Association of Opioids With Incisional Ocular Surgery

Key Points

Question  What is the rate of filled opioid prescriptions around the time of incisional ocular surgeries, and how has that rate changed with time?

Findings  In this cohort study querying a large national US insurer’s claims database, we found an increasing trend of filled opioid prescriptions over the course of all years studied (2000-2016). Race/ethnicity, education, yearly income, and geographic location affected the rate of filled opioid prescriptions.

Meaning  Further research may be warranted to study the trend of increasing rates of filled perioperative opioid prescriptions identified in this investigation.

Importance  Opioid abuse has been declared a public health emergency. Currently, little is known about the association between opioids and ocular surgery.

Objective  To characterize rates of filled opioid prescriptions after incisional ocular surgeries.

Design, Setting, and Participants  This cohort study included patients with incisional ocular surgeries within a large national US insurer’s administrative medical claims database. All incisional ocular surgeries from January 2000 through December 2016 were evaluated. An opioid prescription was eligible if it occurred from 1 day before to 7 days after a surgery. Any surgery on a patient who was younger than 18 years, had more than 30 consecutive days of an opioid prescription in the prior 6 months, or had less than 6 months of data in the database prior to surgery was excluded. Data analysis occurred from May 2018 through November 2018.

Main Outcomes and Measures  The rate of opioid prescriptions filled for all incisional ocular surgeries from 2000 through 2016. Primary analysis looked at the rate of filled opioid prescriptions for each ophthalmic subspecialty surgery over time. Secondary analysis assessed which patient or surgical characteristics (ie, age, sex, race/ethnicity, geographic locations, yearly income, educational level, and type of eye surgery) were associated with filling an opioid prescription. Multivariate logistic regression using generalized estimating equations was used to determine odds ratios (ORs) of filling an opioid prescription.

Results  A total of 2 407 962 incisional ocular surgeries were included, of which 45 776 (1.90%) were associated with an opioid prescription. The rate of filled opioid prescriptions varied considerably over time, with the lowest rate occurring in the 2000-2001 cohort year (671 of 45 776 [1.24%]) and the highest in 2014 (5559 of 45 776 [2.51%]). An increasing trend was seen over the course of the study (2000-2001: 671 of 45 776 [1.24%]; 2016: 5851 of 45 776 [2.07%]; P < .001). Multivariate logistic regression showed that year of surgery was significantly associated with filling an opioid prescription, with the highest odds in 2014 (OR, 3.71 [95% CI, 3.33-4.1]), 2015 (OR, 3.33 [95% CI, 2.99-3.70]), and 2016 (OR, 3.27 [95% CI, 2.94-3.63]) compared with 2000 to 2001 (P < .001).

Conclusions and Relevance  These findings suggest the rate of filled opioid prescriptions are increasing for all types of incisional ocular surgery over time. Given the ongoing national opioid epidemic, understanding patterns of use can help in reversing the epidemic.

The last few decades have seen a considerable increase in opioid-associated deaths in the United States. As per the US Centers for Disease Control and Prevention, more than 200 000 people have died from opioid-associated overdoses in the United States from 1999 through 2016, with 5 times more people dying from prescription opioids in 2016 than in 1999; in US adults younger than 50 years, drug overdose has become the number 1 cause of mortality, and the rates are higher in men than women and higher in non-Hispanic white individuals than other racial/ethnic groups.^1,2 In addition, the rates are increasing in the older population as well.³ This prompted the establishment of a 5-point plan to combat opioid abuse, misuse, and overdose by the US Department of Health and Human Services in 2017.⁴

While opioids have proven effective analgesics in a wide range of ophthalmic settings, attention is turning to how they are used in ophthalmology.^5-10 Patel and Sternberg¹¹ conducted a study characterizing opioid-prescription patterns by ophthalmologists within Medicare. They showed that about 90% of ophthalmologists wrote fewer than 10 opioid prescriptions annually, and approximately 1% wrote more than 100 prescriptions annually. Charlson and coworkers¹² specifically focused on oculoplastic surgeons and found that they wrote a mean of 45 opioid prescriptions per year, with 45% writing fewer than 10 and 14.8% writing more than 100 prescriptions per year; male surgeons wrote higher numbers of prescriptions (at higher rates) than female surgeons did; and older surgeons wrote fewer prescriptions than younger surgeons did.

