sign in sign up
<<
Home , Cataract surgery

Surgery Complications and Revisit Rates Among Three States

SUZANN PERSHING, DOUGLAS E. MORRISON, AND TINA HERNANDEZ-BOUSSARD

PURPOSE: To characterize population-based 30-day surgery readmissions, and suggest a relationship between procedure-related readmissions (revisits) following black or Hispanic race, Medicaid insurance, and diabetes cataract surgery. associated with higher risk for cataract surgery SETTING: Ambulatory cataract surgery performed in complications. (Am J Ophthalmol 2016;171: California, Florida, or New York. 130–138. Ó 2016 Elsevier Inc. All rights reserved.) DESIGN: Retrospective cohort study. METHODS: This study used all-capture state adminis- trative datasets. Cataract procedures from California, ATARACT SURGERY IS THE SINGLE MOST FREQUENT Florida, and New York state ambulatory surgery settings surgical procedure in developed countries, were identified using ICD-9-CM and CPT codes. Thirty- C including the United States, where approximately day readmissions (revisits) were identified in inpatient, 3 million cases are performed annually on ambulatory, and emergency department settings across patients. It is presently the largest single source of Medicare each state. expenditures, and rates of surgery are rising. Between 1990 RESULTS: Across the 3 states, the all-cause 30-day and 2010, incidence of cataract surgery increased between 1–4 readmission rate was 6.0% and the procedure-related 2.5- and 6.5-fold, varying by region. This is a trend likely readmission (revisit) rate was 1.0%. Procedure-related only to continue, given improved technology, an aging revisits were highest for patients aged 20–29 (2.9%) population with higher expectations for visual function, and 30–39 (2.3%) and lowest for patients aged 70–79 and expanding surgical criteria, including lower visual (0.9%). Multivariate associations between clinical char- acuity thresholds for surgery, more bilateral surgery, and 5–7 acteristics and 30-day procedure-related revisits included surgery in younger patients. Despite extremely high age 20–29 (odds ratio [OR]: 3.13; 95% confidence inter- rates of surgical success, broader criteria and large surgical vals [CI]: 2.33–4.20) and age 30–39 (OR: 2.35; CI: volume yield a significant number of complications, 1.91–2.89) compared with age 70–79, male sex (OR: simply as a proportion of total cases. 1.29; CI: 1.24–1.34), races black (OR: 1.37; CI: Several international population studies have examined 1.27–1.48) and Hispanic (OR: 1.16; CI: 1.08–1.24) aggregate incidence and risk factors for surgery-related 8–11 compared with white, and Medicaid insurance (OR: complications —including national efforts in the 1.18, CI: 1.07–1.30) compared with Medicare. Diabetes United Kingdom, Australia, Sweden, and Malaysia, was also associated with increased 30-day procedure- among others, as well as newer initiatives to collect related revisits (OR: 1.093, CI: 1.024–1.168). internationally standardized outcomes data (International CONCLUSIONS: Cataract surgery is a common and, in Consortium for Health Outcomes Measurement, aggregate, expensive procedure. -related American Academy of Intelligent revisits follow a similar trend as surgical complications Research In Sight registry, and the European Registry of in large-scale population data, and may be useful as a Quality Outcomes for Cataract and Refractive Surgery). preliminary, screening outcome measure. Our results There is more limited evidence in the United States highlight the importance of age as a risk factor for cataract regarding complication rates, risk factors, and their relative impact. A retrospective cross-sectional analysis of Medicare claims data examined adjusted 1-year rates of Supplemental Material available at AJO.com. severe postoperative complications (, Accepted for publication Aug 26, 2016. suprachoroidal hemorrhage, and ) and From the Veterans Affairs Palo Alto Health Care System, Byers Institute at Stanford University, Palo Alto, California (S.P.); found declining rates of serious adverse events Department of Surgery, Stanford University, Palo Alto, California between 1994 and 2006, with a 0.5% rate of at least 1 (D.E.M.); and Stanford School of Medicine, Stanford, California severe complication.12 (T.H.-B.). Douglas E. Morrison is currently affiliated with the Department of National quality analysis and reporting is increasingly Biostatistics, UCLA, Los Angeles, California. relying on standardized metrics across specialties—particu- Inquiries to Tina Hernandez-Boussard, Associate Professor of Surgery, larly post-hospitalization readmissions, or post-procedure Medicine (Biomedical Informatics), and Biomedical Data Science, Stanford School of Medicine, 1070 Arastradero, #225, Stanford, CA readmissions for surgical disciplines. Although cataract 94305-5733; e-mail: [email protected] patients (and ophthalmology patients more generally) are

