Highlights
- •
Nearly 1 in 5 patients who undergo primary rhegmatogenous retinal detachment repair require reoperation within 90 days.
- •
Reoperation occurs most frequently after pneumatic retinopexy (29% of cases).
- •
Reoperation rates are similar after scleral buckle (19%) and pars plana vitrectomy (18%).
- •
Patient characteristics also influence likelihood of reoperation.
Purpose
To examine associations between primary repair, patient characteristics, and rhegmatogenous retinal detachment (RRD) reoperation.
Design
Retrospective cohort study.
Methods
We used administrative claims to identify enrollees with incident RRD treatment by laser barricade, pneumatic retinopexy (PR), pars plana vitrectomy (PPV), or scleral buckle (SB) between 2003 and 2016. Analysis excluded patients with less than 3 years of continuous enrollment, previous RRD diagnosis, or repair. We determined reoperation frequency (PPV, PR, or SB) within 90 days postrepair and used multivariable logistic regression to identify associations between reoperation and patient and primary repair characteristics.
Results
Of 16,190 patients with documented primary RRD repair, 2,918 (18.0%) required reoperation within 90 days. Reoperation was significantly associated with male sex (odds ratio [OR] 1.24, P < .001), pseudophakia (OR 1.25, P < .001), vitreous hemorrhage (OR 1.22, P = .001), and worse systemic health (OR 1.19-1.25, P < .05, for Charlson Comorbidity Index ≥3). Pseudophakia had higher reoperation odds after all primary procedures except PPV. In addition, 28.7% of primary PR cases required reoperation, vs 19.1% of SB and 17.9% of PPV repairs. Adjusting for other patient characteristics, PR had highest odds of reoperation (OR 1.90, P < .001, vs primary PPV). Primary laser barricade had lowest odds of reoperation (OR 0.49, P < .001). PPV was the most frequent reoperation procedure.
Conclusions
Nearly 1 in 5 patients require reoperation within 90 days after primary RRD repair. Cases requiring only primary laser barricade had lowest reoperation odds, likely representing less severe RRDs. Primary PR had highest reoperation odds; PPV and SB were similar to each other. These findings are important for patient education and surgical decision-making.
R hegmatogenous retinal detachments (RRDs) have an annual incidence of approximately 1 in 10,000 and are treated with a variety of methods. Used alone or in combination, management approaches include pneumatic retinopexy (PR), pars plana vitrectomy (PPV), scleral buckle (SB), laser retinopexy, and cryopexy.
Scleral buckling has shown robust historic success rates ranging from 75% to 95%. However, long-term outcomes show that up to 13% of patients undergoing SB still require 1 or more additional vitreoretinal procedures over a follow-up period of 20 years. Recent years have seen advances in microsurgical techniques with improved outcomes for difficult cases. PPV has been increasingly adopted for RRD treatment, and is recommended for pseudophakic patients with primary retinal detachment, for example. , Single operation success rates (SOSRs) for PPV historically fall between 71%-98%, with some comparative trials suggesting there is no significant statistical difference in SOSR for scleral buckle vs pars plana vitrectomy; however, this is controversial. , A systematic review examining retinal detachment after primary repair revealed redetachment rates of 21% in the PPV population, less than the 28% seen in the SB repair group, over follow-up ranging from 3-36 months. PPV may be superior to SB for successful reattachment of complicated cases such as those with proliferative vitreoretinopathy. , , Pseudophakic patients had better SOSR with PPV than SB (20% vs 40% redetachment) at 12 months. Findings are overall mixed.
Unlike PPV and SB, pneumatic retinopexy offers ease of treatment performed in clinic rather than the operating room. Possible reasons for reoperation include missed or induced retinal breaks, vitreous traction on breaks inhibiting closure of the break, persistent retinal detachment, and later proliferative vitreoretinopathy. Patient factors such as pseudophakia may also affect surgery results. , Among phakic patients undergoing RRD repair via PR, 71%-84% required only a single surgery for successful reattachment, but this was notably lower for pseudophakic patients (SOSR of 41%-67%). A recent prospective, randomized trial comparing PR vs PPV for primary management of uncomplicated RRD found success rates of 80.8% in the PR group and 93.2% in the PPV group.
