Medication Adherence in Patients With Glaucoma and Disability

Key Points

Question  What is the level of glaucoma medication adherence for patients with glaucoma and disability?

Findings  In this population-based cohort study that included 46 468 patients with glaucoma, having limb disability, being in a vegetative state, and having dementia were associated with 0.226 to 3 fewer outpatient visits and lower medication adherence (by 0.7-17.6 percentage points).

Meaning  These results suggest that policies targeting glaucoma medication adherence should consider disability types.

Importance  People with disabilities tend to have lower medication adherence. Glaucoma medication adherence has been scantly studied for people with disability.

Objective  To determine whether disability leads to reduced glaucoma medication adherence and whether this decrease varies by type of disability.

Design, Setting, and Participants  This population-based case-control study enrolled individuals with glaucoma and without disability, who were followed up until they received disability certification. All patients in Taiwan with confirmed glaucoma in 1 or both eyes were identified using National Health Insurance claims data. All patients with glaucoma who required glaucoma medication adherence (confirmed glaucoma, suspected glaucoma, and patients with ocular hypertension) and had newly obtained disability status after December 31, 2013, were identified and matched to counterparts without disability based on age and sex. The study period was January 1, 1997, to December 31, 2017. Data were analyzed from May 2021 to August 2021.

Exposures  All patients were followed up until they obtained confirmed disability status, which was identified using the National Disability Registry in Taiwan.

Main Outcomes and Measures  Secondary adherence was measured using frequencies of glaucoma medication refills and outpatient visits at 1-year and 2-year intervals.

Results  A total of 46 468 patients with glaucoma (23 234 with disability and 23 234 without disability; 24 508 men [52.7%]; 21 960 women [47.3%] mean [SD] age, 72.5 [14.3] years) were included in the study. Overall, the frequency of glaucoma outpatient visits was higher in people with disabilities than those without disabilities both before the index dates (difference, 0.64 [95% CI, 0.57-0.72]; P < .001) and after the index dates (difference, 0.34 [95% CI, 0.27-0.41]; P < .001) when using 1-year intervals. However, when stratified by the type of disability, having limb disability, being in a vegetative state, and having dementia were associated with fewer outpatient visits and lower medication adherence (at a maximum of 17.60 [95% CI, 8.90-26.30] percentage points lower; P < .001) compared with people without disability. Adjusted regression results revealed that people with visual disability had a mean of 2.50 (95% CI, 2.34-2.67) times more glaucoma outpatient visits than their matches who were disability free (P < .001).

Conclusions and Relevance  Certain types of disability can reduce glaucoma medication adherence by up to 17.60%. Policies targeting medication adherence should consider these disability types.

Glaucoma is a leading cause of blindness worldwide, and medication adherence is essential to slow or reverse its progression.^1,2 However, nonadherence to glaucoma medications is common.³ As many as 80% of patients deviate from their prescribed medication regimen, depending on the target population studied. Newman-Casey et al⁴ reported that 61% of these individuals cited multiple barriers to adherence, mainly forgetfulness, poor self-efficacy, and beliefs regarding glaucoma medication (skepticism about whether glaucoma medication can mitigate visual acuity loss).

Although several studies have investigated the causes of poor glaucoma medication adherence and strategies to overcome them, few have targeted people with disabilities and, to our knowledge, none has investigated the association of new-onset disabilities with medication adherence in patients with glaucoma and whether such adherence is associated with visual acuity. Disability typically requires special care. People with disabilities and major health conditions often have lower access to medical care, even though they generally incur higher medical expenses than their peers who are healthy.^5-7

A new-onset disability can be emotionally challenging to a patient with glaucoma. One hypothesis is that because disability and glaucoma treatment can both be demanding, patients may be forced to lower adherence to glaucoma medication. The magnitude of such a reduction, if large, warrants policy intervention. For disabilities that preclude a person from being physically or mentally able to adhere to their glaucoma medication, such as dementia and certain limb disabilities, medication adherence depends heavily on caregivers’ perceptions, attitudes, and abilities.^8,9 Having a caregiver make the adherence decision affects the degree of adherence compared with the patient making the decision. Regardless of the underlying mechanism, understanding the association of different disability types with glaucoma medication adherence is essential.