130 © 2016 ELSEVIER INC.ALL RIGHTS RESERVED. 0002-9394/$36.00 http://dx.doi.org/10.1016/j.ajo.2016.08.036 seldom admitted or readmitted to the hospital, readmis- DATA SOURCES: The study examined discharge data sions—broadly defined as revisits, including return to the from the State Inpatient Databases, State Ambulatory operating room or unscheduled visits in an ambulatory Databases, and State Emergency Department Databases care setting—are a robust choice for analyzing significant of the Healthcare Cost and Utilization Project, Agency complications in cataract surgery. Detected events may for Healthcare Research and Quality, between 2008 and encompass retained cataract fragments, wound 2011 in California and Florida and between 2009 and leakage/dehiscence, severe endophthalmitis, and retinal 2011 in New York.21 These states were selected owing to detachment, among others. With the exception of a 1999 their geographic spread and the quality of available data, Canadian study that looked broadly at readmissions including all-capture discharges and unique identifiers to following outpatient (including cataract) surgeries at a track patients across hospital and outpatient or emergency single high-volume ambulatory surgery center with a settings. Available data included patient sociodemo- 0.2% rate of complication-related readmissions for cataract graphics, primary expected payer, and clinical variables surgery after 30 days,13 little has been published regarding related to diagnoses, comorbidities, and procedures readmissions as a marker for cataract surgery quality.4,14 performed. Although they do not capture all complications, read- Adult patients receiving cataract surgery in the missions do identify the most serious that have an impact ambulatory setting were identified by records contain- on cost as well as on quality of life. Readmissions are also ing one of the ICD-9 diagnosis codes 366.* (excluding less subject to the accuracy concerns implicit in using 366.52 and 366.53) and also containing CPT codes claims data to identify specific complications by Interna- 66984 or 66982 or one of the ICD-9 procedure codes tional Classification of Diseases, Ninth Revision (ICD-9) 13.* (used in ambulatory records in Florida but not in diagnosis and Current Procedural Terminology (CPT) pro- California or New York); 19% of the records had mul- cedure codes. And, with increased focus on value-based tiple cataract procedure codes. We excluded patients payment systems, including Affordable Care Act mandates under 20 and over 100 years of age, owing to limited to tie payments to quality outcomes,15–20 understanding number of patients and patients lacking the encounter how cataract surgery will be viewed by policymakers in a linkage identifier. broader context is increasingly important—part of identifying, tracking, and predicting complications as MAIN OUTCOME MEASURES: Procedure-related revisits. Our well as their cost and outcomes. primary outcome was 7-day and 30-day procedure-related In this study, we sought to evaluate the incidence of readmissions. We defined a procedure-related readmission as complications and revisits requiring a return for additional an emergency room visit, return to , or evaluation or treatment (in the clinic, operating room, or inpatient admission with complication-related ICD-9-CM emergency department setting) as a reliable quality indica- codes (Supplemental Table 1; Supplemental Material tor for cataract surgery in a large, multipayer, population- available at AJO.com). Planned postoperative visits were based observational study. We characterized patterns of identified using the primary diagnosis code (V5*, V67*, readmissions following cataract surgery, identified key V68*) and were not considered as readmissions. Records for patient characteristics (such as age and comorbidities) second cataract surgery procedures were also not considered associated with higher readmission rates, and analyzed as readmissions, presuming that they represented cataract data at the state level to identify important geographic surgery on the opposite eye. trends. All-cause readmissions. As a secondary outcome we analyzed all-cause readmissions, again excluding planned postoperative visits and cataract surgery on the opposite eye. METHODS PATIENT AND ORGANIZATIONAL CHARACTERISTICS: STUDY DESIGN: This study was a retrospective cohort Our descriptive statistics were based on patient character- study using all-capture state administrative datasets. Cata- istics obtained during the index visit (first cataract proced- ract procedures from California, Florida, and New York ure). Evaluated patient and health system variables state ambulatory surgery settings were identified using included age, sex, race/ethnicity, primary expected payer, International Classification of Diseases, Ninth Revision, number of chronic conditions (available in the ambulatory Clinical Modification (ICD-9-CM) and CPT codes. care records starting in 2009), and coded clinical Thirty-day readmissions (revisits) were identified in inpa- comorbidities (diabetes, myopic or hyperopic refractive tient, ambulatory, and emergency department settings error, vascular disease, hypertension, and prostate across each state. The Stanford Institutional Review Board disease—selected as a proxy for alpha blocker medication (IRB) determined that this study was exempt from IRB use, which is associated with intraoperative complexity approval in adherence to all state and federal laws. and complications).22 Clinical comorbidities were

VOL. 171 CATARACT SURGERY COMPLICATIONS AND REVISIT RATES 131 TABLE 1. Cataract Surgery Complications and Revisit Rates in Three States: Clinical Characteristics by Procedure-Related 30-Day Revisits, 2008–2011

Total No Revisit Revisit Variable N ¼ 1 223 733 N ¼ 1 211 805 (99.0%) N ¼ 11 928 (1.0%) P Value

