Bradford W. Lee, Parthasarathi Sathyan, Rajesh K. John, Kuldev Singh, Alan L. Robin. Predictors of and Barriers Associated With Poor Follow-up in Patients With Glaucoma in South India. Arch Ophthalmol. 2008;126(10):1448–1454. doi:10.1001/archopht.126.10.1448
To determine predictors of and reasons for poor longitudinal glaucoma follow-up in South India.
This 1-to-1, matched, case-control study enrolled 300 patients with established glaucoma. We defined cases (poor follow-up) and controls (good follow-up) based on number of and maximum interval between glaucoma follow-up visits attended in the preceding year. We collected data by oral questionnaire and used stepwise multivariate logistic regression to calculate odds ratios (ORs) for poor follow-up.
Adjusting for age and sex, independent predictors of poor follow-up included lack of formal education (adjusted OR, 4.13; 95% confidence interval [CI], 1.44-11.90), no use of prescribed glaucoma medications (adjusted OR, 2.17; 95% CI, 1.06-4.43), and belief that follow-up is less important if one uses glaucoma medications and has no noticeable visual changes (adjusted OR, 10.59; 95% CI, 3.74-29.97). Age, sex, and disease severity were not significant predictors. The most prevalent barriers to follow-up were belief that there was no problem with one's eyes (44.4%) and lack of escort (19.7%).
Knowing predictors of poor follow-up can help identify patients who need individualized strategies to improve follow-up. Because believing one's eyes are problem-free and lacking escorts are significant barriers to follow-up, novel strategies in patient education (eg, intensive counseling, audiovisual aides, and patient support groups) and escort provision may improve longitudinal glaucoma follow-up and disease management.
Glaucoma is the third leading cause of blindness worldwide after cataract and uncorrected refractive error and accounts for 10% of the world's blind.1India is no exception, with 4% to 10% of bilateral blindness attributable to glaucoma.2- 4Multicenter studies have found that lowering intraocular pressure can significantly reduce the development or progression of visual loss.5- 9As such, even marginal improvements in glaucoma care in India would have a vast effect in decreasing vision loss and blindness.
Improvements in glaucoma care, both in developed and developing countries, have been limited by inadequate screening and diagnosis, low use of eye care services, and poor adherence to treatment and follow-up recommendations. First, there exists a low level of glaucoma diagnosis. In India, studies have found that 93% to 94% of persons with open-angle glaucoma had not been diagnosed until the time of the survey, of which 1.5% and 3.3% were already blind bilaterally and unilaterally owing to glaucoma.4,10Even in developed nations, approximately one-half of persons with glaucoma are unaware of their disease.11
In already diagnosed individuals, an additional challenge may be a low number of patients who actually seek care. Community-based studies in rural India such as the Aravind Comprehensive Eye Survey have found that, of the people diagnosed with glaucoma in the community, 50% had never had an eye examination and only 16% had ever visited an ophthalmologist.10Other studies have identified reasons why people with vision problems do not seek eye care, such as lack of funds, time constraints, inability to leave family and work responsibilities, need for escorts, fear, apathy, or fatalistic perceptions about their disease.10,12,13
Finally, once diagnosed individuals seek care as patients with glaucoma, poor medication adherence and poor attendance of periodic glaucoma follow-up visits serve as obstacles to proper disease management. Many studies have documented poor adherence to glaucoma medication regimens,14- 18with one study finding that nearly half of individuals filling glaucoma prescriptions discontinued ocular hypotensive therapies within 6 months.19However, no studies, to our knowledge, have examined the problem of poor longitudinal glaucoma follow-up in a developing country.
We performed a case-control study to determine independent predictors of poor glaucoma follow-up in India, with particular interest in whether severity of glaucoma, transportation-related access to clinic, and financial ability were predictive of poor follow-up. We then evaluated patient-reported reasons for failure to attend follow-up visits.
This 1-to-1, matched, case-control study enrolled 300 existing glaucoma patients attending follow-up at the Aravind Eye Hospital Glaucoma Clinic in Coimbatore, Tamil Nadu, India from July to September 2006. The hospital is a multispecialty tertiary care ophthalmology hospital and part of a 5-hospital system responsible for 1 825 826 outpatient visits and 261 772 ocular surgical procedures and laser procedures performed in 2007. At Aravind, patients with glaucoma are scheduled to attend routine follow-up visits at regular intervals based on disease severity (Table 1), with more severe glaucoma requiring follow-up at shorter time intervals. We enrolled 150 cases (subjects with poor follow-up) and 150 controls (subjects with good follow-up). Evaluation of follow-up status was based on disease severity as well as data on all glaucoma follow-up visits attended. For each subject, all glaucoma follow-up visits attended at Aravind in the previous 12 to 14 months were documented directly from Aravind's medical records. Patients self-reported whether they had attended any glaucoma follow-up visits at outside facilities and, if so, the number and dates of all visits.
