To determine provider compliance with hydroxychloroquine screening following the revised recommendations published in 2011 by the American Academy of Ophthalmology.
Evaluation of adherence to a screening protocol.
Subjects were identified with hydroxychloroquine as a medication by electronic query at a large multispecialty ophthalmic practice. Patients were excluded if patients: (1) were screened by an outside physician; (2) lacked recorded height, weight, start date, or dosing; or (3) took hydroxychloroquine for malaria prophylaxis. Screening tests were stratified by ophthalmic subspecialty. Guidelines define proper screening as 1 subjective test—Humphrey visual field (HVF), and 1 objective test—spectral-domain optical coherence tomography (SD OCT), fundus autofluorescence (FAF), or multifocal electroretinography (mfERG). Adherence to guidelines was determined by categorizing practices as: (1) “appropriate”—consistent with guidelines; (2) “underscreened”—insufficient testing; or (3) “inappropriate”—no testing.
The study comprised 756 patients with a mean age of 56 years undergoing 1294 screening visits. Twenty-one patients received initial screenings outside the institution. Most common screening tests employed included SD OCT (56.6%), 10-2 HVF (55.0%), and Amsler grid (40.0%). Of the 735 initial screenings, 341 (46.4%) were appropriately screened, 204 (27.8%) underscreened, and 190 (25.9%) inappropriately screened. Of those who presented solely for screening (560), 307 (54.8%) were appropriately screened, 144 (25.7%) underscreened, and 109 (19.5%) inappropriately screened.
Of patients presenting for hydroxychloroquine screening, 54.8% of patients received appropriate evaluation, indicating lack of adherence to guidelines. Overall, SD OCT and 10-2 HVF were the preferred screening modalities, with FAF and mfERG less frequently ordered.
Hydroxychloroquine retinopathy is a well-described long-term potential side effect of chronic therapy. Permanent vision loss may occur with development of the characteristic bilateral bull’s-eye maculopathy. Although hydroxychloroquine has demonstrated a more favorable side effect profile with decreased ocular toxicity compared to chloroquine, the risk for retinopathy is still present, with rates varying from 1% to as high as 7.5% in patients with long-term exposure.
In 2011, the American Academy of Ophthalmology (AAO) revised its 2002 guidelines for hydroxychloroquine retinal toxicity screening. The intended goal of the new screening guidelines was to detect functional and anatomic abnormalities related to toxicity at an early stage, with the hope of minimizing irrevocable central blindness. In contradistinction to the 2002 guidelines, the 2011 guidelines recommended subjective testing with a 10-2 Humphrey visual field (HVF) that could no longer be substituted by the previously accepted Amsler grid test. An alternatively accepted objective test in the new guidelines includes a multifocal electroretinogram (mfERG). Additional recommended anatomic tests to detect subtle anatomic change include spectral-domain optical coherence tomography (SD OCT) and fundus autofluorescence (FAF).
The guidelines further reiterated specific risk factors associated with toxicity and made screening guidelines accordingly. The designation of high-risk included patients with: (1) cumulative hydroxychloroquine consumption >1 kg; (2) daily dosing >6.5 mg/kg/day of ideal body weight; or (3) concomitant renal or liver disease. Additional, albeit less definitive, risk factors included advanced age or comorbid retinal or macular disease. For those without high-risk characteristics, a baseline screening upon initiation of hydroxychloroquine was recommended followed by a 5-year examination-free window. For patients with high risk factors, annual screening was recommended.
Despite the refined guidelines and increased availability of testing, overall adherence to the guidelines, particularly in long-term hydroxychloroquine users, has been reportedly poor. Nika and associates evaluated long-term users of chloroquine and hydroxychloroquine and demonstrated that a third of high-risk patients did not receive appropriate diagnostic testing and just under a third lacked regular eye care. Additionally, Browning showed poor documentation of patient height, weight, daily dose, and cumulative dose that would otherwise be helpful in risk stratification; a reliance on mfERG and SD OCT that was likely responsible for increasing the cost of screening; and the absence of increased toxicity detection since the new recommendations. These studies suggest poor overall compliance and highlight the difficulty of determining the true rate of hydroxychloroquine toxicity.
In this study, we report clinician patterns of hydroxychloroquine screening in a large multispecialty ophthalmology practice. Furthermore, we investigate whether differences in eye care specialization affect choice of screening test and whether or not that correlates with screening guideline compliance.
