eTable 1.International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) and Current Procedural Terminology (CPT) Codes Included in Definition of Cataract and Related Procedures
eTable 2. Diseases Included in Mortality Attributed to Specific Systemic Conditions
eTable 3.International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) Codes Included in the Definition of Comorbidities
eTable 4. Systemic and Ocular Comorbidities Collected within the Women’s Health Initiative (WHI) and Medicare Databases
eTable 5. Baseline Characteristics of Patients With Cataract in the Women’s Health Initiative from 1993-2013 (n = 74 044)
eTable 6. Source of Diagnosis for Selected Systemic Comorbidities Dually Assessed in the Women’s Health Initiative (WHI) and Medicare Databases
Retracted article with errors highlighted
Replacement article with corrections highlighted
Customize your JAMA Network experience by selecting one or more topics from the list below.
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.2017.4512
How is cataract surgery associated with mortality outcomes in women with cataract?
In this prospective cohort study of 74 044 participants of the Women's Health Initiative with cataract, cataract surgery was associated with increased risk for all-cause mortality and mortality attributed to vascular, cancer-related, accidental, pulmonary, and infectious causes, even after adjusting for demographics and systemic and ocular comorbidities.
In women with cataract, cataract surgery may be associated with higher mortality risk related to multiple types of systemic illness. However, this study cannot determine whether death events are attributable to the surgery itself, a result of surgery postponement, or other factors.
Previous studies have suggested an association between cataract surgery and decreased risk for all-cause mortality potentially through a mechanism of improved health status and functional independence, but the association between cataract surgery and cause-specific mortality has not been previously studied and is not well understood.
To examine the association between cataract surgery and total and cause-specific mortality in older women with cataract.
Design, Setting, and Participants
This prospective cohort study included nationwide data collected from the Women’s Health Initiative (WHI) clinical trial and observational study linked with the Medicare claims database. Participants in the present study were 65 years or older with a diagnosis of cataract in the linked Medicare claims database. The WHI data were collected from January 1, 1993, through December 31, 2015. Data were analyzed for the present study from July 1, 2014, through September 1, 2017.
Cataract surgery as determined by Medicare claims codes.
Main Outcomes and Measures
The outcomes of interest included all-cause mortality and mortality attributed to vascular, cancer, accidental, neurologic, pulmonary, and infectious causes. Mortality rates were compared by cataract surgery status using the log-rank test and Cox regression models with time-varying covariate cataract surgery status adjusting for demographics, systemic and ocular comorbidities, smoking, alcohol use, body mass index, and physical activity.
A total of 74 044 women with cataract in the WHI included 41 735 who underwent cataract surgery. Mean (SD) age was 70.5 (4.6) years; the most common ethnicity was white (64 430 [87.0%]), followed by black (5293 [7.1%]) and Hispanic (1723 [2.3%]). The mortality rate was 2.56 per 100 person-years in the cataract surgery group and 1.45 per 100 person-years in the cataract diagnosis group. In covariate-adjusted Cox models, cataract surgery was associated with higher all-cause mortality (adjusted hazards ratio [AHR], 1.07; 95% CI, 1.02-1.11) as well as higher mortality specific to vascular (AHR, 1.36; 95% CI, 1.26-1.46), cancer (AHR, 1.27; 95% CI, 1.18-1.38), accidental (AHR, 1.36; 95% CI, 1.05-1.76), pulmonary (AHR, 1.96; 95% CI, 1.62-2.37), and infectious (AHR, 1.37; 95% CI, 1.14-1.65) diseases. Neurologic death causes were not associated with cataract surgery (AHR, 0.98; 95% CI, 0.83-1.17)
Conclusions and Relevance
In older women with cataract in the WHI, cataract surgery was associated with higher risk for total and cause-specific mortality (except for neurologic causes). Further study of the interplay of cataract surgery, systemic disease, disease-related mortality, and the best timing of when to undergo cataract surgery would be informative for improved patient care.
