approaches can be used to deal with unmeasured confounding when estimating treatment effects in observational studies, each with its own pros and cons. This study aimed to compare treatment effects as estimated by different statistical approaches for two interventions in observational stroke care data. Patients and methods We used prospectively collected data from the MR CLEAN registry including all patients (n = 3279) with ischemic stroke who underwent endovascular treatment (EVT) from 2014 to 2017 in 17 Dutch hospitals. Treatment effects of two interventions - i.e., receiving an intravenous thrombolytic (IVT) and undergoing general anesthesia (GA) before EVT - on good functional outcome (modified Rankin Scale [less than or equai to]2) were estimated. We used three statistical regression-based approaches that vary in assumptions regarding the source of unmeasured confounding: individual-level (two subtypes), ecological, and instrumental variable analyses. In the latter, the preference for using the interventions in each hospital was used as an instrument. Results Use of IVT (range 66-87%) and GA (range 0-93%) varied substantially between hospitals. For IVT, the individual-level (OR ~ 1.33) resulted in significant positive effect estimates whereas in instrumental variable analysis no significant treatment effect was found (OR 1.11; 95% CI 0.58-1.56). The ecological analysis indicated no statistically significant different likelihood ([beta] = - 0.002%; P = 0.99) of good functional outcome at hospitals using IVT 1% more frequently. For GA, we found non-significant opposite directions of points estimates the treatment effect in the individual-level (ORs ~ 0.60) versus the instrumental variable approach (OR = 1.04). The ecological analysis also resulted in a non-significant negative association (0.03% lower probability). Discussion and conclusion Both magnitude and direction of the estimated treatment effects for both interventions depend strongly on the statistical approach and thus on the source of (unmeasured) confounding. These issues should be understood concerning the specific characteristics of data, before applying an approach and interpreting the results. Instrumental variable analysis might be considered when unobserved confounding and practice variation is expected in observational multicenter studies. Keywords: Unmeasured confounding, Confounding by indication, Statistical approaches, Ecological-analysis, Instrumental variable, Acute ischemic stroke, Intravenous thrombolysis, General anesthesia