In light of the current opioid epidemic and the increased risk of addiction and overdose with each episode of opioid use,^13,14 we set out to characterize perioperative opioid use by using data from a large national US insurer’s administrative medical claims database. While data exist on overall opioid-prescribing patterns by ophthalmic surgeons,^11,12 to the best of our knowledge, patient-associated and surgery-associated data on opioids prescribed around the time of ophthalmic surgery have yet to be examined.

The Clinformatics Data Mart Database (OptumInsight) is a large national US insurer’s administrative medical claims database containing deidentified beneficiary data. It includes all outpatient medical claims, associated diagnoses, pharmaceutical prescriptions filled, and demographic data during enrollment in the insurance plan. The University of Pennsylvania’s institutional review board declared this study exempt from review because all involved data were deidentified, and patient consent was waived.

All surgeries on patients 18 years or older who underwent incisional ocular surgery from 2000 through 2016 and had least 6 months of uninterrupted enrollment in the database prior to day of surgery were eligible for inclusion. Each surgery (including patients with multiple surgeries) was individually assessed for eligibility to be certain it met all inclusion and exclusion criteria. The date of surgery was considered the index date. The patients must also have remained in the database for at least 14 days after the index date. To decrease the chance that a prescription was given for something other than the eye surgery, all surgeries were excluded if the patient had greater than 30 continuous days of an opioid prescription within the 6 months prior to the index date. (We considered such patients to be recipients of long-term pain medication.) Surgeries were then analyzed as a single combined cohort, as well as divided into subspecialty surgery categories. (The eTable in the Supplement presents all Current Procedural Terminology [CPT] codes and subspecialty surgery groupings used in this study.)

Incisional eye surgery was defined as any surgical code in which a scleral or corneal incision would be made. In addition, strabismus surgery was considered for any CPT code that involved eye-muscle transposition, and ocular trauma was considered for all CPT codes for open or ruptured globe repair. Surgeries were further classified as single, combined, or sequential, with single surgeries being further subdivided into cataract, glaucoma, cornea, retina, strabismus, and trauma categories. Combined surgeries were defined as any 2 or more surgeries from different ophthalmology subspecialties performed on the same day. Sequential surgeries were defined as at least 1 additional surgery from the same or different subspecialty within 30 days of the index date. The lone exception to the definition of sequential surgery was when a cataract surgery was performed on a defined eye, followed by cataract surgery on the fellow eye. This situation was simply counted as 2 distinct cataract surgeries. Both combined and sequential surgeries were thought to represent unique situations, which could increase the likelihood of filling an opioid prescription through increased duration of surgery, invasiveness, and degree of induced inflammation. Each combined or sequential surgery was also counted within each appropriate subspecialty category (ie, a combined cataract and glaucoma surgery was also counted separately in both the cataract surgery and glaucoma surgery categories). Accounting for surgeries in this manner raised the possibility of double counting; therefore, an additional analysis was run that removed combined and sequential categories and divided the retina group into vitrectomy and scleral-buckle subcategories.

An incident of opioid prescription was considered to have occurred when a patient filled an opioid prescription between 1 day prior to the index date or up to 7 days after that date. National Drug Codes (NDC) were used to determine which prescriptions contained opiates. A list of all NDC codes was searched for any oral drugs that were included within the pharmaceutical class opioid. Our primary analysis evaluated the trend in filled opioid prescriptions from 2000 through 2016. To more easily assess changes in prescription fill rates over time, we performed an analysis with grouped cohort years (2000-2004, 2005-2008, 2009-2012, and 2013-2016). We also ran a sensitivity analysis that would only count a prescription if it were filled 1 day before the index date and up to 4 days, instead of 7 days, after that date.

Multivariate logistic regression using generalized estimating equations was used to account for possible nonindependence for patients who had multiple surgeries within the observation period. A secondary analysis aimed to identify patient and surgical characteristics associated with filling an opioid prescription. All patient characteristics were determined at the time of the index date. Tested categories included age, sex, race/ethnicity, geographic locations, yearly income, educational level, and type of eye surgery. Means and SDs were used to summarize continuous variables, while frequencies and percentages were used for categorical variables.

Statistical analysis was performed from May 2018 through November 2018 using SAS version 9.4 (SAS Institute Inc). Findings were considered statistically significant at P < .05.

A total of 2 984 918 incisional ocular surgeries occurred on patients 18 years or older from 2000 through 2016 (Figure 1). Overall, 1 163 940 patients (mean [SD] age, 69.8 [10.6] years and 937 688 male [41.4%]) had 2 407 962 incisional ocular surgeries that were included in the analysis.