Because life expectancy in most countries has increased over time, a higher proportion of older adults may be living with disabilities. People with disabilities are at an increased risk of secondary health conditions that may not be directly associated with their disability, and quality of life is often adversely affected by disability,^10,11 management of chronic illnesses, and major health conditions.^12-14 Thus, glaucoma medication adherence is crucial for people with disabilities. This study used a national longitudinal database to determine whether and to what extent a new-onset disability in patients with glaucoma affects glaucoma medication adherence compared with patients with glaucoma without disability.

This case-control study included data collected from patients from January 1, 1997, to December 31, 2017. We used the nationwide claims database (for approximately 23 million individuals) of the Taiwan National Health Insurance (NHI) scheme, which contains information on inpatient and outpatient utilization, including all health care utilization, prescriptions, and medical procedures used under the NHI. Enrollment in the NHI program is compulsory for all residents of Taiwan. All individuals are included in the database, even if they do not visit a medical institution while enrolled. The data sets are managed and distributed by the Taiwan Ministry of Health and Welfare. The data were anonymized before being released to the researchers.

This study adheres to the Declaration of Helsinki. It was approved by the Institutional Review Board of National Yang-Ming University.

International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 365 and 364.22 and International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes H40 and H42 were used for identifying glaucoma diagnoses. A patient considered to require glaucoma medication adherence was defined as having 1 or more outpatient visit with a glaucoma diagnosis (including those with confirmed glaucoma, suspected glaucoma, and patients with ocular hypertension) in 1 or both eyes and a glaucoma drug prescription.

The Disability Registry, managed by the Taiwan Ministry of the Interior,¹⁵ is a robust data source to identify people with newly approved disability status. The registry includes all individuals with certified disability status in Taiwan. The Disability Registry can be linked to the NHI claims database by using an individual-specific, anonymous identification number.

To identify people with new-onset disability, we included people with a first record in the Disability Registry after January 2014, because in 2012, all individuals with previously approved disability status were required to renew their registrations. To ensure we included only people with new-onset disability, we excluded all individuals who were present in the Disability Registry before December 2013.

We used the following exclusion criteria. First, individuals who obtained disability status in 2017 were excluded, because 2017 was the final year of the study period, and we required at least 1-year follow-up to calculate medication adherence post-disability certification. We also excluded participants who died within 1 year of obtaining disability status. Second, participants younger than 20 years and older than 100 years in 2017 were excluded to enhance the comparability of the participants. Third, people with congenital disabilities were excluded, because even if they had new-onset disability in adulthood, such disabilities are less likely to represent a health shock.

The outcome variable of interest was secondary glaucoma medication adherence as defined by Newman-Casey and Myers.¹⁶ Secondary nonadherence was defined as patients ceasing glaucoma medication usage after their first prescription.

Prescription information was obtained from the NHI database. Buying glaucoma medication out of pocket is uncommon in Taiwan; thus, the claims database provides robust measures of secondary adherence. Refill records provide more accurate measures than self-reported drug refill data; self-reported refills are often overestimated.¹⁷ To identify refills, we first identified all ophthalmology outpatient visits that led to glaucoma medication prescription. In Taiwan, physicians can provide patients with long-term prescriptions for chronic diseases if the physician judges that the chronic condition is stable and no physician visits are required for the subsequent 3 months. In this case, the patient obtains the first month’s medication on the day of the visit and the subsequent 2 refills at a pharmacy. A maximum of 2 refills per prescription is allowed. However, not all patients return to pharmacies for a refill. In this case, secondary medication adherence is not met.

To ensure that the results are robust to different specifications of medication adherence, we used strict and loose definitions to identify patients who obtained a full year’s medication. Under the strict definition, patients who made at least 3 ophthalmology outpatient visits with glaucoma medication plus 7 refill records were included. Although 4 visits plus 8 refill records would represent a full-year refill, we used 3 visits plus 7 refills instead because we included 1 year before and after the date of confirmed disability; thus, the patients may have made an outpatient visit shortly before disability certification, causing 1 fewer outpatient visit or refill to obtain a full-year refill. For the 2-year measure, 7 outpatient visits plus 15 refills were used. For patients not provided with a long-term prescription, 11 or more outpatient visits were also considered to meet the strict definition. We subtracted 1 visit for the same reason outlined above. For the 2-year measure, 23 outpatient visits were used for this definition.