Cataract surgery is the mainstay of treatment for visually significant cataract, and its primary purpose is to reverse the associated visual impairment.1,2 Previous studies from administrative databases have demonstrated that in addition to visual improvement, cataract surgery was also associated with a lower mortality risk.3-5 Although these studies hypothesized that the observed associations are mediated by improvements in health status and functional independence after cataract surgery, they have not examined whether the association between cataract surgery and lower mortality risk applies to women with different medical conditions that affect overall functioning to varying degrees, such as cancer, cardiovascular disease, and dementia.
The Women’s Health Initiative (WHI) database contains detailed and extensively validated information on demographic, comorbidity, and lifestyle factors in a large cohort of US women with more than 20 years of follow-up.6 Furthermore, the WHI database contains information on total and cause-specific mortality, which is not available in other large databases of patients with cataract surgery in the United States.7 These factors make the WHI an ideal cohort in which to examine the interplay between eye disease and a wide variety of systemic and lifestyle factors in US women. Given these strengths of the WHI database and the need for further understanding of the association between cataract surgery and improved survival, the purpose of the present study is to examine the association between cataract surgery and total and cause-specific mortality in the WHI cohort, with the hypothesis that cataract surgery improves survival among women with cataract in the WHI.
The WHI is a study of US postmenopausal women aged 50 to 79 years. The WHI included 4 overlapping, randomized clinical trials evaluating hormone therapy, diet modification, and calcium and vitamin D supplementation that enrolled 68 132 women and an observational study cohort that enrolled 93 676 women. Details of the study design have been previously published.8-10 The protocols were approved by the institutional review boards at all clinical sites, and all women provided written informed consent. The present study was additionally approved by the institutional review board at UCLA.
Data collection for the main WHI study started in January 1, 1993, and ended December 31, 2005, at 40 clinical centers, and additional observational data were collected in 2 WHI extension studies from January 1, 2005, through December 31, 2010,11 and from January 1, 2010, through December 31, 2015.12 For WHI participants who are also enrolled in Medicare, WHI data are linked on an individual level to all International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM)13 diagnosis codes and Current Procedural Terminology (CPT)14 codes in the administrative Medicare database.
The study population consisted of all WHI participants who had an ICD-9-CM diagnosis code for cataract in the linked Medicare database at 65 years or older (eTable 1 in Supplement 1). Participants younger than 65 years at the timeof cataract diagnosis in Medicare were excluded owing to selective indications for Medicare enrollment before 65 years of age. Participants who were younger than 65 years at the time of WHI enrollment were not included in the study population until the age of cataract diagnosis in the Medicare database, which defined their age at study baseline.
The exposure of interest was cataract surgery. The cataract surgery group consisted of all participants with an ICD-9-CM diagnosis code for cataract and a CPT code for cataract surgery (eTable 1 in Supplement 1). The cataract diagnosis group consisted of all participants with an ICD-9-CM diagnosis code for cataract without a CPT code for cataract surgery. Patients in both groups were followed up starting from the date of earliest diagnostic code for cataract.
The primary outcome of interest was all-cause mortality before December 31, 2015. Death and underlying cause of death were determined at the clinical coordinating center by review of medical records or death certificates and, in some cases, by reports from participants’ relatives or by the National Death Index.15 Of 31 311 total discovered deaths in the WHI cohort as of December 31, 2015, cause of death was ascertained from medical record review for 21 887 participants (69.9%), from the initial cause of death form for 2466 (7.9%), and from the ICD-9-CM code for 6958 (22.2%). In this study, causes of death were grouped into vascular, cancer-related, accident-related, neurologic, pulmonary, and infectious conditions (eTable 2 in Supplement 1).
For the cataract diagnosis group, time to death was calculated as the number of days from cataract diagnosis to death. For the cataract surgery group, time between cataract diagnosis and surgery was classified into follow-up time for cataract diagnosis, and time after surgery was classified into follow-up time for cataract surgery. Participants who did not have a recorded death in the study were censored on the last known date of WHI follow-up or on December 31, 2015, if they were still enrolled.