This is an evaluation of adherence to a screening protocol of patients taking hydroxychloroquine between March 11, 2011 (1 month after the 2011 guidelines publication) and September 18, 2014. Waiver of informed consent and waiver of HIPAA authorization were approved by the Cleveland Clinic Institutional Review Board. Inclusion criteria were patients of the Cleveland Clinic Cole Eye Institute who had hydroxychloroquine listed as an active medication in the electronic medical record. Medical records of 881 patients were identified. The initial 881 patients were filtered to 756 patients based on the following exclusion criteria: (1) screening was performed by an outside physician (12 patients, 1.4%); (2) lacked recorded height or weight in the electronic medical record (25 patients, 2.8%); (3) had unavailable hydroxychloroquine start date or dosing (35, 4.0%); and (4) hydroxychloroquine was prescribed for malaria prophylaxis (53, 6.0%). Of the 756 patients, 735 had initial screening encounters and 559 underwent subsequent follow-up screening examinations within the evaluation period. Baseline examination was defined as patients who had been seen for hydroxychloroquine screening within 1 year of starting medication.
Data collected from the electronic medical record included baseline demographic factors (such as height/weight), date hydroxychloroquine was initiated, hydroxychloroquine dose, ancillary tests used, past medical history, pre-existing ophthalmologic disease, examination findings, screening tests employed, recommended follow-up interval, and provider recommendations. Follow-up intervals were calculated based on provider follow-up recommendation rather than date of next visit. Screening tests included: Amsler grid, color vision (Ishihara color plates), 10-2 HVF, 24-2 HVF, 30-2 HVF, SD OCT, mfERG, FAF, fluorescein angiography (FA), and fundus photographs. While all physicians had access to these tests at the Cole Eye Institute, FAF, FA, and mfERG were not available at satellite offices. However, it was possible during the reporting period to refer the patient into the main office to complete the testing with these modalities.
Patients’ most recently reported height, weight, and presence of comorbid liver or renal diseases based on electronic medical records of the initial visit were recorded. Cumulative dosing was calculated based on available information concerning years of therapy and total daily dosage. Assumptions in this calculation included 100% adherence and no breaks in treatment. Ideal body weight was calculated from height using the Devine formula. Daily dosing (mg/kg/day) was calculated for both the actual body weight at time of initial presentation and ideal body weight. Stratification into a high-risk group was based upon having 1 of the 3 high-risk criteria: (1) total hydroxychloroquine consumption >1 kg; (2) daily dosing ≥6.5 mg/kg/day based on ideal body weight; and (3) concomitant renal or liver disease. Although the original guidelines suggested age, macular disease, and genetic factors as possible risk factors, these features were not included in the designation of high-risk characteristics, given a theoretical but not substantiated increased risk. Screening tests and follow-up intervals for high- and low-risk patients were analyzed and stratified by ophthalmic subspecialty of the doctor performing the visit.
To evaluate adherence to the 2011 screening practice guidelines, tests were defined as either subjective tests (10-2 HVF) or objective tests (SD OCT, mfERG, FAF). An “appropriate” screening was defined as a 10-2 HVF and at least 1 objective test. Subcategorization differed by the amount of objective testing. “Underscreened” was defined as having only a 10-2 HVF or objective tests. “Inappropriate” was defined as no 10-2 HVF or objective testing (SD OCT, FAF, mfERG). These terms specifically applied to initial baseline screening of patients whose encounter was solely for hydroxychloroquine toxicity screening.
A clinical diagnosis of hydroxychloroquine toxicity was made based on history, parafoveal attenuation or loss of the ellipsoid zone on SD OCT, parafoveal areas of abnormal hyperautofluorescence on FAF, or funduscopic examination demonstrating concentric or central macular retinal pigmentary epithelial changes. These findings were correlated with functional testing via static perimetry testing (HVF).
Statistical analysis comparing practice patterns between subspecialists was carried out using a Poisson regression with pairwise comparisons for individual subspecialist differences with JMP 10 Pro Software ( www.jmp.com ). Statistical significance was determined with a P value less than .05 for overall practice patterns. Bonferroni correction for statistical significance was used for multiple comparisons with P values determined by .05/n (the number of tests performed). Statistical significance for subspecialist practices was set to P < .002 (.05/28), while P value for doctor of medicine (MD) vs doctor of optometry (OD) practices was set to P < .005 (.05/10).