Under the loose definition, patients who made at least 3 ophthalmology outpatient visits with glaucoma medication, regardless of their pharmacy refill record, were included. This provides the loosest definition of adherence. This definition is likely to overestimate adherence. For the 2-year measure, 7 outpatient visits were used for this definition. For both definitions of adherence, we calculated the number of patients who met the criteria 1 and 2 years before and after disability certification.

Each patient with glaucoma with a new-onset disability was age matched and sex matched to 1 patient with glaucoma without disability. For the patients with new-onset disability, the index date was defined as the date of confirmed disability status, as indicated in the Disability Registry. For patients without disabilities, the index date was that of their matches. We could not match on more variables because of the limited number of people with disability. We therefore controlled for potential confounders in the final regression model instead.

We first used negative binomial regression analysis to determine the association of disability with the number of outpatients in ophthalmology. We subsequently used logistic regression analysis to determine the association of disability with the probability that adherence was met according to the 2 definitions. For both regression models, we controlled for patient’s age, sex, area of residence, the number of inpatient episodes 1 year before the index date (as a proxy for health status), and the patients’ insurable income under the NHI. For all models, we also included an interaction between time (ie, before and after the index date) and disability status (with and without disability). We combined hearing and verbal disabilities (hereafter, hearing/verbal disability) and combined facial, balance, and epilepsy disabilities to form the group of other disabilities. Despite the low number of people with intellectual disability, autism, and vegetative states, we did not combine them with other groups because they have specific characteristics in medication utilization adherence.^18-20 All statistical analyses were conducted using Stata 15 (StataCorp).²¹ We used P < .05 as the threshold of significance for all analyses.

The study included 46 468 patients with glaucoma (23 234 with disability and 23 234 without disability). eTable 1 in the Supplement presents further information on the number of patients with glaucoma with new-onset disabilities between 2014 and 2016. eTable 2 in the Supplement presents the patients’ sociodemographic characteristics. In brief, the full group included 24 508 men (52.7%) and 21 960 women (47.3%), with a mean (SD) age of 72.5 (14.3) years. Because of age and sex matching, these variables did not differ between the 2 groups. People without disability had higher mean monthly income for the first tertile (difference, $41.63 [95% CI, $31.18-46.07]; P < .001) and third tertile (difference, $158.86 [95% CI, $140.41-$177.31]; P < .001) monthly income categories. Moreover, Northern Taiwan had a higher proportion of people without disability than people with disability (53.1% [95% CI, 52.67%-53.57%]; P < .001).

eTable 3 in the Supplement presents the unadjusted number of visits and percentage of participants meeting the strict and loose definitions of glaucoma medication adherence before and after the index date. People with disabilities had a higher number of glaucoma outpatient visits both before and after the index date regardless of whether we used 1-year measures (differences: before the index date, −0.64 [95% CI, −0.72 to −0.57] percentage points; P < .001; after the index date, −0.34 [95% CI, −0.41 to −0.27] percentage points; P < .001) or 2-year measures (differences: before the index date, −1.23 [95% CI, −1.36 to −1.09] percentage points; P < .001; after the index date, −0.54 [95% CI, −0.67 to −0.41] percentage points; P < .001). People with disability also had a higher overall percentage of medication adherence both before and after the index date (example differences: by the strict definition, at 1 year, before the index date, −0.21 [95% CI, −0.21 to −0.20] percentage points; P < .001; after the index date, −0.08 [95% CI, −0.09 to −0.08] percentage points; P < .001; by the loose definition, at 1 year, before the index date, −0.08 [95% CI, −0.09 to −0.07] percentage points; P < .001; after the index date, −0.05 [95% CI, −0.06 to −0.04] percentage points; P < .001).