Hospital state, n (%) California 413 682 (33.8) 409 823 (99.1) 3859 (0.9) Ref Florida 531 176 (43.4) 525 202 (98.9) 5974 (1.1) <.0001 New York 278 875 (22.8) 276 780 (99.2) 2095 (0.8) <.0001 Procedure code, n (%) CPT:66984 907 654 (74.2) 899 776 (99.1) 7878 (0.9) Ref CPT:66982 76 685 (6.3) 75 404 (98.3) 1281 (1.7) <.0001 PR:13* 239 394 (19.6) 236 625 (98.8) 2769 (1.2) <.0001 Age, n (%) 70–79 481 014 (39.3) 476 850 (99.1) 4164 (0.9) Ref 20–29 1751 (0.1) 1700 (97.1) 51 (2.9) <.0001 30–39 4474 (0.4) 4371 (97.7) 103 (2.3) <.0001 40–49 25 770 (2.1) 25 399 (98.6) 371 (1.4) <.0001 50–59 104 780 (8.6) 103 594 (98.9) 1186 (1.1) <.0001 60–69 323 734 (26.5) 320 603 (99.0) 3131 (1.0) <.0001 80–89 265 910 (21.7) 263 248 (99.0) 2662 (1.0) <.0001 90–100 16 300 (1.3) 16 040 (98.4) 260 (1.6) <.0001 Sex, n (%) Female 704 300 (58.1) 698 209 (99.1) 6091 (0.9) Ref Male 507 495 (41.9) 501 754 (98.9) 5741 (1.1) <.0001 Race, n (%) White 772 934 (63.2) 765 247 (99.0) 7687 (1.0) Ref Black 63 011 (5.1) 62 143 (98.6) 868 (1.4) <.0001 Hispanic 123 455 (10.1) 122 088 (98.9) 1367 (1.1) .0002 Asian/Pacific Islander 53 108 (4.3) 52 625 (99.1) 483 (0.9) .0554 Other or missing 211 225 (17.3) 209 702 (99.3) 1523 (0.7) <.0001 Primary expected payer, n (%) Medicare 779 524 (63.7) 771 876 (99.0) 7648 (1.0) Ref Medicaid 41 900 (3.4) 41 360 (98.7) 540 (1.3) <.0001 Private 325 028 (26.6) 321 901 (99.0) 3127 (1.0) .3535 Self pay 51 408 (4.2) 51 097 (99.4) 311 (0.6) <.0001 Other or missing 25 873 (2.1) 25 571 (98.8) 302 (1.2) .0029 Discharge year, n (%) 2008 305 650 (25.0) 302 494 (99.0) 3156 (1.0) Ref 2009 373 578 (30.5) 369 971 (99.0) 3607 (1.0) .0057 2010 294 850 (24.1) 292 032 (99.0) 2818 (1.0) .00272 2011 249 655 (20.4) 247 308 (99.1) 2347 (0.9) .0005 5þ chronic conditions, n (%) No 866 251 (94.4) 858 090 (99.1) 8161 (0.9) Ref Yes 51 832 (5.6) 51 221 (98.8) 611 (1.2) <.0001 Vascular disease, n (%) No 1 164 910 (95.2) 1 153 618 (99.0) 11 292 (1.0) Ref Yes 58 823 (4.8) 58 187 (98.9) 636 (1.1) .5993 Hypertension, n (%) No 943 714 (77.1) 934 584 (99.0) 9130 (1.0) Ref Yes 280 019 (22.9) 277 221 (99.0) 2798 (1.0) .1330 Diabetes, n (%) No 1 101 523 (90.0) 1 090 954 (99.0) 10 569 (1.0) Ref Yes 122 210 (10.0) 120 851 (98.9) 1359 (1.1) <.0001 Hyperopia, n (%) No 1 222 687 (99.9) 1 210 767 (99.0) 11 920 (1.0) Ref Yes 1046 (0.1) 1038 (99.2) <20 (<1%) .4904

Continued on next page

132 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER 2016 TABLE 1. Cataract Surgery Complications and Revisit Rates in Three States: Clinical Characteristics by Procedure-Related 30-Day Revisits, 2008–2011 (Continued)

Total No Revisit Revisit Variable N ¼ 1 223 733 N ¼ 1 211 805 (99.0%) N ¼ 11 928 (1.0%) P Value