The years 2000 and 2001 were combined because of the low number of surgeries meeting inclusion criteria in 2000. Surgical numbers and rates of opioid prescription filled overall, and subspecialty-specific categories for each study year can be seen in Table 1. Of all incisional ocular surgeries, 45 776 were associated with filling an opioid prescription, for a total rate of 1.90%. However, the rate varied considerably, with the lowest rate occurring in the 2000-2001 cohort year (671 of 45 776 [1.24%]) and the highest rate in 2014 (5559 of 45 776 [2.51%]). An increasing trend was seen over the course of the study (2000-2001: 671 of 45 776 [1.24%]; 2016: 5851 of 45 776 [2.07%]; P < .001) (Figure 2).

Within subspecialty categories, cataracts had the highest number of surgeries (2 060 766). Although the total number of opioid prescriptions (19 494) was the highest in this group, given the large number of cataract surgeries performed, this category had the lowest rate of opioid prescriptions filled per surgery (0.95%). Because of the drastically higher number of cataract surgeries compared with other subcategories, this category had the most influence on the overall 1.90% opioid prescription fill rate (the next highest was glaucoma, at 141 429 surgeries). Excluding cataract surgeries, the rate of filled opioid prescriptions per incisional surgery was 26 282 (7.57%), with the highest rates associated with strabismus surgery (5296 [19.52%]), trauma surgery (459 [19.22%]), and retina surgery (overall, 15 716 [11.15%]; 3569 [14.72%] for scleral-buckle surgery and 12 147 [10.41%] for vitrectomy).

Table 2 shows the raw numbers and rates of opioid prescription filled by surgery for individual subcategories of surgery. Many of these subcategories were found to be significantly associated with filling an opioid prescription compared with cataract surgery. The year of surgery was significantly associated with filling an opioid prescription; after controlling for all covariates in the multivariate analysis, these showed steadily increasing odds ratios (ORs) for filling an opioid prescription, with the highest odds in 2014 (OR, 3.71 [95% CI, 3.33-4.13]), 2015 (OR, 3.33 [95% CI, 2.99-3.70]), and 2016 (OR, 3.27 [95% CI, 2.94-3.63]) compared with 2000 and 2001 (P < .001 for all comparisons). Grouping the years into 4-year and 5-year cohorts showed a similar trend, with the highest OR occurring in the 2013-2016 group (2013-2016 vs 2000-2004: OR, 2.51 [95% CI, 2.39-2.64]; P < .001). The types of surgeries with the highest odds of filling an opioid prescription were trauma surgery (OR, 19.91 [95% CI, 17.50-22.65]), strabismus surgery (OR, 16.74 [95% CI, 15.85-17.69]), and retina surgery (OR, 12.35 [95% CI, 12.03-12.68]) compared with cataract surgeries. An additional analysis that removed combined and sequential surgery-type categories and also separated the retina group into vitrectomy and scleral buckle subcategories showed that the opioid prescription fill rates associated with scleral-buckle surgery (OR, 14.05 [95% CI, 13.45-14.68]) and vitrectomy surgery (OR, 11.11 [95% CI, 10.81-11.42]) were significantly greater than the rate for cataract surgery (P < .001 for both comparisons). Further details are in Table 3.

The sensitivity analysis that only counted prescriptions that were filled either 1 day before the index date or up to 4 days instead of 7 days after the index date saw a small drop in total numbers of prescriptions to 39 090 (instead of 45 776). This reduction in number of outcomes did not change any of the results for either surgeries or associations with patient characteristics.

Many patient characteristics were significantly associated with filling an opioid prescription. Race/ethnicity group data were mixed, with black individuals having higher odds of filling a prescription (OR, 1.24 [95% CI, 1.19-1.29]) and Asian individuals (OR, 0.61 [95% CI, 0.56-0.66]) and Hispanic individuals (OR, 0.76 [95% CI, 0.72-0.80]) having lower odds (P < .001 for all comparisons) compared with white individuals. Geographic location was highly associated with odds of filling an opioid prescription, with all regions in the United States having higher rates than those in the Northeast (reference) and all regions having lower rates than those in the Mountain region (OR, 1.87 [95% CI, 1.77-1.98]; P < .001). Yearly income showed an association with filled opioid prescriptions, in which patients in the $75 000 to $99 000 income group (OR, 0.96 [95% CI, 0.92-1.00]) and those in the $50 000 to $59 000 income group (OR, 0.94 [95% CI, 0.89-0.99]) were less likely to fill a prescription than those making more than $100 000 (P < .001 for both comparisons). Education was also significantly associated with prescription filling, with patients who had a high school diploma or less more likely to fill a prescription than those with more than a bachelor’s degree (OR, 1.07 [95% CI, 1.03-1.12]; P < .001). Lastly, male patients had significantly but only slightly higher odds (OR, 1.02 [95% CI, 1.00-1.05]; P = .02), whereas each additional year of age was associated with lower odds (OR, 0.98 [95% CI, 0.97-0.98]; P < .001). Table 2 presents full multivariate results.