Demographics that were collected included age at the time of cataract diagnosis in Medicare, race, US region of residence, educational level, annual income, health insurance status at the time of WHI enrollment, use of preventive health services (mammography and Papanicolaou smear), and WHI study arm. Systemic comorbidities in the Charlson Comorbidity Index (CCI)16 were assessed at the time of surgery and the time of diagnosis for the cataract surgery group and at the time of diagnosis for the cataract diagnosis group (eTable 3 in Supplement 1). The CCI is a weighted index of systemic disease burden (range, 0-33), and a higher CCI score indicates a higher burden of systemic disease.16 Ocular comorbidities collected included age-related macular degeneration and glaucoma. A participant was counted as having a systemic or ocular comorbidity if she had the disease based on the WHI indicator variable for the disease, an ICD-9-CM code for the disease in Medicare, or both (eTable 4 in Supplement 1). The WHI verified the following outcomes by central medical record review: all cancers, myocardial infarction, congestive heart failure, stroke, other cardiovascular events, pulmonary emboli, hip fracture, and dementia (only in the WHI Memory Study subcohort).
Because objective visual acuity data were not available, the presence of severe cataract was examined as a proxy for poor vision. Patients with ICD-9-CM codes for subcapsular cataracts, total or mature cataract, hypermature cataract, and combined cataracts were considered to have severe subtypes of cataract (eTable 1 in Supplement 1).
Additional covariates collected included smoking status, alcohol intake, body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), and physical activity. Smoking status and alcohol intake were self-reported at study baseline. We determined BMI based on height and weight measurements from study baseline. Physical activity was based on metabolic equivalent tasks per week, which were calculated from self-reported physical activity habits at baseline and at years 3 and 6.
Data were analyzed for this study from July 1, 2014, through September 1, 2017. Patient characteristics and mortality rates were compared by cataract surgery status descriptively. To compare the source of diagnosis for systemic comorbidities that were ascertained from the WHI and Medicare databases, we performed χ2 tests for the diagnosis source for selected cardiovascular and cancer-related comorbidities. Log-rank tests were used to compare crude mortality rates by surgical status. Cox regression models with time-varying covariate cataract surgery status (exposure) were used to obtain covariate-adjusted associations from cataract surgery and time to death due to any cause and from cataract surgery and time to death attributed to vascular, cancer, accidental, neurologic, pulmonary, and infectious conditions. All demographic, CCI score, ocular comorbidities, cataract severity, smoking status, alcohol intake, physical activity, and BMI data were included as covariates in the model. To account for potential confounding by indication, propensity scores were created by regressing all covariates on cataract surgery status to determine the probability of cataract surgery. Participants were then grouped by propensity score strata, and time to all-cause mortality was compared between participants in the cataract surgery and cataract diagnosis groups within propensity score deciles with Cox models with cataract surgery as a time-varying covariate. P < .05 indicated significance.
Baseline characteristics are summarized in Table 1 and eTable 5 in Supplement 1. The study cohort included 74 044 women (mean [SD] age, 70.5 [4.6] years), of whom 41 735 were in the cataract surgery group and 32 309 were in the cataract diagnosis group. The largest proportion of patients in the cataract surgery and cataract diagnosis groups were aged 65 to 69 years at the time of cataract diagnosis (45.6% and 55.6%, respectively). The remainder of demographic factors were similar in both groups. The most common systemic comorbidity in both groups was solid malignant neoplasm (29.9% in the cataract surgery and 23.6% in the cataract diagnosis groups), and the most common ocular comorbidity was glaucoma (31.6% in the cataract surgery and 17.5% in the cataract diagnosis groups). Higher proportions of most comorbidities were found in the surgery group at the time of surgery. When diagnosis sources for selected systemic comorbidities were compared, 6.2% to 40.8% of comorbidities were diagnosed in the WHI and Medicare databases, whereas the remainder were diagnosed in 1 database only (eTable 6 in Supplement 1). In both groups, most participants were never smokers (51.5% in the cataract surgery and 51.8% in the cataract diagnosis groups, of those with available data), had 1 to less than 7 alcoholic beverages per week (25.8% in the cataract surgery and 27.1% in the cataract diagnosis groups, of those with available data), had a BMI of less than 25 (36.2% in the cataract surgery and 36.7% in the cataract diagnosis groups), and had activity levels of 5 to 12 MET/wk (64.8% in the cataract surgery and 64.5% in the cataract diagnosis groups, of those with available data).