Need for reoperation is influenced both by primary procedure type and underlying patient comorbidities. However, although there are recommendations for primary RRD repair technique based on factors such as detachment severity, location of the break(s), and patient characteristics, head-to-head real-world data on likelihood of needing repeat RRD repair is limited. In this investigation, we used national managed care administrative claims among patients receiving RRD repair between 2003 and 2016 to assess the association of primary repair procedure and patient factors—including demographics, ocular and systemic comorbidities, and geographic region—with need for reoperation after primary repair.
Methods
Data Source
Our retrospective cohort study analyzed administrative claims filed between 2003 to 2016, inclusive of both Medicare Advantage (Part C) and commercial health insurance enrollees. Claims were drawn from the Clinformatics DataMart database (OptumInsight, Eden Prairie, Minnesota, USA), which represents approximately 67.6 million distinct patients and provides dates of member enrollment, demographics, and applicable diagnosis as well as procedure codes. Based on range of study years, both International Classification of Diseases, 9th ( ICD-9 ) and 10th ( ICD-10 ), diagnosis codes are used, in addition to Current Procedural Terminology ( CPT ) codes. This analysis was deemed exempt by the Stanford University Institutional Review Board.
Sample and Variable Selection
Only patients with 3 or more years of continuous health plan enrollment were considered in this analysis, to avoid missing potential unobserved health care during unenrolled periods. For those patients with many periods of enrollment satisfying this requirement, the longest period was used for analysis. We identified incident RRD based on the first recorded diagnosis after a 2-year lookback period, excluding patients with pre-existing RRD diagnosis or a previous RRD repair procedure code documented during the lookback period (Supplementary Table S1). We also excluded patients with confounding ocular comorbidities, including tractional or serous retinal detachment, proliferative retinopathy (eg, diabetic or sickle cell retinopathy), retinoschisis, chorioretinitis or endophthalmitis, penetrating ocular trauma, and retinal vascular occlusion (Supplementary Table S1).
We limited our analysis to patients who underwent primary RRD repair within 1 year after incident RRD diagnosis. We used lookback period data to determine baseline demographics and clinical characteristics, including age, race, sex, Charlson Comorbidity Index (as a measure of general systemic health), geographic region (US census division), and selected comorbidities (myopia, pseudophakia, vitreous hemorrhage, lattice degeneration, and diabetes mellitus) using ICD-9 and ICD-10 codes (Supplementary Table S1). We determined the type of primary repair procedure performed for each patient—PR, SB, PPV with or without SB, or laser barricade—using CPT procedure codes (Supplementary Table S2) and calculated time from initial diagnosis to primary repair procedure. Our outcome of interest was any reoperation by PR, SB, or PPV (defined by the same set of CPT codes for RRD repair) occurring within 90 days after the primary repair.
Statistical Analysis
Statistical analyses were conducted in SAS, version 9.4 (SAS Institute Inc, Cary, North Carolina, USA), Stata, version 14.2 (StataCorps, College Station, Texas, USA), and Excel, version 16.45 (Microsoft, Redmond, Washington, USA). Baseline patient characteristics were summarized with frequencies and percentages. Among the subset of patients who underwent primary RRD repair, we determined incidence of RRD reoperation and compared characteristics for patients who underwent secondary repair vs those who did not. We used multivariable logistic regression models to study associations of primary RRD repair type and patient characteristics with need for reoperation. Because the influence of patient characteristics on odds of reoperation may vary based on primary repair type, we also stratified our multivariable regression analysis based on primary repair procedure. Finally, we used multivariable regression modeling to evaluate odds of specific secondary repair procedures based on primary repair procedure and baseline characteristics. For all analyses, P <.05 was considered statistically significant.
Results
Study Sample Demographics
In total, 25,417 patients with incident RRD diagnoses met inclusion criteria between 2003 and 2016. On average, patients were continuously enrolled in their insurance plan for 7.9 consecutive years and were 59.8 years old at time of first RRD (range 3-89 years). Primary repair procedures were documented within 1 year after incident RRD diagnosis in 16,190 of these patients (63.7%), and most repairs completed within 1 week after initial diagnosis.