Myopia, n (%) No 1 221 510 (99.8) 1 209 609 (99.0) 11 901 (1.0) Ref Yes 2223 (0.2) 2196 (98.8) 27 (1.2) .2503 Prostate , n (%) No 1 210 787 (98.9) 1 199 012 (99.0) 11 775 (1.0) Ref Yes 12 946 (1.1) 12 793 (98.8) 153 (1.2) .0161

identified by ICD-9 diagnosis code: diabetes (250.*), in Florida were coded under ICD-9 procedure code 13.*. hypertension (401.*–405.*), hyperopia (367.0), myopia Demographic and clinical characteristics of the study (367.1), prostate hyperplasia (600.*), and vascular disease sample are presented in Table 1. Across all states, the (central and peripheral arterial atherosclerosis, thrombosis, majority of patients in the sample fell between 60 and 80 or embolism, ischemic heart disease, or cerebrovascular dis- years of age. Mean age was 71.8. Most patients were rela- ease—410–414.*, 431.*, 433–435.*, 437.0–437.1, tively healthy, with fewer than 5 identified chronic medical 438.*440.*, 443.9, 444–445.*). conditions. Medicare was the primary expected payer in most cases (63.7%), followed by private insurance (26.6%). STATISTICAL ANALYSES: We calculated descriptive We also evaluated clinical comorbidities that were felt statistics for our sample of patients and analyzed to influence general health status or anatomic surgical procedure-related readmission and all-cause readmission outcomes. The prevalence of diabetes mellitus in our aggre- rates by patient characteristics, using 2-sided Wald tests gate sample was 10.0% and the prevalence of hyperopic of univariate logistic regression coefficient estimates. We and myopic refractive errors was 0.1% and 0.2%, respec- used the Pearson-Clopper method to construct 95% confi- tively (compared with prevalence of <0.1% and 0.1% in dence intervals for procedure-related readmission rates by our dataset as a whole—substantially lower than the decade of age. We compared clinical comorbidity and read- expected population prevalence, a findings that likely mission rates by state using the Pearson x2 test. We built reflects coding behavior rather than a true difference in multivariate models of readmission using mixed-effects patient characteristics). Hypertension appeared in 22.9% logistic regression, with a random effect by hospital. Using of patients in our sample and vascular disease in 0.7% of Census Bureau population data for 2010 as denominators,23 patients. Prostate disease was less than 2%. Estimated prev- we computed the total number of cataract procedures per alence of key clinical comorbidities was compared across 1000 persons according to patient age and the state in states (Supplemental Table 2; Supplemental Material which care occurred. available at AJO.com). All analyses were conducted using SAS version 9.3 (SAS Institute Inc, Cary, North Carolina, USA) or READMISSION RATES: We found low revisit rates at R version 3.0.3 (R Foundation for Statistical Computing, 7 days and 30 days (1.6% and 6.0% all-cause, 0.4% and Vienna, Austria). Because no direct patient-identifiable 1.0% procedure-related, respectively). Results were data were used, this study was exempt from review by the compared across states, with the highest procedure- Stanford University Institutional Review Board. related readmission rates seen in Florida (Table 2). The leading primary diagnoses recorded in all-cause revisit records were retained cataract lens fragments (6.6% of all-cause revisits at 7 days and 3.6% at 30 days), intraocular RESULTS lens implant malfunction (3.0% at 7 days and 2.1% at 30 days), and posterior capsule opacification/after- STUDY SAMPLE CHARACTERISTICS: We found discharge cataract obscuring vision (2.9% at 30 days). We found records with codes for cataract surgery for a total of 1 223 similar results for procedure-related revisits (Table 3). 733 patients between 20 and 100 years of age across the 3 Other non–procedure-related reasons for readmission states between 2008 and 2011—22.8% of these in New included syncope, hypertension, urinary tract infection, York (2009–2011), 33.8% in California, and 43.4% in pneumonia, and chest pain/cardiovascular disease. Other Florida. The majority of cases were identified with the reasons for readmission each accounted for less than 2% CPT code for routine extracapsular cataract surgery of readmissions and were likely unrelated to surgery (66984); however, a large proportion of surgeries performed (eg, benign colon neoplasm). Most procedure-related

VOL. 171 CATARACT SURGERY COMPLICATIONS AND REVISIT RATES 133 TABLE 2. Cataract Surgery Complications and Revisit Rates in Three States: All-Cause and Procedure-Related Revisit Rates, 2008–2011

All States California Florida New York N ¼ 1 223 733 N ¼ 413 682 N ¼ 531 176 N ¼ 278 875

All-cause revisits, n (%) 7-day 19 246 (1.6) 6360 (1.6) 9492 (1.8) 3394 (1.2) 30-day 73 239 (6.0) 23 778 (5.8) 36 207 (6.8) 13 254 (4.8) Procedure-related revisits, n (%) 7-day 4179 (0.4) 1498 (0.4) 1918 (0.4) 763 (0.3) 30-day 11 928 (1.0) 3859 (0.9) 5974 (1.1) 2095 (0.8)

All calculations were statistically significant with P < .0001 for a difference among states.