With an increase in opioid-associated deaths,¹ many surgical subspecialties have recently examined their opioid-prescribing patterns.^11,15-17 To date, only 2 studies^11,12 appear to have specifically focused on ophthalmology, with one concluding that “ophthalmologists show discretion in their opioid prescribing patterns.”^11(p1216) While reassuring that most ophthalmologists are not writing prescriptions for large amounts of opioids, the characteristics of the physicians who write fewer than 10 prescriptions per year is unclear.^11,12 It is possible that this proportion includes numerous physicians who do not perform surgery or offer opioids but treat patients who have insurance besides Medicare, which was the focus of the previous studies.^11,12

We chose to compliment this work by evaluating the situation most likely to elicit an opioid prescription (ie, incisional surgery) while accounting for the patient and surgical factors associated with these prescriptions. We found an increasing rate of filled opioid prescriptions after incisional eye surgery throughout the study period (Tables 1 and 3). Importantly, this finding was consistent over every type of subspecialty surgery evaluated. Trauma, strabismus, and retina surgeries had the highest rates of filled opioid prescriptions per surgery, as well as highest odds of a patient filling an opioid prescription in comparison with cataract surgery (Table 2). The US Department of Health and Human Services declared opioid abuse a public health emergency in 2017, outlining a 5-point strategy to reducing the toll opioids are having on US society.⁴ Unfortunately, the data available were only inclusive of 2016, and as such, any assessments of this declaration’s influence on opioid prescribing for ocular surgery would only be speculative.

Demographic characteristics found to be associated with a significantly higher rate of filled opioid prescriptions included black race, male sex, education to high school completion or less, non-Northeast location, and performance of any incisional surgery other than cataract extraction. Similar socioeconomic characteristics have been previously shown to be significantly associated with long-term opioid use or abuse and deaths associated with illicit opioid use.^17-19 Significantly decreased opioid prescription rates in the Northeast have been found in other studies,^11,15 and these may be attributable to strict, state-specific opioid-prescribing regulations.²⁰ To help reduce unnecessary prescribing, the US Centers for Disease Control and Prevention has published guidelines for prescribing opioids for chronic pain; selecting specific opioids and dosing regimens, duration, and follow-up; and assessing opioid risks and addressing harms.²¹ Other alternatives to opioids also need to be evaluated. One recent report showed topical drops reduced intravenous opioid use in the postoperative setting.²²

It is unclear why the odds of filling an opioid prescription would be 3.27 to 3.71 times higher in 2014 through 2016 compared with 2000 and 2001. The recent trends in ophthalmic surgery have included smaller corneal incisions vs scleral tunnel cataract surgery, endothelial vs full-thickness corneal transplants, microincisional vs traditional glaucoma surgery, small (23-gauge to 27-gauge) vitrectomy vs 20-gauge vitrectomy, and decreasing numbers of scleral-buckle procedures performed. The reduced invasiveness of these procedures in this time frame would suggest that the surgeries inherently would not be responsible for the increasing rates of filled opioid prescriptions for all these subspecialty specific incisional surgeries (Tables 1 and 3). An additional major factor to consider is the shift toward topical vs peribulbar or retrobulbar anesthesia, which has been associated with increased perception of surgical pain and possibly with patients desiring stronger postoperative analgesia.²³

While the cause of this paradoxical increase remains unclear, it is likely multifactorial. Some potential reasons may include (1) insufficient physician education regarding proper opioid-prescribing patterns and/or assessment of pain²⁴; (2) increased focus on pain management since the introduction of pain scales in the early 2000s and signing of the Pain Relief Promotion Act of 2000 (HR 2260) by the Clinton administration²⁵; (3) lack of standardization or consensus of state and/or national regulations on opioid prescription²⁶; (4) increased physician emphasis on pain control as patients’ so-called positive perception of pain has been correlated with improved hospital and physician satisfaction scores²⁷; and (5) drug-company marketing practices.

