Survey questions. The questions are numbered as referred to in the text. The numbering in each of the 3 surveys varied. After each question, the survey (state) in which the question was asked is given in brackets with any changes that may have been made in the question. The American Academy of Ophthalmology (AAO) Practice Pattern response is included. FL indicates Florida; MA1, first Massachusetts survey; MA2, second Massachusetts survey; NY, New York; and DM, diabetes mellitus.
Khadem JJ, Buzney SM, Alich KS. Practice Patterns in Diabetic RetinopathyPart 1: Analysis of Retinopathy Follow-up. Arch Ophthalmol. 1999;117(6):815-820. doi:10.1001/archopht.117.6.815
To evaluate ophthalmologists' management of diabetic patients.
A multiple-choice questionnaire was mailed to all ophthalmologists in New York (1985), Florida (1990), and Massachusetts (1993 and 1996). Questions included practice patterns, methods used in examination, use of photography and fluorescein angiography, indications for laser treatment, and intervals for follow-up of selected conditions. Responses were tabulated and compared between surveys and with the American Academy of Ophthalmology Preferred Practice Pattern.
In this first report, we detail follow-up patterns of various grades of retinopathy. Physicians increasingly used duration of diabetes as a criterion in determining the length of follow-up for adults, despite persistent ambiguities for children. There were notable changes over time in nearly all aspects of follow-up for both general ophthalmologists and retina specialists. Retina specialists were less likely to treat proliferative disease, more likely to follow up patients with preproliferative disease sooner, and used longer follow-up intervals for short-duration diabetic patients, whereas a small percentage of general ophthalmologists continued to recommend treatment for background disease.
There were noticeable trends toward Diabetes 2000 recommendations over time, although there remained many areas where further education appeared warranted.
RESULTS of recent nationwide controlled studies1- 4 demonstrate that effective therapy exists for diabetic retinopathy. The Diabetes 2000 program of the American Academy of Ophthalmology (AAO), formally established in 1990, is aggressively disseminating the recommendations of these various study groups in an effort to improve the overall quality of care for patients with diabetes and those specifically at risk for diabetic retinopathy. We surveyed Massachusetts ophthalmologists in 1993 and 1996 with regard to their care of patients with diabetes and compared these results with survey data from New York (performed in 1985)5 and Florida (performed in 1990).6 The surveys contained almost identical questions initially developed by the State of New York Department of Health (New York State Diabetes Control Program).
Our goal is to monitor the practice patterns of surveyed ophthalmologists in these successive surveys "over time." We demonstrate trends in patient care, including gains and shortcomings, compared with those promoted by the AAO and Diabetes 2000. In this first part of our report, we focus on the demographics of the populations polled and their recommendations for follow-up intervals of various diabetic patients and conditions.
This study was conducted by the Massachusetts Department of Public Health using a survey administered by the State of New York Department of Health5 in 1985 and the Florida Diabetic Retinopathy Task Force6 in 1990. The Florida and Massachusetts surveys differed from the original New York questionnaire by minor variations in wording and the addition of new questions (Figure 1). The surveys were mailed to all ophthalmologists known to the boards of medicine in these states, excluding retired, in-training, and out-of-state practitioners. The second Massachusetts survey (1996) was sent only to ophthalmologists who responded to the first mailing (1993).
For the New York and Florida studies, mailed questionnaires were followed up with 3 mailings: reminder cards at 1 week and questionnaire copies at weeks 3 and 7. In the Massachusetts surveys, mailed questionnaires were followed up with 3 facsimile transmissions to the physicians' offices at weeks 1, 3, and 7 requesting completion. Afterward, a telephone call was placed to each physician to personally request completion.
Survey results were processed by a database program and analyzed using a commercially available software package (SAS Institute Inc, Cary, NC). Tests of significance were performed using the χ2 test with P<.05. Comparisons were made among the 4 surveys and to the Diabetes 2000 program recommendations initiated in 1990, as found in the AAO's Preferred Practice Pattern Guideline: Diabetic Retinopathy.7
Results are presented in Table 1, Table 2, Table 3, Table 4, and Table 5. The absence of data from a particular state in these tables indicates that such information was not available to the authors. Column percentages may not always total 100% for certain tables because respondents could choose more than one response per question.
A comparison of the survey statistics is presented in Table 1. Almost all ophthalmologists in Florida and Massachusetts reported asking patients about a history of diabetes. Most respondents in Massachusetts were aware of the Diabetes 2000 project, with almost half interested in receiving more information about this program.