readmissions were to ambulatory care (Supplemental Table 3; Supplemental Material available at AJO.com). TABLE 3. Cataract Surgery Complications and Revisit Rates in Three States: Top 10 Primary Diagnoses for 30-Day EFFECT OF PATIENT AGE: Analysis of procedure-related Procedure-Related Revisits readmission rates according to patient age revealed a bimodal N (%) distribution, with worse outcomes among younger and older ICD-9 Code and Name (Total ¼ 11 928) ;patients. Figure 1 illustrates aggregate 30-day procedure- related readmission rates according to age—ranging from 998.82: Cataract fragments after surgery 2.9% (20–29 years of age) to 0.87% (70–79 years of age). No 9250 (77.5) Yes 2678 (22.5) The lowest rate of procedure-related readmissions was 366.53: After-cataract obscuring vision observed between 70 and 79 years of age, having a 70% No 9744 (81.7) reduction in odds of procedure-related readmission Yes 2184 (18.3) compared with a patient in his or her 20s. 996.53: Lens prosthesis malfunction No 10 326 (86.6) Yes 1602 (13.4) OTHER CHARACTERISTICS ASSOCIATED WITH PROCEDURE-RELATED READMISSIONS: Although differ- 379.32: Subluxation of lens ences were small, a higher relative percentage of patients No 11 601 (97.3) Yes 327 (2.7) with procedure-related readmissions was seen among 379.34: Posterior dislocation of lens male patients, those of black or Hispanic race, those with No 11 652 (97.7) Medicaid or unknown insurance, those with presence of 5 Yes 276 (2.3) or more chronic conditions, and those with diabetes, 367.31: Anisometropia myopia, and prostate hyperplasia (Table 1). No 11 724 (98.3) In multivariate analyses, procedure-related readmissions Yes 204 (1.7) varied by patient sex, race/ethnicity, and diabetic status, as 361.00: Retinal detachment with defect, NOS well as by insurance/payer type (Table 4). Male sex and No 11 747 (98.5) black or Hispanic race/ethnicity were consistently associ- Yes 181 (1.5) ated with more readmissions (an increase in odds ranging 360.00: Purulent endophthalmitis, NOS from 16% to 37%). Among the clinical comorbidities in No 11 755 (98.5) Yes 173 (1.5) the model, only diabetes was statistically significantly asso- 379.31: ciated with procedure-related readmission (odds ratio No 11 772 (98.7) [OR]: 1.09; 95% confidence interval [CI]: 1.02–1.17). Yes 156 (1.3) Referenced to Medicare, Medicaid was significantly asso- 361.9: Retinal detachment, NOS ciated with higher procedure-related readmission rates at No 11 774 (98.7) 30 days (OR: 1.18; CI: 1.07–1.30). Private insurance was Yes 154 (1.3) associated with lower odds of procedure-related read- ¼ missions (OR: 0.94; CI: 0.90–0.99). Notably, non-Medicare ICD-9 International Classification of Diseases, Ninth ¼ payers had a higher proportion of younger (50–70 year old) Revision; NOS not otherwise specified.

134 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER 2016 FIGURE 1. Patient age and 30-day procedure-related revisit rates, 2008–2011. Shown are 95% confidence intervals; numbers above each confidence interval represent odds ratios for procedure-related revisits (adjusted for sex, hospital state, race, primary expected payer, and clinical comorbidities), with the 70–79 years age group held as reference.

patients and fewer chronic medical conditions. Even performed annually at a cost of $3.4 billion to Centers for Medicaid had fewer chronic medical conditions than Medi- Medicare and Medicaid Services. Although cataract sur- care—5.7% of patients having over 5 chronic conditions, gery is appropriately considered safe and effective, compli- compared to 7.0% of Medicare patients. However, Medicaid cations have far-reaching impact—financial as well as had the highest percentage of patients with diabetes, over functional. Direct clinical costs of managing complications twiceasmanypatientsasMedicare(22.3%vs10.4%). range from $500 to $6000 for an episode of care,24 and in- direct costs are even greater. Aside from decreased quality PRACTICE AND OUTCOME VARIATION BY STATE: of life, visual impairment also has societal costs, including All-cause and procedure-related readmission rates were lost potential wages (or unpaid societal contribution) for higher in Florida than in California and New York the patient and the implicit costs of caregiver time. Thus, (Table 2)—a difference that was statistically significant as cataract surgery becomes increasingly common owing (P < .0001). Incorporating 2010 U.S. state census data, to an aging population, improving technology, and lower we found higher rates of cataract surgery (per 100 000 pop- thresholds for performing surgery, it will play an important ulation) in Florida compared with California and New part in health system–wide efforts to improve value by York, across all ages but particularly at 50–84 years lowering costs and improving quality. (Figure 2). The majority of all cataract surgeries (70.5%) Recent national and international guidelines have are performed between 65 and 84 years of age. Within endorsed cataract quality assessment. The U.S. Agency this age group, in 2010 a higher proportion of younger for Healthcare Research and Quality National Quality residents 65–74 years old underwent surgery in Florida Measures Clearinghouse (NQMC) includes recommenda- than in California or New York (31.9% of patients vs tions to collect data on visual outcomes after cataract sur- 13.4% and 17.3%, respectively). Also, although most gery, readmissions within 28 days of cataract surgery, and procedure-related readmissions occurred in patients 70–84 presence of retained cataract lens fragments, endophthal- years of age regardless of state, the highest odds of mitis, dislocated or wrong prosthesis, all-cause and procedure-related readmissions were seen in retinal detachment, or wound dehiscence within 30 days 20- to 49-year-old patients in Florida. of cataract surgery (NQMC: 007860-007863 and 003800- 003802; HHS: 004298-004299, 004915-004916, and 004489-004490).25 Additionally, the International Con- DISCUSSION sortium for Health Outcomes Measurement selected cata- ract surgery as 1 of 4 initial key conditions to lead an CATARACT SURGERY IS THE LEADING PROCEDURE UNDER international initiative for collecting clinical outcomes Medicare, with approximately 3 million surgeries data.26 Recommended outcome metrics included presence

