Notice of Retraction and Replacement. Tseng et al. Association of cataract surgery with mortality in older women: findings from the Women’s Health Initiative. JAMA Ophthalmol. 2018;136(1):3-10 | JAMA Ophthalmology | JAMA Network
[Skip to Navigation]
Sign In
Comment & Response
November 2018

Notice of Retraction and Replacement. Tseng et al. Association of cataract surgery with mortality in older women: findings from the Women’s Health Initiative. JAMA Ophthalmol. 2018;136(1):3-10

Author Affiliations
  • 1Center for Community Outreach and Policy, Stein Eye Institute, David Geffen School of Medicine, UCLA (University of California, Los Angeles)
  • 2Los Angeles Biomedical Research Institute, Harbor-UCLA Medical Center, Torrance
  • 3Department of Biostatistics, Fielding School of Public Health, UCLA
  • 4Department of Epidemiology, University of Pittsburgh, Pittsburgh, Pennsylvania
  • 5Department of Medicine, University of Massachusetts Medical School, Worcester
  • 6Department of Preventive Medicine, University of Tennessee Health Sciences Center, Memphis
  • 7Division of Health Policy and Management, School of Public Health, University of Minnesota, Minneapolis
  • 8Department of Epidemiology, Fielding School of Public Health, UCLA
JAMA Ophthalmol. 2018;136(11):1313-1314. doi:10.1001/jamaophthalmol.2018.3347

We write to report a serious error and incorrect analysis that has affected the results of our study, “Association of Cataract Surgery With Mortality in Older Women: Findings from the Women’s Health Initiative,” that was published online on October 26, 2017, and in the January 2018 print issue of JAMA Ophthalmology.1 In that article, we reported the results of a cohort study using data from the Women’s Health Initiative randomized clinical trial and data in the Medicare claims database. Participants were women 65 years or older with a diagnosis of cataract. We assessed the association between cataract surgery and total and cause-specific mortality in older women with cataract, and we erroneously reported that cataract surgery was associated with lower risk for total and cause-specific mortality.

After publication of the article, a reader contacted us and questioned the results from our time-varying covariate model, which was 1 of the models used in our study. In response, 2 of us (V.L.T. and F.Y.) examined the analysis in which cataract surgery was treated as a time-varying covariate and found that the time-to-event variable for the exposed group (women with cataract surgery) was mistakenly defined as time since the date of surgery rather than the date of diagnosis for this model. When we corrected the time-to-event duration for this group, results from the model with cataract surgery as a time-varying covariate differed from the results originating from the conventional Cox model we had used, and cataract surgery was no longer associated with a lower risk for mortality.

We separately double checked and reran all analyses from the remainder of the study and confirm that there are no additional technical errors. However, because we did not observe differing results when computing a time-varying exposure model in our initial analysis for all-cause mortality due to the above identified programming error, we incorrectly and regrettably concluded that it would suffice to only compute and report the conventional Cox models for the cause-specific outcomes. After correcting the programming error, we are now aware of a likely immortal time bias for all originally reported analyses. In epidemiology, an immortal time bias refers to a time interval between study inclusion and exposure onset where no outcome events can be observed for the exposed group, which may lead to underestimation of the outcome for exposed participants. In our study, this refers to the time between cataract diagnosis and surgery for participants with cataract surgery, during which mortality could not have occurred. This is especially important when cataract surgery may have been delayed for a year or more because the patient’s health status may change during the time interval. We have rerun all our analyses allowing for the time-varying exposure that results in a markedly different association of cataract surgery with all-cause or cause-specific mortality.

We have corrected the results of the original article.1 These results now indicate that cataract surgery was associated with a higher risk for all-cause mortality and mortality attributed to multiple causes of death, except for neurologic-related causes of death. Our approach now accounts for the time from diagnosis to actual surgery reflecting that women at the time of diagnosis initially continue to contribute observational time to the nonsurgery group.

We regret the errors in our definition of the time-to-event for the women with cataract surgery and in our incorrect analyses and any confusion this has caused the readers of our article.1 We have requested that our article be retracted and replaced with the correct data, analyses, and findings. The abstract, text, and Tables 2, 3, and 5 in the original article have been corrected and replaced online. An additional online supplement has been added that includes a version of the original article with the errors highlighted and a version of the replacement article with the corrections highlighted.

Back to top
Article Information

Corresponding Author: Anne L. Coleman, MD, PhD, Stein Eye Institute, David Geffen School of Medicine, UCLA, 100 Stein Plaza, Room 2-118, Los Angeles, CA 90095 (coleman@jsei.ucla.edu).

Published Online: August 23, 2018. doi:10.1001/jamaophthalmol.2018.3347

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

References
1.
Tseng  VL, Chlebowski  RT, Yu  F,  et al.  Association of cataract surgery with mortality in older women: findings from the Women’s Health Initiative.  JAMA Ophthalmol. 2018;136(1):3-10. doi:10.1001/jamaophthalmol.2017.4512PubMedGoogle ScholarCrossref
×