Table 2 (question 1) concerns the start of dilated retinal examinations in patients with type 1 diabetes mellitus, defined as those diagnosed as having diabetes before the age of 12 years. The percentage requesting examination on diagnosis of diabetes decreased from the Florida survey to the first Massachusetts survey (MA1) and still further to the second Massachusetts (MA2) survey. The percentage taking duration of diabetes into account increased notably from the Florida to MA2 surveys. Five years with diabetes was the most frequent response from those choosing duration as a criterion in both the MA1 and MA2 surveys, increasing from 16% to 67% of respondents.
Table 3 (question 4) compares care for postpubertal children with diabetes compared with adults. Approximately one third of physicians responded that both groups should be treated the same. Another one third of respondents reported relying on duration of disease. The remaining third responded that adults and children should receive different recommendations, with fewer visits for children than for adults over time.
Table 4 (question 2) presents data for follow-up of retinopathy-free patients of various ages and duration of diabetes. Respondents recommended longer follow-up intervals with each successive survey. There were fewer 6-month and more 2-year follow-ups in the later surveys, with a 1-year follow-up interval becoming the most popular choice. To determine the influence of duration of diabetes in recommending a follow-up interval, comparisons were made between recommendations for patients with long-standing diabetes (3 and 4) and those with shorter durations. In all 4 surveys, patients 3 and 4 had more reexaminations at 6 months and fewer at 2 years than the other patients. Over time, the 4 surveys demonstrated fewer 6-month and more 1-year examinations for this patient pair. Despite a great variation in age (25  vs 60  years old), these patients received almost identical recommendations for follow-up in every interval and every survey.
The 2 patients with the shortest duration of diabetes (2 and 6) had fewer recommendations for reevaluation at 6 months and more recommendations for reevaluation at 2 years in all 4 surveys than most patients with diabetes of longer-duration. Over time, there was an increasing tendency to see patients 2 and 6 at 1- and 2-year intervals, with fewer 6-month interval recommendations. General ophthalmologists and retinal specialists both followed this trend (Table 4), with some differences noted in the Florida and Massachusetts surveys. In these surveys, there was a greater difference between general ophthalmologists and retina specialists for 2-year than for 6-month follow-ups in patients with diabetes of short duration, which increased with successive surveys. For the 1-year follow-up interval, the difference between general ophthalmologists and retina specialists increased in successive surveys. These differences were not seen for patients 4 and 5—who had long duration, or patients 3 and 4—who had their age controlled for. General ophthalmologists and retina specialists treated patients 4 and 5 similarly during the 4 years, except MA2, in which no retina specialist gave patients 4 and 5 2-year return intervals.
By comparing patients 3 and 4, who had similar ages but different durations of disease, we were able to control for age and determine that longer duration was associated with shorter follow-up interval in all 4 surveys. The effect of age was studied by comparing patients 1 and 3 and controlling for duration. The younger patient had more 6-month and fewer 1-year follow-up recommendations than the older patient. General ophthalmologists and retina specialists (Table 4) had similar profiles for these patients.
Table 5 (question 3) presents treatment or follow-up recommendations for types of diabetic retinopathy without regard to patient age or duration of diabetes. For background diabetic retinopathy, there was a preponderance of follow-up visits between 6 months and 1 year. Over time, there was a definite trend toward more 1-year and fewer 6- to 11-month evaluations so that, by MA2, no one recommended a follow-up at less than 4 months. Table 5 also compares responses from general ophthalmologists and retina specialists in Massachusetts. For background diabetic retinopathy, retina specialists recommended more follow-ups at 4 to 5 months and less follow-up at 6 months to 1 year than general ophthalmologists. No retina specialist elected immediate treatment for this group.
There was a trend toward earlier follow-up visits for patients with preproliferative retinopathy, with the highest percentage of physicians requesting a 1- to 3-month interval visit, and few (almost none) requesting a 1-year follow-up (Table 5). Between 13% and 22% of respondents suggested a 6- to 11-month follow-up. Eight percent of Florida physicians and 5% to 9% of Massachusetts physicians suggested immediate treatment, but 21% of Massachusetts physicians suggested referral of patients with preproliferative retinopathy to a retina specialist. Six percent to 9% of retina specialists in Massachusetts suggested immediate treatment, but they were more likely to follow up at earlier intervals of 1 to 5 months than at 6 to 11 months.