The study comprised 756 patients with a mean age of 56 years undergoing 1294 screening visits. Six hundred and sixty-one patients (87.4%) were female. Two hundred and eighty-one (37.2%) and 253 (33.5%) were diagnosed with rheumatoid arthritis and systemic lupus erythematosus, respectively. The mean duration of treatment was 6.3 years (range: 0–46.1 years) with 354 (46.8%) having received hydroxychloroquine treatment for 5 years or greater. The most common dose prescribed was 400 mg/day (517 patients, 68.4%). Additional patient demographics are reviewed in Table 1 .
|Average age (median, mean, range)||56, 55.84, 12–93|
|Male (n, %)||95, 12.6%|
|Female (n, %)||661, 87.4%|
|Diagnosis of patients screened (n, %)|
|Rheumatoid arthritis||281, 37.2%|
|Juvenile rheumatoid arthritis||7, 0.9%|
|Systemic lupus erythematosus||253, 33.5%|
|Sjogren syndrome||114, 15.1%|
|Other a||221, 29.23%|
|Dose of hydroxychloroquine (n, %)|
|>400 mg/day||2, 0.26%|
|400 mg/day||517, 68.4%|
|201–399 mg/day||29, 3.8%|
|≤200 mg/day||208, 27.5%|
|Actual body weight in kg (mean, range)||79.6, 34.9–169.2|
|Ideal body weight in kg (mean, range)||56.7, 20.2–89.1|
|Dosing by ideal body weight (n, %)|
|Patients receiving <6.5 mg/kg/day||374, 49.5%|
|Patients receiving ≥6.5 mg/kg/day||382, 50.5%|
|Duration of hydroxychloroquine treatment|
|0–5 years (n, %)||402, 53.2%|
|5–15 years (n, %)||292, 38.6%|
|15+ years (n, %)||62, 8.2%|
|Patients receiving <1 kg lifetime dose (n, %)||553, 73.2%|
|Patients receiving ≥1 kg lifetime dose (n, %)||203, 26.9%|
|Comorbid systemic disease (n, %)|
|Renal disease||17, 2.3%|
|Liver disease||60, 7.9%|
|Risk stratification (n, %)|
|High risk b||497, 65.7%|
|Low risk||259, 34.3%|
a Ankylosing spondylitis, arthritis, arthropathy, autoimmune disorder, autoimmune enteropathy, autoinflammatory syndrome, chronic idiopathic urticarial, chronic polychondirits, connective tissue disease (CTD), CREST, Crohn disease, cutaneous vasculitis, dermatomyositis, diffuse CTD, osteoarthritis, fibromyalgia, Kikuchi-Fujimoto syndrome, inflammatory arthritis, inflammatory polyarthropathy, juvenile idiopathic arthritis, Lambert-Eaton myasthenic syndrome, lichen planopilaris, lymphomatoid hyperplasia, mixed CTD, multiple sclerosis, palimdromic rheumatism, paraneoplastic arthritis, pemphigus foliaceous, polymyalgia rheumatica, psoriatic arthropathy, Raynaud syndrome, reactive arthritis, sarcoidosis, scleroderma, seronegative arthritis, seronegative inflammatory arthritis, Still disease, synovitis, systemic sclerosis, ulcerative colitis, undifferentiated CTD, unknown, urticarial vasculitis, Wegener granulomatosis.
A total of 203 patients (26.9%) received a cumulative lifetime dose exceeding 1 kg. Based on ideal body weight calculations, 382 patients (50.5%) exceeded ≥6.5 mg/kg/day. Seventy-seven patients (10.2%) had either comorbid liver or renal disease. Cumulatively, 497 (65.7%) were identified as high-risk ( Tables 1 and 2 ).
|Descriptions of macular findings on physical examination a (n, %)|
|Retinal pigment epithelium changes||312, 12.0%|
|Bull’s-eye maculopathy||8, 0.3%|
|Other b||424, 16.4%|
|Comorbid macular disease (n, %)|
|Age-related macular degeneration||47, 6.2%|
|Cystoid macular edema||11, 1.5%|
|Epiretinal membrane||27, 3.6%|
|Macular hole||5, 0.7%|
|Vitreomacular traction||2, 0.3%|
|Other c||9, 1.2%|
b Choroidal folds, cotton-wool spots, diabetic macular edema, flat macula, sclerotic vessels, retinal edema, retinal whitening, dot-blot hemorrhage, intraretinal hemorrhage, intraretinal edema, subretinal fluid, macular hemorrhage, no record, congenital hypertrophy and hyperplasia of the retinal pigment epithelium, subfoveal deposits, microaneurysm, starfold, lamellar hole, pigmented lesion, fibrosis.