Among patients who underwent primary RRD repair, 2,918 (18.0%) required reoperation within 90 days after initial surgery ( Table 1 ). On average, time from initial RRD diagnosis to primary repair was shorter for patients requiring reoperation compared with those who did not (mean 4.40 days vs 5.75 days, P = .007). A higher proportion of patients requiring secondary repair were male (66.7%, vs 33.3% female). The largest proportion of reoperation patients were white (80.6%); however, black enrollees were slightly more likely to require reoperation ( P = .02). Age distribution was similar for RRD patients with and without reoperations. Among those patients requiring a reoperation, age at RRD diagnosis ranged from 5 to 88 years, with a mean of 60.0 years.
| Cases Requiring Reoperation, (n = 2,918) | Cases Without Reoperation, (n = 13,272) | ||
|---|---|---|---|
| Characteristic | n (%) | n (%) | P Value |
| Year | |||
| 2005 | 160 (5.48) | 734 (5.53) | .92 |
| 2006 | 181 (6.20) | 816 (6.15) | .91 |
| 2007 | 218 (7.47) | 913 (6.88) | .26 |
| 2008 | 245 (8.40) | 1,069 (8.05) | .54 |
| 2009 | 281 (9.63) | 1,070 (8.06) | .006 |
| 2010 | 235 (8.05) | 1,079 (8.13) | .89 |
| 2011 | 286 (9.80) | 1,144 (9.07) | .22 |
| 2012 | 280 (9.60) | 1,259 (9.97) | .54 |
| 2013 | 326 (11.17) | 1,291 (10.29) | .16 |
| 2014 | 284 (9.73) | 1,240 (9.86) | .84 |
| 2015 | 222 (7.61) | 1,169 (9.30) | .004 |
| 2016 | 200 (6.85) | 1,098 (8.71) | .001 |
| Sex | |||
| Male | 1,946 (66.69) | 8,024 (60.46) | <.001 |
| Female | 972 (33.31) | 5,248 (39.54) | <.001 |
| Age | |||
| 0-9 y | 4 (0.14) | 15 (0.11) | .73 |
| 10-19 y | 30 (1.03) | 154 (1.16) | .54 |
| 20-29 y | 31 (1.06) | 234 (1.76) | .007 |
| 30-39 y | 96 (3.29) | 441 (3.32) | .93 |
| 40-49 y | 334 (11.45) | 1,408 (10.61) | .19 |
| 50-59 y | 907 (31.08) | 3,949 (29.75) | .16 |
| 60-69 y | 851 (29.16) | 4,189 (31.56) | .011 |
| 70-79 y | 495 (16.96) | 2,162 (16.29) | .37 |
| 80-89 y | 170 (5.83) | 720 (5.42) | .39 |
| Race | |||
| Asian | 78 (2.67) | 424 (3.19) | .14 |
| Black | 215 (7.37) | 821 (6.19) | .02 |
| Hispanic | 155 (5.31) | 765 (5.76) | .34 |
| White | 2,352 (80.60) | 10,726 (80.82) | .79 |
| Unknown | 118 (4.04) | 536 (4.04) | .99 |
| Clinical comorbidities | |||
| Myopia | 96 (3.29) | 380 (2.86) | .22 |
| Pseudophakia | 598 (20.49) | 2,175 (16.39) | <.001 |
| Vitreous hemorrhage | 408 (13.98) | 1,741 (13.12) | .21 |
| Lattice degeneration | 359 (12.30) | 1,838 (13.85) | .03 |
| Diabetes | 546 (18.71) | 2,090 (15.75) | <.001 |
| Charlson Comorbidity Index | |||
| 0 | 1,589 (54.46) | 7,603 (57.29) | .005 |
| 1-2 (mild) | 871 (29.85) | 3,976 (29.96) | .91 |
| 3-4 (moderate) | 288 (9.87) | 1,084 (8.17) | .003 |
| 5+ (severe) | 170 (5.83) | 609 (4.59) | .005 |
| Mean days (95% CI) from primary diagnosis to primary RD repair | 4.40 (3.69-5.10) | 5.75 (5.32-6.19) | .007 |
Patients with pseudophakia ( P < .001) and diabetes ( P < .001) were disproportionately represented among patients who underwent secondary repair, whereas fewer patients with lattice degeneration required reoperation ( P = .03). There was no statistically significant difference in the prevalence of other ocular comorbidities such as myopia or vitreous hemorrhage by reoperation status. Patients requiring reoperation had more systemic comorbidities, based on Charlson Comorbidity Index scores in the moderate or severe categories ( P < .01).