Tables 1, 2, and 3 present the adjusted marginal association of disability with the number of glaucoma outpatient visits in patients meeting the strict and loose definitions of glaucoma medication adherence. The results revealed that the association of disability with glaucoma outpatient visits varied greatly by disability type. People with visual disability (differences: 1 year, 2.50 [95% CI, 2.34 to 2.67] percentage points; 2 years, 4.34 [95% CI, 4.03 to 4.64] percentage points; both P < .001), verbal disability (differences: 1 year, 0.56 [95% CI, 0.35 to 0.78] percentage points; 2 years, 1.11 [95% CI, 0.73 to 1.49] percentage points; both P < .001), and autism (differences: 1 year, 4.83 [95% CI, 2.11 to 7.54] percentage points; 2 years, 10.33 [95% CI, 5.29 to 15.37] percentage points; both P < .001) exhibited a higher number of visits than their matches without disability for both the 1-year and 2-year measures. However, people who had a limb disability (differences: 2 years, −0.27 [95% CI, −0.53 to 0] percentage points; P = .048), were in a vegetative state (difference: 2 years, −3.00 [95% CI, −5.37 to −0.63] percentage points; P = .01), or had dementia (differences: 1 year, −0.27 [95% CI, −0.44 to −0.10] percentage points; P = .002; 2 years, −0.42 [95% CI, −0.73 to −0.10] percentage points; P = .01) had fewer glaucoma outpatient visits compared with their matches.

In terms of meeting the definitions of medication adherence (Table 2 and Table 3), as logistic regressions were used, we excluded some disability groups because of perfect predictions, mostly those with small sizes. Among the remaining disability types, people with visual and verbal disability consistently exhibited a higher probability of meeting medication adherence, regardless of strict or loose definition or whether the 1-year or 2-year measures were used. For people with visual disability, adherence was 9.80 (95% CI, 8.90-10.60) percentage points higher than for their matches who were disability free when using a strict 1-year definition (P < .001). People with mental disabilities also had higher probability of adherence at 2 years when we used the loose definition (3.80 [95% CI, 0.90 to 6.70] percentage points; P = .01). People with limb disability or dementia, by contrast, had a lower probability of medication adherence compared with their counterparts who were disability free. For people with dementia, the probability of adherence using a strict definition dropped by approximately 1.40 (95% CI, −2.30 to −0.60) percentage points compared with their matches without disability (P < .001). For people in a vegetative state, the decrease in medication adherence was as high as 17.60 (95% CI, −8.90 to −26.30) percentage points (P < .001) when the loose 1-year definition was used. Perfect prediction of failure to meet adherence was observed for this group when using the strict definition of adherence.

In this study, we found that glaucoma medication adherence increased or decreased depending on the type of disability. However, overall, glaucoma outpatient visits and medication adherence were higher among people with disability.

In a 4-year follow-up study, Newman-Casey et al²² reported that an increased number of ophthalmology visits by patients with glaucoma was associated with longer-term glaucoma medication adherence. Given that the number of ophthalmology visits is a modifiable factor, this may be used to improve adherence. How outpatient visits actually affect adherence is unknown; however, increased visits may reflect a higher level of patient health knowledge and physician feedback to the patients, which are essential for glaucoma because it is a relatively asymptomatic disease.²²

We found that people with visual disability had higher glaucoma medication adherence. In Taiwan, people who qualify for a visual disability status are not necessarily completely blind. This may explain why greater effort was made by these patients to ensure the maintenance of their vision. This is in contrast to a Singaporean study, which reported that people with visual impairment had low eye care utilization²³; the discrepancy is likely attributable to different definitions for visual disability and visual impairment in their study. Although lack of eye care utilization can lead to visual impairment,²⁴ our study adds to existing knowledge in 2 ways. First, people with visual impairment may not have lower utilization prior to the impairment than their counterparts who are disability free; however, this higher level of utilization may not necessarily prevent them from developing visual disability. Second, eye care utilization increases after visual disability, leading to a higher burden for both the individual and the economy; however, these patients continue to have suboptimal outcomes.

A Taiwanese study used national survey data and found that older people with poorer scores in activities of daily living, a measure of physical disability, were less likely to access eye care.²⁵ This agrees with the findings that people with limb disability were less likely to refill their glaucoma medication. Policies should have the aim of ensuring that people with such disabilities are able to take their glaucoma medications.