Overall screening characteristics are reviewed in Table 3 . Of the 735 initial screening encounters, 560 (76.2%) presented exclusively for hydroxychloroquine screening while 175 (23.8%) presented with a non-hydroxychloroquine-related chief complaint. The most common initial screening tests employed included SD OCT (56.6%), 10-2 HVF (55.0%), and Amsler grid (40.0%). Follow-up screening tests recapitulated this trend with 63.5% receiving SD OCT, 42.4% receiving 10-2 HVF, and 31.3% receiving Amsler grid testing. FAF was infrequently performed, in 97 (13.2%) patients, and mfERG was rarely performed, in 3 (0.4%) patients, at initial screening. Of note, 12 patients (1.6%) were found to have signs of toxicity determined by treating physicians.
|Number of patients screened||756|
|Total number of screening encounters||1294|
|Initial hydroxychloroquine screening encounters (n)||735|
|Chief complaint hydroxychloroquine screening (n, %)||560, 76.2%|
|Chief complaint non-hydroxychloroquine-related (n, %)||175, 23.8%|
|Baseline examinations a (n, %)||207, 28.2%|
|Follow-up hydroxychloroquine screening encounters (n)||559|
|Number of doctors screening||50|
|Tests employed (initial screening) (n, %)|
|Spectral-domain optical coherence tomography||416, 56.6%|
|Humphrey visual field|
|Fundus autofluorescence||97, 13.2%|
|Multifocal electroretinogram||3, 0.4%|
|Amsler grid||294, 40.0%|
|Color testing, Ishihara||236, 32.1%|
|Fluorescein angiography||17, 2.3%|
|Fundus photographs||70, 9.5%|
|Tests employed (follow-up screening) (n, %)|
|Spectral-domain optical coherence tomography||355, 63.5%|
|Humphrey visual field|
|Fundus autofluorescence||91, 16.3%|
|Multifocal electroretinogram||5, 0.9%|
|Amsler grid||175, 31.3%|
|Color testing, Ishihara||138, 24.7%|
|Fluorescein angiography||13, 2.3%|
|Fundus photographs||45, 8.1%|
|Hydroxychloroquine toxicity (determined by provider) (n, %)|
|Recommended follow-up (initial screening) Patients with no toxicity and no comorbid macular disease (n = 653) (n, %)|
|<6 months||54, 8.3%|
|6–11 months||68, 10.4%|
|1 year||445, 68.2%|
|1–5 years||3, 0.5%|
|5+ years||2, 0.3%|
|As needed or pro re nata||81,12.4%|
|Patients with no toxicity and comorbid macular disease (n = 70) (n, %)|
|<6 months||18, 25.7%|
|6–11 months||17, 24.3%|
|1 year||24, 34.3%|
|1–5 years||1, 1.4%|
|5+ years||0, 0.0%|
|As needed or pro re nata||10, 14.3%|
|Patients with hydroxychloroquine toxicity (n = 12) (n, %)|
|<6 months||6, 50.0%|
|6–11 months||1, 8.3%|
|1 year||1, 8.3%|
|1–5 years||0, 0.0%|
|5+ years||0, 0.0%|
|As needed or pro re nata||4, 33.3%|
There were a total of 50 providers (38 MDs and 12 ODs) performing screening examinations. Optometrists (24%) made up the largest screening group among the various specialties. Table 4 lists tests ordered by specialty. Notable differences in screening patterns include: (1) retina specialists were more likely to order SD OCT ( P = 1.1E-04), FAF ( P = 1.5E-0.5), and fundus photographs ( P = 2.9E-0.5) over all specialties; and (2) comprehensive ophthalmologists were more likely to order Amsler ( P = 2.3E-10) and Ishihara color plates ( P = 2.4E-10). Between ophthalmologists’ and optometrists’ screening practices, ophthalmologists were more likely to order all screening tests except HVF 30-2 ( P = .31) and mfERG ( P = .03) ( Supplemental Table , available at AJO.com ).
|Subspecialty (n)||Encounters||Amsler||Color||10-2 HVF||24-2 HVF||30-2 HVF||SD OCT||mfERG||FAF||FA||Fundus Photographs|
|n||% a||n||% a||n||% b||n||% b||n||% b||n||% b||n||% b||n||% b||n||% b||n||% b||n||% b||n||% b|
|Comprehensive (9)||303||41.2||185||33.1||243 c||49.8||177 c||36.3||150||30.7||12||2.5||1||0.2||240||49.2||0||0.0||61||12.5||9||1.8||29||5.9|
|Retina (9)||156||21.2||178||31.8||99||29.6||59||17.7||170||50.9||21||6.3||3||0.9||274 d||82.0||2||0.6||99 d||29.6||20||6.0||64 d||19.2|