VOL. 171 CATARACT SURGERY COMPLICATIONS AND REVISIT RATES 135 readmission rates following cataract surgery were predict- TABLE 4. Cataract Surgery Complications and Revisit Rates ably low. Including non–procedure-related reasons for in Three States: Multivariate Associations Between Clinical readmission, we conservatively estimated 6.0% readmis- Characteristics and 30-Day Procedure-Related Revisits, sions at 30 days—well below rates for other noted medical 2008–2011 conditions.20 Narrowing results further to include only

Variable Odds Ratio LCI UCI P Value cataract surgery procedure-related readmissions yielded a 30-day rate of 1.0%, with retained cataract lens fragments Sex and intraocular lens malfunction as the leading primary Female Ref associated diagnosis codes. These rates were higher than Male 1.289 1.242 1.337 <.0001 Hospital state in a Canadian study including 4700 outpatient cataract California Ref surgeries; however, the calculated readmission rate of Florida 1.138 0.995 1.303 .0565 0.2% in the study included only procedure-specific compli- 13 New York 0.843 0.720 0.987 .0391 cations leading to surgical readmission. Race Factors associated with readmissions in our analysis White Ref included male sex, black or Hispanic race, diabetes, and Black 1.374 1.272 1.484 <.0001 Medicaid as primary expected payer—a constellation of Hispanic 1.156 1.080 1.237 <.0001 characteristics that may be seen among underserved Asian/Pacific Islander 1.021 0.919 1.133 .7030 patients. In addition to these, we found age to have an Other or missing 0.924 0.86 0.994 .0335 important association with procedure-related readmission, Primary expected payer with patients in their early 70s having the lowest readmis- Medicare Ref Medicaid 1.181 1.071 1.302 .0008 sion rates. Both young and older patients were at higher risk Private 0.944 0.897 0.993 .0243 compared with this age group—an association that makes Self pay 0.888 0.765 1.032 .1215 intuitive clinical sense because younger patients are likely Other or missing 1.050 0.919 1.199 .4762 to have other eye pathology causing earlier cataract Age (eg, prior eye trauma or ) and also causing the 70–79 Ref surgery to be riskier and more complex. Alternative expla- 20–29 3.130 2.332 4.202 <.0001 nations could be that younger patients may have better 30–39 2.351 1.911 2.893 <.0001 access to care (eg, transportation and resources) or higher < 40–49 1.608 1.432 1.805 .0001 levels of anxiety than the elder, making the younger < 50–59 1.263 1.174 1.358 .0001 patient more likely to seek additional care and/or atten- 60–69 1.096 1.044 1.151 .0002 tion. Older patients have denser ; are more prone 80–89 1.197 1.139 1.258 <.0001 90–100 1.923 1.693 2.185 <.0001 to corneal scarring, chronically dry ocular surface, or other Clinical comorbidities opacity impairing the surgical view; and may have a Hyperopic refractive 0.674 0.334 1.361 .2712 that dilates poorly owing to systemic alpha blocker medica- 28 error tions, diabetes, or ischemia. Myopia 1.082 0.731 1.600 .6941 Important clinical comorbidities were inconsistently Prostate disease 0.99 0.839 1.169 .9058 recorded in our dataset. Although the prevalence of Vascular disease 1.051 0.963 1.147 .2615 diabetes mellitus in our aggregate sample was similar to Hypertension 0.959 0.905 1.017 .1623 expected U.S. population prevalence (8.3%–11.5%),29 Diabetes 1.093 1.024 1.168 .0080 the prevalence of other key clinical comorbidities was LCI ¼ lower 95% confidence interval; UCI ¼ upper 95% significantly lower than the expected population preva- confidence interval. lence, particularly for hyperopic and myopic refractive errors (<0.1% and 0.1%, respectively, in our dataset, compared with an expected prevalence of 9.95% and 23.9% per National Eye Institute data).30,31 Similarly, of dropped nucleus (retained cataract lens fragments), hypertension, vascular disease, and prostate disease endophthalmitis, capsule problems, persistent corneal appeared less prevalent in our sample than expected for edema, and complication-related return to operating the population.32 Because these findings were not specific room within 90 days. Multiple studies have investigated to our sample, but rather were seen in the dataset as a visual outcomes and center-specific complication rates whole, we believe they likely reflect coding practices. with cataract surgery under different techniques and with Hospital readmissions in general have demonstrated differing experience.27 substantial variation across hospitals and regions, and we We used large representative population datasets for similarly identified differences in cataract surgery practice California, New York, and Florida to assess incidence of and outcomes among states. Though our study only procedure-related revisits and to evaluate those revisits as included analysis of 3 states, there was a suggestion of a comprehensive measure of quality. In our analysis, different rates of surgery by state, particularly among