Follow-up of both proliferative retinopathy and macular edema showed changes over time toward earlier treatment (Table 5). While Florida ophthalmologists were much more willing to follow up these conditions for various intervals, Massachusetts ophthalmologists who were surveyed later in time either treated these conditions immediately or referred the patient, with approximately 10% willing to wait up to 3 months to follow up these conditions. Retina specialists were much more likely to treat proliferative retinopathy, macular edema, and vitreous hemorrhage at the time of diagnosis than were general ophthalmologists, and were also less likely to wait 1 to 3 months for follow-up. Sixty-five percent to 74% of retina specialists elected to immediately treat proliferative retinopathy or macular edema.
For patients with maculopathy without edema, approximately one fourth were referred to a retina specialist. A greater percentage of retina specialists followed up these patients at shorter intervals (≤5 months) than general ophthalmologists. Less than 10% of retina specialists from Massachusetts advised immediate treatment; this percentage decreased from MA1 to MA2, whereas there was no change among the 3% of general ophthalmologists who advised immediate treatment.
Although these data suggested important trends in clinical practice, χ2 testing of the differences between general ophthalmologists and retina specialists was not significant at the P<.05 level. Practice size (determined as the number of patients seen per week) (Table 1) in MA1 and MA2 had no significant association with follow-up intervals recommended for patients with or without diabetic retinopathy (Table 4 and Table 5; P<.05).
Diabetic retinopathy is the leading cause of new blindness among working-age Americans.8 In an effort to educate ophthalmologists about the optimal care of patients with diabetes, especially as suggested by the results of multicenter national studies,1- 4 the AAO developed Preferred Practice Patterns (PPP) for diabetic retinopathy and established Diabetes 2000. We investigated the manner in which ophthalmologists manage patients with diabetes via a self-administered survey conducted 4 times in 3 states for 11 years. In contrast to a recent study9 that assessed the conformance of one multispecialty group practice to PPP, this study reports on a cross section of ophthalmologists.
We acknowledge that differences among the survey populations, including access to educational programs, age of practitioners polled, and quality of programs offered may weaken comparisons and conclusions. Since the data derive from a self-administered questionnaire, respondents' answers may not fully mirror actual practice patterns. Nevertheless, beyond a detailed, exhaustive chart review or prospective study, the use of a questionnaire offers a practical means of assessing practice data.
The importance of duration of diabetes as a determinant in the progression of retinopathy is stressed in the PPP. The surveys dealt extensively with duration of disease and patient age as determinants of initial and follow-up examinations. Over time, duration proved to be a more important criterion in determining follow-up for adults, but its use remained ambiguous for pediatric patients. In adults, duration was independent of patient age for patients with both type 1 and type 2 diabetes mellitus. A trend to mirror the PPP recommendation of 1-year follow-ups was seen for both general ophthalmologists and retina specialists. This 11-year trend toward follow-ups at 1 year strongly suggests that results of national studies and Diabetes 2000 have succeeded in modulating practice patterns along suggested guidelines for adults. Moreover, the smaller trend toward an increase in 2-year follow-ups for adults free of retinopathy suggests a growing understanding of the natural history of the disease.10
In contrast to trends toward AAO recommendations in adults, practitioners exhibited a variable and inconsistent approach to the management of pediatric patients. The data for time of initial examination and frequency of follow-up for pediatric patients free of retinopathy suggest confusion regarding the importance of duration within the natural history of this group.11 Specifically, the surveys demonstrate that a large percentage of ophthalmologists fail to appreciate that a well-defined time of diagnosis of diabetes in the prepubertal patient obviates the need for both an initial evaluation at that time and frequent follow-ups. Although duration of diabetes and a 5-year waiting period were increasingly regarded in successive surveys to determine the time of initial examination, most responders in all surveys reported an initial examination of pediatric patients at the time of diagnosis of diabetes. Over time, the percentage of examinations at onset of diabetes decreased, whereas those based on duration increased, as recommended by Diabetes 2000. It is worth noting, however, that primary care referral patterns may influence the time of initial examination, suggesting that Diabetes 2000 needs to emphasize the natural history of diabetic retinopathy to both primary care physicians and ophthalmologists.
Follow-up of prepubertal diabetic patients was also a source of confusion. Most practitioners recommended close follow-up for short-duration, prepubertal patients who are free of retinopathy. Despite the PPP-recommended interval of 1 year for these prepubertal patients, over time, 1-year follow-up visits remained stable and 2-year visits increased. Moreover, in pubertal and immediate postpubertal patients, only one third of practitioners used duration to guide follow-up. Practitioners may be uncertain as to the management of pediatric diabetics, especially for patients who are peripubertal. Although the PPP offers guidelines for examination of patients up to 30 years old, specific recommendations regarding the management of peripubertal patients with diabetes are lacking. Inasmuch as puberty can have a destabilizing effect on the diabetic retinal vasculature—especially in a poorly controlled patient—the preferred practice guidelines could be further refined to address this point.