Data on mortality incidence are summarized in Table 2. The crude incidence of all-cause mortality was 2.56 deaths per 100 person-years in the cataract surgery group and 1.45 deaths per 100 person-years in the cataract diagnosis group (P < .001 for log-rank test comparing crude mortality incidence in both groups).
Multivariable associations between cataract surgery and each type of mortality treating cataract surgery as a time-varying covariate are summarized in Table 3. Cataract surgery was associated with a higher risk for all-cause mortality in the unadjusted model (hazards ratio [HR], 1.22; 95% CI, 1.18-1.27) and adjusted model (adjusted HR [AHR], 1.07; 95% CI, 1.02-1.11). Cataract surgery was also associated with higher hazards for all-cause mortality and cause-specific death for every type of death except for neurologic causes of death in the unadjusted and adjusted models. Additional results are given in Table 3.
Results from propensity score analyses are outlined in Table 4 and Table 5. A higher likelihood of receiving cataract surgery at some point during follow-up (dichotomous outcome cataract surgery reported yes/no) was associated with being in the clinical trial arm of the WHI, living outside of the Northeast, having insurance, undergoing routine mammography, having a CCI score higher than 0, having any ocular comorbidity, and having a high level of physical activity. When grouping the women into deciles according to the propensity of having a cataract surgery during follow-up and estimating Cox models for the subgroups so obtained, we observed a high hazard rate ratio for all-cause mortality for the groups of women most likely to receive cataract surgery during follow-up but not in other subgroups (Table 5).
In participants with cataract in the WHI, cataract surgery was associated with a higher risk for total and cause-specific mortality (except for neurologic cause) after adjusting for demographics, systemic and ocular comorbidities, and selected lifestyle factors. Specifically, WHI participants after cataract surgery had relatively higher risks for dying due to any cause and due to vascular, cancer, accidental, pulmonary, and infectious conditions compared with participants with a cataract who did not receive surgery. Our study design does not allow us to distinguish whether the empirically observed increases in the hazard for these death events are attributable to the surgery itself or whether they are a result of individually postponing surgery until a point in time where hazard rates are increasing in addition to the general effects of the covariates we do adjust for, such as age at cataract diagnosis.
A previous study of cataract surgery and mortality in the US Medicare population3 reported that cataract surgery was associated with a lower hazard for all-cause mortality in patients with cataract after adjusting for demographics and systemic and ocular comorbidities, which was also seen in a subgroup analysis of women within the Medicare population. In the present study, we were able to adjust additionally for smoking, alcohol use, BMI, and physical activity. In addition, we used an approach treating cataract surgery status as a time-varying covariate, and our overall results suggested that cataract surgery is associated with higher all-cause and (most) cause-specific causes of mortality. This approach 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. While the empirically observed increases in the hazard rates for all-cause or cause-specific mortality might be attributable to the surgery itself, other possible explanations remain: our results might be affected by unobserved confounding that we are not aware of, as is typical for observational studies. Furthermore, the empirically observed increases in the hazard rates might be the result of individually postponing surgery until a point in time where hazard rates would increase in addition to the general effects of the covariates we do adjust for, such as age at cataract diagnosis.
To our knowledge, this study is the first to examine the association between cataract surgery and cause-specific mortality, and results from time-varying covariate Cox models suggest that cataract surgery is associated with a higher hazard rate for all-cause and cause-specific mortality.
In the present study, a mediational analysis was not performed to examine potential mechanisms to explain the association between cataract surgery and mortality from different systemic causes. However, we hypothesize that the mechanism of association is multifactorial and can vary across systemic conditions. The cataract surgery group demonstrated a higher mortality rate compared to the cataract diagnosis group, as well as an overall sicker systemic disease profile; we suspect that these differences cannot be explained by cataract surgery alone and that patients with cataract who do and do not receive surgery have inherent demographic, socioeconomic, and systemic differences that must be further explored in future studies.