136 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER 2016 FIGURE 2. Population rates of cataract surgery by hospital state and age in 2010. Numbers based on recorded surgical cases and 2010 U.S. census population data for each state.

younger patients, with important associated differences in readmission for cataract procedures, as the data do not rates of procedure-related readmissions. Although not capture patients who seek additional treatment in a purely accounting for potential confounders such as differences outpatient setting or in another state. Furthermore, gener- in coding practices, number of ophthalmologists, number alizability of our results may vary by state, depending on of residency programs, or number of residents, these find- system resources, predominance of rural vs urban settings, ings suggest the importance of judicious patient selection et cetera. This study highlights the need for more detailed and counseling. As we seek to standardize best practices, evidence, paving the way for interventions to improve an important area for future research will be to investigate outcomes and optimize patient selection and counseling. the extent and etiology of variation in cataract surgery practice and its effect on visual outcomes.

LIMITATIONS: Our analysis was inherently limited by CONCLUSIONS the weaknesses of retrospective claims-based data, particu- 33,34 larly coding inaccuracies. For example, the low CATARACT SURGERY IS ALREADY ONE OF THE MOST COM- apparent prevalence of refractive error in our sample may mon and, in aggregate, most expensive procedures indicate cataract surgery performed less frequently in performed nationally. Although complication rates are patients with myopia or hyperopia, but more likely low, understanding and preventing complications and reflects coding practices (particularly given consistent complication-related readmissions is an important chal- differences in coded comorbidities across states, as shown lenge, particularly as the number of surgeries performed in Supplemental Table 2). However, equivalent large- continues to grow. Procedure-related readmissions follow volume data would be impossible to generate via a random- a similar trend as surgical complications in large-scale ized controlled clinical trial, and the empiric nature of the population data, and may be useful as a preliminary, or data itself enhances generalizability. We believe that this screening, outcome measure, particularly to identify underscores the need for better data collection, such as regional variation in care. Our results additionally intended under ICD-10 and through clinical data registries highlight the importance of age as a risk factor for cataract including the new national ophthalmology Intelligent surgery readmissions, and suggest a relationship Research in Sight clinical registry, as well as the need for between black or Hispanic race, Medicaid insurance, and a reliable comprehensive measure of quality—such as read- diabetes associated with higher risk for cataract surgery mission rates. Our study may also be understating the risk of complications.

VOL. 171 CATARACT SURGERY COMPLICATIONS AND REVISIT RATES 137 FUNDING/SUPPORT: A PORTION OF THIS RESEARCH WAS FUNDED BY STANFORD UNIVERSITY, SPECTRUM INNOVATION Award in Populations Sciences, Stanford, California. Financial disclosures: The following authors have no financial disclosures: Suzann Pershing, Douglas E. Morrison, and Tina Hernandez-Boussard. All authors attest that they meet the current ICMJE criteria for authorship.