Because patients may attend multiple visits in short succession but then go many months without proper follow-up, patients deemed to have good follow-up had to have attended follow-up visits with a certain maximum time interval between visits as advised by their ophthalmologist. Owing to unpredictable follow-up patterns and the need for clinical judgment in evaluating good vs poor follow-up, 3 members of the research team independently confirmed each patient's follow-up status and discussed any initial discrepancies in classification until a consensus was reached. The chief of the glaucoma service evaluated disease severity (ie, mild, moderate, or severe) according to the American Academy of Ophthalmology preferred practice patterns guidelines for primary open-angle glaucoma20and based on visual field testing, cup-to-disk ratios, and an ophthalmology examination. Patients who did not attend follow-up within 1 month after their scheduled follow-up date were considered to have failed to attend a follow-up visit.
All subjects were patients with glaucoma treated at Aravind for at least 1 year prior to enrollment. Eligibility criteria included (1) a medical record with documentation of glaucoma follow-up visits attended in the past year, (2) age of 40 years or older, and (3) a diagnosis of primary open-angle glaucoma, primary angle-closure glaucoma, normal-tension glaucoma, or exfoliation syndrome with glaucoma. After obtaining oral informed consent, all questionnaires were administered orally by a single trained social worker in the patient's preferred language (Tamil or Malayalam). We recruited all eligible patients who came to the glaucoma clinic while the social worker was available to conduct oral questionnaires. Subjects who had failed to attend at least 1 prior follow-up visit in the past year were asked about barriers and reasons for nonattendance.
Prior to the formal study, a pilot study was conducted to validate the questionnaire, to assist in coding classifications, and to test protocol feasibility and acceptability. Based on pilot study results, we estimated that 300 patients would be adequate to identify predictors of poor follow-up with an odds ratio (OR) of 2 or greater, with 80% power and an α of .05. The study protocol was approved by the institutional review boards of Stanford University, Stanford, California and Aravind Eye Care System, Madurai, India. The study followed the tenets of the Treaty of Helsinki.
Data entry was performed using a customized SQL program (Aravind Eye Hospital, Madurai, India) to ensure there were no missing data and to check for inconsistencies in responses. Statistical analysis was done using a commercially available software package (SPSS 15.0.1; SPSS, Chicago, Illinois). Adjusted ORs were calculated using a stepwise multivariate logistic regression model. In creating the multivariate logistic regression model, variables with Pvalues of less than .20 were initially included, along with age and sex, and then successively eliminated based on Pvalues. Variables retained in the final multivariate logistic regression model needed to have a Pvalue less than .05, although age and sex were retained in the final model despite having higher Pvalues and lacking statistical significance.
Of all subjects recruited for the study, only 5 declined to participate owing to time constraints. In total, 300 subjects completed the questionnaire, including 150 subjects with poor follow-up (cases) and 150 subjects with good follow-up (controls) according to our 1-to-1, matched, case-control protocol. Sixteen subjects (10.7%) in each of the case and control groups had seen an ophthalmologist for a glaucoma follow-up visit outside of Aravind in the 12 to 14 months prior to the study. Potential demographic predictors of poor follow-up and their unadjusted ORs are shown in Table 2. The mean subject age was 64 years and more than 60% of subjects were men. Cases were more likely to have had no schooling than controls.
Other potential predictors of poor follow-up relating to medication use, transportation, finances, knowledge about glaucoma, and perceptions about the importance of glaucoma follow-up care are shown in Table 3. More than 70% of subjects in both groups reported using their prescribed topical glaucoma medications, although cases were more likely to report not using their prescribed eye drops and not being advised to use eye drops. Cases were also more likely to receive means-tested waiving of clinic fees. Regarding knowledge about glaucoma's effects on vision, cases were noticeably more poorly informed. Cases were less likely than controls to remember being formally counseled on glaucoma, to understand glaucoma's potential to cause “severe or total vision loss,” or to recognize the permanency of glaucoma-induced vision loss. Cases were also more likely to consider attending follow-up visits to be only “somewhat or not too important,” both in general and in the setting of medication use with no noticeable vision changes.
A stepwise multivariate logistic regression model identified factors that were independent predictors of poor follow-up, including having no schooling, not using prescribed eye drops, and feeling that it is only “somewhat important” or “not too important” to attend follow-up visits in the setting of medication use and no noticeable vision changes (Table 4). Interestingly, one factor that predicted better follow-up was undergoing 3 or more vehicle changes to travel to the clinic. Unlike in the univariate analysis, neither nonpaying patient status (Aravind categorizes patients as paying or nonpaying patients, with the fees of nonpaying patients waived on an honor system; the hospital uses revenues from paying patients to cross-subsidize care for nonpaying patients), lack of recollection of being counseled on glaucoma, or uncertain knowledge about the effects of glaucoma were independent predictors of poor follow-up.