Although limited studies examine the association between cataract surgery and systemic conditions, previous studies17-19 have demonstrated associations between cataract surgery and lower risk for fall and fracture, and others20-22 have demonstrated that patients with cataract have higher scores on standardized cognition assessments after cataract surgery. However, these studies did not examine the associations between these systemic conditions and eventual mortality. Because our observed elevated hazard rates might be a result of the individual timing of the surgery, which is elective for the respective patient, studies specifically targeting at which point in life cataract surgery is best performed hold promise for further insight into the benefits or harms of cataract surgery. Also, additional studies could specifically examine the association of cataract surgery with factors such as quality of life, activities of daily living, access to medical care, and mental health.
This study is mainly limited by its observational nature. Participation in the WHI is voluntary, and generalizability of findings from this cohort may be limited by differential participant dropout. In addition, because the WHI cohort is all female, findings from this study may not be generalizable to male patients, and additional studies of cataract surgery and mortality in a male population would be beneficial. Furthermore, WHI participants are long-term study participants, and their demographic and systemic disease profiles may differ from those of the general population of women in the United States. Another finding of interest in this study that may limit its generalizability is the unusually high proportion of patients with systemic comorbidities compared with previous studies,3 especially in the cataract surgery group. We hypothesize that the explanation for these differences is multifactorial and mainly related to the ability to combine the WHI and Medicare databases, which may lead to increased detection of comorbidities compared with a single data collection modality that was supported by the significant proportion of cardiovascular and cancer-related comorbidities that were only diagnosed in 1 of the 2 databases between WHI and Medicare. In addition, follow-up with primary care physicians may have been better among WHI participants who have proven reliable follow-up with regular study visits, which would lead to increased detection of systemic disease. The observational data in this study are also subject to misclassification bias, including the accuracy of cause-of-death data. Previous studies have demonstrated that cardiovascular disease is overrepresented as a cause of death,23 in addition to suggesting high rates of misclassification of causes of death from other sources of illness.24-26 In the present study, higher rates of metastatic cancer but lower rates of cancer-related deaths in the cataract surgery vs cataract diagnosis group were found, suggesting that our findings should be interpreted in light of the inherent limitations of cause-of-death data. In addition, unmeasured confounding is a possibility. Finally, although higher mortality rates were observed in the cataract surgery group, they do not prove a causal association between cataract surgery and increased mortality, and additional study of the mechanisms and interplay between these associations is warranted as is further study of the timing of cataract surgery for those women who ultimately elect to have such a surgery.
This study found associations between cataract surgery and higher risks for total and several types of cause-specific mortality in participants of the WHI. Our study design does not allow us to distinguish whether the empirically observed increases in the hazard for these death events are attributable to the surgery itself or whether they are a result of individually postponing surgery until a point in time where hazard rates are increasing in addition to the general effects of the covariates we do adjust for, such as age at cataract diagnosis. Further studies of the associations cataract surgery, systemic disease, and disease-specific mortality would be informative for improving patient selection and use of cataract surgery and for improved understanding of the benefits of cataract surgery beyond vision improvement.
Accepted for Publication: September 11, 2017.
Retraction and Replacement: This article was retracted and replaced on August 23, 2018, to
fix errors throughout the article and
tables (see Supplement 2
or the retracted article with errors highlighted and Supplement 3 for the replacement article with corrections highlighted).
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 (firstname.lastname@example.org).
Published Online: October 26, 2017. doi:10.1001/jamaophthalmol.2017.4512
Author Contributions: Dr Tseng had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Tseng, Chlebowski, Coleman.
Acquisition, analysis, or interpretation of data: Tseng, Chlebowski, Yu, Cauley, Li, Thomas, Virnig.
Drafting of the manuscript: Tseng.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Tseng, Yu, Li, Thomas.
Obtained funding: Tseng, Chlebowski, Cauley.
Administrative, technical, or material support: Tseng, Cauley.
Study supervision: Chlebowski, Coleman.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was supported by the Center of Community Outreach and Policy, Stein Eye Institute, UCLA, and the National Institutes of Health Loan Repayment Program. The Women’s Health Initiative program is supported by contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C from the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services.
Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; in the collection, analysis and interpretation of the data; or in the preparation, review or approval of the manuscript.
Additional Contributions: We thank the original investigators of the Women’s Health Initiative. The list of investigators can be found at the following link: https://www.whi.org/researchers/Documents%20%20Write%20a%20Paper/WHI%20Investigator%20Short%20List.pdf.