REFERENCES 18. McCarthy D, Johnson MB, Audet AM. Recasting readmis- sions by placing the hospital role in community context. 1. Gollogly HE, Hodge DO, St Sauver JL, Erie JC. Increasing JAMA 2013;309(4):351–352. incidence of cataract surgery: population-based study. J Cata- 19. Berry JG, Hall DE, Kuo DZ, et al. Hospital utilization and char- ract Refract Surg 2013;39(9):1383–1389. acteristics of patients experiencing recurrent readmissions 2. Klein BE, Howard KP, Lee KE, Klein R. Changing incidence within children’s hospitals. JAMA 2011;305(7):682–690. of lens extraction over 20 years: the Beaver Dam eye study. 20. Dharmarajan K, Hsieh AF, Lin Z, et al. Diagnoses and timing Ophthalmology 2014;121(1):5–9. of 30-day readmissions after hospitalization for heart failure, 3. Javitt JC, Kendix M, Tielsch JM, et al. Geographic variation in acute myocardial infarction, or pneumonia. JAMA 2013; utilization of cataract surgery. Med Care 1995;33(1):90–105. 309(4):355–363. 4. Schein OD, Cassard SD, Tielsch JM, Gower EW. Cataract 21. Healthcare Cost and Utilization Program. Healthcare Cost surgery among Medicare beneficiaries. Ophthalmic Epidemiol and Utilization Project (HCUP) 1993-2011. Available at: 2012;19(5):257–264. http://hcupnet.ahrq.gov/. Accessed October 15, 2015. 5. Erie JC. Rising cataract surgery rates: demand and supply. 22. Chang DF, Braga-Mele R, Mamalis N, et al. ASCRS White Ophthalmology 2014;121(1):2–4. Paper: clinical review of intraoperative floppy- syndrome. 6. Hatch WV, Campbell Ede L, Bell CM, El-Defrawy SR, J Cataract Refract Surg 2008;34(12):2153–2162. Campbell RJ. Projecting the growth of cataract surgery during 23. U.S. Census Bureau. Census 2010. United States Census the next 25 years. Arch Ophthalmol 2012;130(11):1479–1481. Bureau; 2010. 7. Lundstrom M, Goh PP, Henry Y, et al. The changing pattern 24. Busbee BG, Brown MM, Brown GC, Sharma S. Incremental of cataract surgery indications: a 5-year study of 2 cataract sur- cost-effectiveness of initial cataract surgery. Ophthalmology gery databases. Ophthalmology 2015;122(1):31–38. 2002;109(3):606–612. discussion 612–603. 8. Goh PP, Salowi MA. The 6th report of the National Eye 25. Agency for Healthcare Research and Quality. National qual- Database 2012. Available at: http://www.acrm.org.my/ned/ ity measures clearinghouse 2014. Available at: http://www. cataractSurgeryRegistry.html. Accessed January 10, 2015. qualitymeasures.ahrq.gov/. Accessed October 22, 2014. 9. Clark A, Morlet N, Ng JQ, Preen DB, Semmens JB. Whole 26. International Consortium for Health Outcomes Measure- population trends in complications of cataract surgery over ment. Medical conditions 2014. Available at: http://www. 22 years in Western Australia. Ophthalmology 2011;118(6): ichom.org/medical-conditions/. Accessed October 22, 2014. 1055–1061. 27. de Silva SR, Riaz Y, Evans JR. Phacoemulsification with 10. Johnston RL, Taylor H, Smith R, Sparrow JM. The Cataract posterior chamber intraocular lens versus extracapsular cata- National Dataset electronic multi-centre audit of 55,567 op- ract extraction (ECCE) with posterior chamber intraocular erations: variation in posterior capsule rupture rates between lens for age-related cataract. Cochrane Database Syst Rev . Eye 2010;24(5):888–893. 2014;CD008812. 11. Behndig A, Montan P, Stenevi U, Kugelberg M, 28. Preferred practice pattern guideline: cataract in the adult eye Lundstrom M. One million cataract surgeries: Swedish Na- 2011 Available at: http://one.aao.org/preferred-practice- tional Cataract Register 1992-2009. J Cataract Refract Surg pattern/cataract-in-adult-eye-ppp–october-2011. Accessed 2011;37(8):1539–1545. October 22, 2014. 12. Stein JD, Grossman DS, Mundy KM, Sugar A, Sloan FA. Se- 29. Center for Disease Control and Prevention. National dia- vere adverse events after cataract surgery among medicare betes fact sheet: national estimates and general information beneficiaries. Ophthalmology 2011;118(9):1716–1723. on diabetes and prediabetes in the United States. In: Human 13. Mezei G, Chung F. Return hospital visits and hospital readmis- and Health Services, ed; 2011. Atlanta, GA. sions after ambulatory surgery. Ann Surg 1999;230(5):721–727. 30. National Eye Institute. Hyperopia 2014. Available at: http:// 14. Menda SA, Driver TH, Neiman AE, Naseri A, Stewart JM. www.nei.nih.gov/eyedata/hyperopia.asp. Accessed October Return to the operating room after resident-performed cata- 22, 2014. ract surgery. JAMA Ophthalmol 2014;132(2):223–224. 31. National Eye Institute. Myopia 2014. Available at: http:// 15. Kocher RP, Adashi EY. Hospital readmissions and the www.nei.nih.gov/eyedata/myopia.asp. Accessed October 22, Affordable Care Act: paying for coordinated quality care. 2014. JAMA 2011;306(16):1794–1795. 32. Paolone DR. Benign prostatic hyperplasia. Clin Geriatr Med 16. Birkmeyer JD, Hamby LS, Birkmeyer CM, Decker MV, 2010;26(2):223–239. Karon NM, Dow RW. Is unplanned return to the operating 33. Romano PS, Mark DH. Bias in the coding of hospital room a useful quality indicator in general surgery? Arch Surg discharge data and its implications for quality assessment. 2001;136(4):405–411. Med Care 1994;32(1):81–90. 17. Tsai TC, Joynt KE, Orav EJ, Gawande AA, Jha AK. Varia- 34. Sacks GD, Dawes AJ, Russell MM, et al. Evaluation of hospi- tion in surgical-readmission rates and quality of hospital tal readmissions in surgical patients: do administrative data care. N Engl J Med 2013;369(12):1134–1142. tell the real story? JAMA Surg 2014;149(8):759–764.

138 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER 2016