Of the 300 subjects, 223 (74.3%) failed to attend at least 1 follow-up visit and collectively cited 402 barriers for an average of 1.8 barriers per subject. Barriers were classified in 5 categories, with the following percentages of total barriers reported: knowledge and perceptional barriers (37.3%), incidental obligations (17.7%), time and inconvenience (17.4%), physical challenges (17.7%), and financial difficulty with follow-up visit–related costs (10.0%). The barriers and reasons reported for nonattendance are reported along with their categorizations in Table 5. The most prevalent barriers reported were thinking “my eyes were okay” (44.4%), lacking an escort to the clinic (19.7%), being unable to leave work responsibilities (16.1%), being unaware of the importance of regular follow-up visits (13.0%), and being out of town (10.8%). The most important reasons cited for nonattendance were thinking “my eyes were okay” (26.0%), lacking an escort to clinic (15.2%), being unable to leave work responsibilities (7.6%), being out of town (7.6%), and being unaware of the importance of regular follow-up visits (6.3%).
This is the first study to examine the predictors of and barriers associated with poor glaucoma follow-up in patients with established glaucoma in a developing country. Our primary hypothesis, that poor attendance of follow-up visits is associated with milder disease severity, inconvenient transportation, and financial difficulties, was based on clinical experience as well as previous studies indicating that treatment costs, transportation issues, and the perception that vision problems were unimportant were reasons why visually impaired community members did not seek eye care.12,21Our study results suggest that none of the hypothesized variables were independently associated with poor follow-up in patients with established glaucoma, based on multivariate logistic regression analysis. However, because our subjects were all established patients with continuing glaucoma follow-up, we cannot rule out the possibility that our hypothesized predictors of poor follow-up may actually be significant issues for those persons with diagnosed glaucoma who never access glaucoma care or are lost to follow-up. We also did not gather data regarding the length of time since glaucoma diagnosis or the length of time subjects had been followed up for glaucoma at Aravind, although it would be interesting to know whether these factors could help predict follow-up compliance.
One factor that independently predicted poor follow-up was having no schooling. This factor was correlated with nonpaying patient status and poor knowledge about glaucoma's effects (lower levels of education were correlated with free payment status [Pearson correlation, 0.309; P = .01, 2-tailed] as well as poor knowledge of glaucoma's effects [Pearson correlation, 0.309 and 0.352 for knowledge of the “worst potential severity of glaucoma-induced vision loss” and “permanency of vision loss”; P = .01 for both, 2-tailed]), which explains why the 2 latter factors were associated with poor follow-up on univariate, but not multivariate, analysis. The fact that patients with no schooling are unable to read printed materials on glaucoma could further explain their poor understanding of the disease and its management. For patients with minimal formal education who are nonpaying or who, on being questioned, lack fundamental understanding of glaucoma's effects, providers might consider providing educational interventions. These might include specialized patient education sessions that end with counselors asking patients (and their escorts) questions that demonstrate thorough comprehension of glaucoma's effects. Another might be better use of audiovisual teaching aides, such as educational video screenings or glaucomatous vision loss simulators.
Complete nonuse of prescribed glaucoma medications independently predicted poor follow-up, although we found no such association for medication use with poor regimen adherence. Despite the stigma attached to nonuse of prescribed medications and its inevitable underreporting, nearly 10% of subjects with poor follow-up admitted to not using prescribed medications. If other more sensitive methods of detecting nonuse of medications (eg, using written questionnaires or prescription refill information) could have been used (written questionnaires and prescription refill information may be relevant in other countries but are not relevant means of assessing medication compliance in India given the prevalence of illiteracy and the process by which patients purchase medications), an even greater proportion of patients not using medications, and thus presumably at greater risk of poor follow-up, would likely have been identified.
The perception that attending follow-up visits is less important was not an independent predictor of poor follow-up, although that same perception in the setting of medication use and no noticeable vision changes was a strong independent predictor of poor follow-up. This underscores an important misconception in patients that, while it is very important to attend follow-up visits in general, it is less important as long as one uses prescribed medications and does not notice visual changes. This implies that convincing patients of the importance of proper follow-up is not enough. Instead, patient counseling and education must focus on persuading patients why it is important to attend glaucoma follow-up visits, even in the context of medication use and no noticeable vision changes. Patients must understand that progressive visual loss may not be noticed early on, that strict adherence to an ineffective medication regimen can allow disease progression, and that consistent follow-up and monitoring of medication efficacy is ultimately required to forestall glaucoma-induced vision loss. For patients with poor follow-up for whom traditional education and counseling have been unsuccessful, providers may consider implementing novel methods of persuasion, such as creating videos with patient testimonials, organizing patient support groups, or matching patients up with a glaucoma-patient buddy who can help motivate them to manage their glaucoma.