General ophthalmologists appeared to be following the recommendation that a patient newly diagnosed as having retinopathy-free type 2 diabetes mellitus be followed up at longer intervals,12 inasmuch as they reported an increasing trend for 2-year follow-ups over time. In contrast, retina specialists reported more frequent 1-year and less frequent 2-year follow-ups over time for this same patient. Perhaps this differing trend between general ophthalmologists and retina specialists reflects the fact that the latter are following up patients who may be medically more complicated, with conditions such as arteriosclerosis, hypertension, nephropathy, and poor metabolic control. General ophthalmologists and retina specialists were similar in their recommendations for patients with diabetes of long duration, again supporting the conclusion that there has been excellent education in the role of duration in the management of diabetics.
For patients with short-duration, retinopathy-free type 1 diabetes mellitus, retina specialists suggested more 2-year and fewer 1-year follow-up examinations over time than did general ophthalmologists. The PPP recommend longer follow-up (5 years after onset of diabetes). Although general ophthalmologists and retina specialists noted more 2-year follow-up examinations, the increase was greater for retina specialists. The apparent caution exhibited by general ophthalmologists may also reflect the confusion, if not anxiety, with regard to the care of peripubertal patients, as discussed herein.
While most physicians chose to follow up background retinopathy (minimal or nonproliferative diabetic retinopathy) at less than 1 year, an increasing percentage of physicians chose to follow up this condition at 1 year. Retina specialists followed up these patients sooner than general ophthalmologists, perhaps because patients followed up by retina specialists may be at greater risk of progression. Treatment for this condition was advised by a small percentage of general ophthalmologists in even the later surveys, but not by any retina specialists.
Preproliferative retinopathy (severe nonproliferative retinopathy) was followed up at slightly shorter intervals than the PPP guideline of 3 to 4 months. Although the responses of most ophthalmologists fell within the PPP, as many as 22% of respondents were willing to wait 6 to 11 months for follow-up. These responses may suggest that the unstable nature of this condition may not be adequately appreciated by many ophthalmologists. On the other hand, 20% of general ophthalmologists recognized an imminent progression of disease and referred their patients with preproliferative retinopathy to a retina specialist. The trend among retina specialists to see these patients at shorter intervals than general ophthalmologists suggests a heightened concern for this subset.
The management of proliferative retinopathy (high-risk proliferative diabetic retinopathy) and macular edema were well within the PPP guidelines for treatment at the time of diagnosis and short follow-up interval. Over time, there was an increase in the number of general ophthalmologists who either treated or referred these patients. Interestingly, retina specialists urgently treated only three fourths of their patients with proliferative disease or macular edema, suggesting that such physicians withheld laser photocoagulation treatment until high-risk criteria were noted as prescribed in the PPP.
The surveys suggested that patients with maculopathy but no edema are either referred or followed up even closer than suggested by the PPP (4-6 months). A small percentage of general ophthalmologists and a larger percentage of retina specialists recommended treatment, perhaps to minimize the progression of maculopathy with foveal deposition of hard exudate.
The finding that the number of patients seen per week was not associated with follow-up recommendations suggests that clinical decision making is independent of time constraints.
Our review of clinicians' practice patterns over time suggests that practitioners are becoming increasingly aware of Diabetes 2000 and adopting the guidelines outlined within the PPP. This indicates that ophthalmologists find these guidelines useful. Although our surveys do not show statistical significance by χ2 testing, they do demonstrate definite trends in the care of patients with diabetes. Monitoring clinical practice patterns through surveys such as this enables educators to address misconceptions and modulate existing programs or develop new ones. Guidelines developed in controlled clinical trials may be scientifically sound in an ideal setting but impractical in the average clinical practice, where patients may be too easily lost to follow-up and where referral sources impose unreasonable demands. Surveys such as this are a means to assess not only the adaptation of guidelines, but also their effectiveness and practicality. We are hopeful that our continuing analysis of the questions contained in this survey, including those concerned with diagnosis and treatment, will provide additional insights.
Accepted for publication December 29, 1998.
Presented at the American Academy of Ophthalmology's Centennial Annual Meeting, Chicago, Ill, October 28, 1996.
Reprints: John J. Khadem, MD, MPH, Retina Specialists of Boston, 100 Charles River Plaza, Fourth Floor, Boston, MA 02114 (e-mail: firstname.lastname@example.org).