Association of Primary Repair Method With Need for RRD Reoperation
Type of primary repair procedure was significantly associated with 90-day reoperation rate ( Table 2 ). The highest frequency of secondary procedures was observed among those patients who underwent primary pneumatic retinopexy (28.7% of cases). SB and PPV were similar to each other, with a slightly higher frequency among SB patients (19.1%, vs 17.9% following PPV).
| Primary Repair Procedure | Cases Requiring Reoperation, n (%)(n = 2,918) | Cases Without Reoperation, n (%)(n = 13,272) | P Value | % of Primary Repair Cases Undergoing Reoperation |
|---|---|---|---|---|
| Pneumatic retinopexy | 611 (20.94) | 1,520 (11.45) | <.001 | 28.67 |
| Scleral buckle | 371 (12.71) | 1,575 (11.87) | .20 | 19.06 |
| Vitrectomy | 1,687 (57.81) | 7,714 (58.12) | .76 | 17.94 |
| Laser | 249 (8.53) | 2,463 ( 18.56 ) | <.001 | 9.18 |
Even after adjusting for patient characteristics, patients who underwent primary PR were most likely to require reoperation—almost 2-fold higher odds of secondary repair after PR compared with primary PPV (odds ratio [OR] 1.90, 95% confidence interval [CI] 1.70-2.12, P < .001). The next highest odds of reoperation were seen after SB alone, with 13% higher odds of reoperation than post-PPV, a finding that just missed the threshold for statistical significance (OR 1.13, 95% CI 0.99-1.29, P = .06) ( Table 3 ). Patients with RRDs that were able to be adequately managed with laser barricade alone had significantly lower odds of needing reoperation compared with primary PPV (OR 0.49, 95% CI 0.42-0.56, P < .001). The influence of primary repair timing (time from initial diagnosis to primary repair) was statistically but not clinically significant in adjusted analysis (OR 0.9978, 95% CI 0.9958-0.9998, P = .03).
| Odds Ratio (95% CI) | P Value | |
|---|---|---|
| Age | ||
| 0-9 y | 2.02 (0.62-6.59) | .25 |
| 10-19 y | 1.41 (0.82-2.44) | .22 |
| 20-29 y | — | — |
| 30-39 y | 1.57 (1.01-2.44) | .04 |
| 40-49 y | 1.51 (1.01-2.24) | .04 |
| 50-59 y | 1.39 (0.94-2.06) | .10 |
| 60-69 y | 1.18 (0.80-1.75) | .40 |
| 70-79 y | 1.28 (0.86-1.91) | .23 |
| 80-89 y | 1.34 (0.88-2.06) | .17 |
| Sex (male) | 1.24 (1.14-1.35) | <.001 |
| Race | ||
| Asian | — | — |
| Black | 1.32 (0.99-1.77) | .06 |
| Hispanic | 1.03 (0.76-1.40) | .84 |
| Unknown | 1.14 (0.83-1.57) | .43 |
| White | 1.10 (0.85-1.41) | .47 |
| Clinical comorbidities | ||
| Myopia | 1.21 (0.95-1.52) | .12 |
| Pseudophakia | 1.25 (1.12-1.39) | <.001 |
| Vitreous hemorrhage | 1.22 (1.09-1.38) | .001 |
| Lattice degeneration | 0.92 (0.81-1.04) | .20 |
| Diabetes | 1.13 (0.99-1.28) | .06 |
| Charlson Comorbidity Index | ||
| 0 | — | — |
| 1-2 | 1.02 (0.92-1.13) | .73 |
| 3-4 | 1.19 (1.01-1.40) | .043 |
| 5+ | 1.25 (1.02-1.52) | .03 |
| Primary RRD repair procedure | ||
| Laser | 0.49 (0.42-0.56) | <.001 |
| Pneumatic retinopexy | 1.90 (1.70-2.12) | <.001 |
| Scleral buckle | 1.13 (0.99-1.29) | .06 |
| Vitrectomy | — | — |
| Region | ||
| East North Central | 1.16 (0.96-1.40) | .13 |
| East South Central | 1.34 (1.02-1.75) | .03 |
| Middle Atlantic | 1.18 (0.94-1.47) | .15 |
| Mountain | 1.16 (0.96-1.42) | .13 |
| New England | 1.29 (1.00-1.66) | .052 |
| Pacific | — | — |
| South Atlantic | 1.23 (1.03-1.46) | .02 |
| Unknown | 1.04 (0.50-2.18) | .91 |
| West North Central | 1.02 (0.83-1.23) | .88 |
| West South Central | 1.32 (1.09-1.60) | .004 |
| Days from primary RRD diagnosis to primary RRD repair | 0.9978 (0.9958-0.9998) | .03 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