Nearly 75% of subjects (n = 223) failed to attend at least 1 follow-up visit in the past year, a surprisingly high percentage that suggests there is much room for improvement in scheduling and reminding patients of their follow-up appointments. Moreover, because our study recruited subjects on arrival to clinic, it does not include those patients who were entirely lost to follow-up. We designed the study primarily to investigate predictors of follow-up, so our data on barriers to follow-up represent barriers only for those patients who ultimately resumed glaucoma follow-up. Knowledge and perceptional barriers to follow-up were the most frequently cited class of barriers reported. Forty-four percent of subjects who failed to attend follow-up visits reported “My eyes were okay” as a reason for nonattendance, of which 26% stated it was their primary reason for nonattendance. The problem of asymptomatic patients prematurely discontinuing their treatment and follow-up is as much a problem for glaucoma as it is for other diseases, like breast cancer, in which recent studies have shown surprisingly high rates of early discontinuation of tamoxifen.22Another 13% also confessed to being “unaware of the importance of follow-up visits,” of which 6% stated it was their primary reason for nonattendance. This indicates that we must carefully examine how doctors and patient counselors are communicating with and educating patients about their disease, and wherein lies the disconnect. Lacking an escort poses another significant challenge and was the primary reason for nonattendance for 15% of respondents. For practical or cultural reasons, many patients are unable or unwilling to attend follow-up visits without an escort. Various strategies could minimize the time and inconvenience to escorts and to patients themselves. Shortening waiting times in the clinic would decrease escorts' burden in time spent and wages lost. More innovative strategies might include employing a hospital-run escort system, organizing a community-based escort-pool by which a single escort can accompany multiple patients from their community, creating a mobile glaucoma van, or establishing local vision stations where patients can more conveniently follow up periodically.
Based on these study findings, it is clear that clinical intuition is not enough to understand the predictors and barriers related to poor glaucoma follow-up. While we can devise plausible explanatory models for why patients fail to attend follow-up, these models can be biased by false assumptions about patients' understandings and beliefs about their disease as well as their barriers to care.
For patients with a history of poor follow-up and those with known predictors of poor follow-up, physicians may need to take a more patient-centered approach to care that considers individual patients' barriers to treatment and follow-up. Additionally, novel strategies to improve patient understanding about glaucoma and motivate adherence to follow-up and medication regimens should be considered when traditional practices remain unsuccessful. Future studies should focus on which types of strategies are effective in improving follow-up patterns, whether predictors can accurately and prospectively identify poor follow-up patients, and how predictors vary across different developing and developed countries.
We restricted our study population to patients with established glaucoma presenting at the glaucoma clinic at the Aravind Eye Hospital in Coimbatore. However, given its broad catchment area of numerous districts and even neighboring states, we believe that our results can be generalized to much of South India. Although our patients had access to free or highly subsidized glaucoma care based on financial need, only 11% of our study population availed themselves of this program. Generalizing our findings to other settings and other countries must be done with caution given site-specific differences in demographics, patient counseling, and availability and quality of care. We believe, however, that our research method can be adapted and applied to determine predictors and barriers associated with poor follow-up in patients with glaucoma or other chronic diseases in other geographic regions. Doing so would enable health care providers to develop patient-centered service models that improve longitudinal follow-up and lead to improved disease management and outcomes.
Correspondence:Bradford W. Lee, MSc, Department of Ophthalmology, Stanford University, 804 Los Robles Ave, Palo Alto, CA 94306 (email@example.com).
Submitted for Publication:March 31, 2008; final revision received June 4, 2008; accepted June 23, 2008.
Financial Disclosure:None reported.
Funding/Support:This study was supported by the Stanford University Medical School Traveling Scholars Research Program and the Aravind Eye Care System.
Role of the Sponsor:The Stanford University Medical School Traveling Scholars Research Program provided a travel stipend for on-site research in India and the Aravind Eye Care System provided room and board during on-site data collection.
Additional Contributions:The authors would like to thank the other members of the research team, Thulasiraj D. Ravilla, MBA, Ramasamy Krishnadas, MD, Rengaraj Venkatesh, MD, Rengappa Ramakrishnan, MD, Ganesh V. Raman, MD, Noela M. Prasad, DO, MSc, and Kamaleshpattu Mahalakshmi, BA at Aravind Eye Care System and Rita A. Popat, PhD, at Stanford University.