An important caveat to our results needs to be noted. These data are the result of a prescription being filled, which is a patient-centered issue, and the data set is unable to account for prescriptions that were written but not filled. We assume that the increase in filled prescriptions is associated with a proportional increase in written prescriptions; however, we are unable to verify this. While other studies reporting increasing rates of opioid prescriptions written in the United States give us confidence in making this assumption,^19,28 another possibility is that patient behavior toward feeling the need for an opioid after surgery has changed. While this cannot be ruled out, no data currently exist to suggest that a patient’s inclination to fill opioid prescriptions has increased over the years. Furthermore, while not specific to ophthalmology, a recent otorhinolaryngologic study also supports the assumption of more frequent opioid prescribing; it showed a near doubling of otorhinolaryngology-associated narcotic prescriptions written from 2008 to 2011, and although the rate decreased from 2011 to 2014, the 2014 rate was still far higher than the pre-2008 levels.²⁹

A number of other important limitations inherent to the study merit consideration. Although the database used is a national claims database with more than 60 million individuals, the data may not generalize to patients outside of this database (such as surgeries conducted within the Veterans Affairs health system or uninsured populations). The database was queried using CPT codes and pharmaceutical claims, which could not be verified by reviewing individual medical records. Also, we were unable to control for the precise indication for each opioid prescription, despite a stringent criterion for the temporal association between prescription and surgery date. Similarly, we were unable to account for the possibility that prescriptions filled were actually written by nonophthalmologists. We attempted to limit this by using a very broad definition of patients with chronic pain and excluding them; however, this may still have affected the results. Similarly, we are unable to be certain that the restriction of removing all patients who had 30 consecutive days of opioids did in fact remove most patients with opioid addictions. Next, the database does not provide information regarding physician sex or level of experience, both of which have been previously shown to be associated with opioid-prescribing patterns.¹⁵ Lastly, similar to other database studies, we were unable to control for family members filling prescriptions for their friends or significant others (which is known as medication sharing). In addition, we are not able to verify how much of the medication was actually used by the patient, if any were at all. However, for this to have affected the results of this study, the rate of prescription filling without use would have had to increase disproportionately in more recent years, which cannot be ruled out but seems unlikely.

In conclusion, we found that the odds of having an opioid prescription filled after incisional ocular surgery was more than 3 times higher in 2014 through 2016 compared with 2000 through 2001. The precise cause of this trend remains unclear but warrants further examination. It is our hope that these findings shed light on the trends in opioid prescription–filling patterns after incisional surgery and will motivate clinicians to evaluate their opioid-prescribing practices. Although we would never argue to deny a patient in pain the help they need, we believe that, given the broader context of the current national opioid crisis, now is the proper time to discuss the role of opioids in postophthalmic surgical management. These results provide a basis for holding such a discussion. We believe that ophthalmologists should engage with their patients with regard to opioid use for any indication and discuss possible nonopioid (alternative) pain-management strategies.

Accepted for Publication: July 14, 2019.

Corresponding Author: Brian L. VanderBeek, MD, MPH, MSCE, Scheie Eye Institute, Perelman School of Medicine, Department of Ophthalmology, University of Pennsylvania, 51 N 39th St, Philadelphia, PA 19104 (brian.vanderbeek@uphs.upenn.edu).

Published Online: September 19, 2019. doi:10.1001/jamaophthalmol.2019.3694

Author Contributions: Dr VanderBeek had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kolomeyer, VanderBeek.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kolomeyer, VanderBeek.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Yu, VanderBeek.

Obtained funding: VanderBeek.

Administrative, technical, or material support: Kolomeyer.

Supervision: VanderBeek.

Conflict of Interest Disclosures: Dr VanderBeek reported grants from the National Eye Institute/National Institutes of Health, Research to Prevent Blindness, and the Paul and Evanina Mackall Foundation during the conduct of the study. No other disclosures were reported.

Funding/Support: This work was supported by the National Institutes of Health K23 Award (1K23EY025729–01 [Dr VanderBeek]) and University of Pennsylvania Core Grant for Vision Research (2P30EYEY001583). Additional funding was provided by Research to Prevent Blindness, the Paul and Evanina Mackall Foundation, and the Heed Ophthalmic Research Foundation in the form of block research grants to the Scheie Eye Institute.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.