A study indicated that the duration of glaucoma, age, ethnicity, and self-reported attitude are associated with glaucoma medication adherence for the general population.²² However, factors associated with poor glaucoma medication adherence for people without disabilities may not be applicable to people with disability. Strategies to improve medication adherence for glaucoma treatments should target the diverse needs of people with various disabilities. Our previous study indicated that the new onset of cancer can impede glaucoma medication adherence, which is likely caused by demanding treatments associated with cancer.²⁶ For people with disability, however, only certain types of disability reduce adherence; thus, the reasons may go beyond the burden of increased treatment demands. For example, we found that patients in a vegetative state were not likely to receive glaucoma medication. Recovery from a vegetative state is not impossible.^27,28 Given the possibility of recovery and that untreated glaucoma is likely to lead to blindness, health education of the patients’ family members or caregiver on the importance of maintaining glaucoma treatment is essential. More support should be given to caregivers of people with disability to ensure they have enough capacity to perform the tasks necessary for medication adherence. This also applies to patients with dementia because caregivers often play key roles in dementia care.²⁹ Studies on general medical care utilization among people with dementia have provided inconsistent results.^30,31 Higher general medical utilization for people with dementia may be explained by poorer physical health.³⁰ For glaucoma, however, it is unreasonable to conclude that people with dementia have lower need of glaucoma medication. Thus, their lower adherence to glaucoma medication after dementia certification likely reflects barriers to health care utilization.

Other plausible explanations exist for the reduced glaucoma medication adherence in people with disabilities other than increased care burden because of the disability per se. First, patients tend to reduce treatment adherence when they perceive that time to clinical benefit is too long. This has been reported for patients with cancer.³² Second, the disability itself may prevent patients from obtaining glaucoma treatment or a medication refill. This explains why only certain types of disability were associated with lower glaucoma medication adherence. Third, limited economic resources may prevent patients from obtaining care for both disability and glaucoma, especially considering that disability often leads to out-of-pocket expenses.^33,34 Despite glaucoma medication being covered by NHI in Taiwan, other indirect costs (such as caregiver time) may hamper health care utilization.

Because life expectancy has been increasing in most countries, determining whether the additional years of life gained are spent in poor health is crucial.³⁵ The financial burden of disability is high for both the affected individuals and the health system.³⁵ Maintaining a high quality of life is essential even after a new-onset disability, and this includes providing appropriate glaucoma care and maintaining vision.³⁶ Medication nonadherence is often accompanied by several unfavorable outcomes, such as treatment failure, increased complications, disease progression, and higher medical costs.³⁷

Our study has some limitations. First, because of data availability, we could only measure secondary medication adherence, defined as prescription refills. However, refill data cannot clarify whether eye-drop instillation was performed. Second, patients potentially only had acute glaucoma and suspended the treatment soon according to their physician’s instructions. However, this is not likely because long-term prescriptions are unlikely for acute glaucoma. Third, certain types of disability, such as autism, were found in only a small number of patients, even though we used the whole-population data set. Studies in countries with larger populations should be conducted to validate our findings. Third, our data preclude the determination of cause of reduction in medication adherence—for example, whether the reduction was caused by the caregiver’s decision. Fourth, we did not consider severity in our analysis because of data limitations. However, severity normally does not influence a supplier’s decision regarding the number of days the medication is to be provided in Taiwan, because of the comprehensive coverage of the NHI. Countries with less comprehensive health care coverage may need to take severity into consideration.

In this analysis, medication adherence decreased in patients with certain types of disabilities, such as people with limb disability, dementia, and people in vegetative states. Policies targeting increased adherence should consider people with different types of disabilities.

Accepted for Publication: September 13, 2021.

Published Online: October 28, 2021. doi:10.1001/jamaophthalmol.2021.4415

Corresponding Author: Christy Pu, PhD, Institute of Public Health, School of Medicine, National Yang Ming Chiao Tung University, 155 Li-Nong St, Sec 2, Peitou, Taipei 11221, Taiwan (cypu@nycu.edu.tw).

Author Contributions: Dr Pu 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: All authors.

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

Drafting of the manuscript: All authors.

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

Statistical analysis: Pu.

Obtained funding: Pu.

Administrative, technical, or material support: All authors.

Supervision: Pu.

Conflict of Interest Disclosures: None reported.

Funding/Support: The study was supported by the Ministry of Science and Technology, Taiwan (grant 110-2628-B-A49A-502).

Role of the Funder/Sponsor: The funder 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.