Marginal Structural Models for Comparing Alternative Treatment Strategies in Ophthalmology using Observational Data

Joffe, Marshall M.; Pistilli, Maxwell; Kempen, John H.

doi:10.3109/09286586.2013.792939

Cited by 5 publications

(2 citation statements)

References 12 publications

(20 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MSMs analyses partially mimic a sequentially randomized trial design and thus allows estimation of the marginal treatment effect through the application of IPTWs. 35 However, similar to the findings from the CC approach, we noticed that the statistical power to detect a pre-identified non-null treatment effect using IPW was decreased compared to MI in our simulated scenarios. This is an important issue since statistical power allows both investigators and readers with information to help interpret potentially null conclusions.…”

Section: Discussionsupporting

confidence: 84%

Missing Data in Marginal Structural Models

et al. 2019

View full text Add to dashboard Cite

Background: The use of marginal structural models (MSMs) to adjust for time-varying confounding has increased in epidemiologic studies. However, in the setting of MSMs, recommendations for how best to handle missing data are contradictory. We present a plasmode simulation study to compare the validity and precision of MSMs estimates using complete case analysis (CC), multiple imputation (MI), and inverse probability weighting (IPW) in the presence of missing data on time-independent and time-varying confounders. Methods: Simulations were based on a cohort sub-study using data from the Osteoarthritis Initiative which estimated the marginal causal effect of intra-articular injection use on yearly changes in knee pain. We simulated 81 scenarios with parameter values varied on missing mechanisms (MCAR, MAR, and MNAR), percentages of missing (10%, 20%, and 30%), type of confounders (time-independent, time-varying, either or both), and analytical approaches (CC, IPW, and MI). The performance of CC, IPW, and MI methods was compared using relative bias, mean squared error (MSE) of the estimates of interest, and empirical power.

show abstract

Section: Discussionsupporting

confidence: 84%

Missing Data in Marginal Structural Models

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Each outcome (≥21, ≥30, rise by ≥10mmHg) was modeled independently for both prevalence and incidence analyses; 95% confidence intervals were presented for all point estimates following the convention where the lower and upper bounds of the confidence interval are presented as subscripts before and after the estimate respectively. 22 All statistical analyses were performed with SAS software version 9.4 (SAS Inc., Cary, NC).…”

Section: Methodsmentioning

confidence: 99%

The Risk of Intraocular Pressure Elevation in Pediatric Noninfectious Uveitis

et al. 2015

Self Cite

View full text Add to dashboard Cite

Purpose To characterize the risk and risk factors for intraocular pressure (IOP) elevation in pediatric non-infectious uveitis. Design Multi-center retrospective cohort study. Participants Nine hundred sixteen children (1593 eyes) <18 years old at presentation with non-infectious uveitis followed between January 1978 through December 2007 at five academic uveitis centers in United States. Methods Medical records review by trained, certified experts. Main outcome measures Prevalence and incidence of IOP≥21 and ≥30mmHg and incidence of a rise in IOP by ≥10mmHg. To avoid under ascertainment, outcomes were counted as present when IOP-lowering therapies were in use. Results Initially 251 (15.8%) and 46 eyes (2.9%) had IOP≥21 and ≥30mmHg, respectively. Factors associated with presenting IOP elevation included age 6–12 years (versus other pediatric ages), prior cataract surgery (adjusted odds ratio≥21mmHg [aOR21]=2.42, P=0.01), pars plana vitrectomy (adjusted odds ratio≥30mmHg[aOR30]=11.1, P=0.03), duration of uveitis ≥6 months (aORs30 up to 11.8, P<0.001), contralateral IOP elevation (aOR21=16.9, aOR30=8.29; each P<0.001), visual acuity worse than 20/40 (aORs21 up to 1.73, P=0.02; aORs30 up to 2.81 P=0.03), and topical corticosteroid use (aORs up to 8.92, P<0.001 in a dose-response relationship). The median follow-up was 1.25 years (interquartile range 0.4–3.66). The estimated risk of any observed IOP elevation to ≥21 mmHg, ≥30 mmHg and of a rise in IOP by ≥10mmHg was 33.4%, 14.8% and 24.4% respectively within 2 years. Factors associated with IOP elevation included pars plana vitrectomy (adjusted hazard ratio≥21mmHg[aHR21]=3.36, P<0.001), contralateral IOP elevation (aHRs up to 9.54, P<0.001), the use of topical (aHRs up to 8.77 that followed a dose-response relationship, P<0.001), periocular (aHRs up to 7.96, P<0.001) and intraocular (aHRs up to 19.7, P<0.001) corticosteroids. Conclusions IOP elevation affects a large minority of children with non-infectious uveitis. Statistically significant risk factors include IOP elevation or use of IOP-lowering treatment in the contralateral eye and local corticosteroid use – that demonstrated a dose-and route of administration-dependent relationship. In contrast, use of immunosuppressive drug therapy did not increase such risk. Pediatric eyes with non-infectious uveitis should be followed closely for IOP elevation when strong risk factors such as the use of local corticosteroids and contralateral IOP elevation are present.

show abstract