Use of different classes of medications to lower intraocular pressure. Of our subjects, 169 (73%) of 230 were using at least 1 eye drop. Subjects using a combination of dorzolamide hydrochloride and timolol maleate were placed in both the β-blocker and topical carbonic anhydrase inhibitor (CAI) columns.
Distribution of symptoms of subjects using eye drops.
Subject-reported consideration of adverse effects of eye drops and general medications.
Henry D. Jampel, Gail F. Schwartz, Alan L. Robin, Donald A. Abrams, Elizabeth Johnson, Rhonda B. Miller. Patient Preferences for Eye Drop CharacteristicsA Willingness-to-Pay Analysis. Arch Ophthalmol. 2003;121(4):540–546. doi:10.1001/archopht.121.4.540
To determine the importance that patients place on the characteristics of topical therapy for lowering intraocular pressure.
We administered a willingness-to-pay instrument to 230 patients from 4 glaucoma subspecialty practices, asking them how much they would be willing to pay to obtain particular characteristics in an eye drop. Data about the subjects' demographics, economic status, attitudes toward eye drops and systemic medications, and symptoms from eye drops were correlated with their willingness to pay using 2-part models.
Of our subjects, 169 (77%) were using eye drops to lower their intraocular pressure. Fatigue, blurred vision, and tearing were the most commonly reported symptoms. Eye drop medications most valued by the subjects did not produce blurring, drowsiness, or inhibition of sexual performance; 85% were willing to pay more for an eye drop that did not cause blurring, and on average they were willing to pay 40% more. Higher educational levels and income were generally associated with a willingness to pay more for eye drops with desirable attributes.
Main Outcome Measure
Willingness to pay more (in dollars).
Patient preferences for eye drop characteristics can be assessed using a willingness-to-pay instrument. Patients place differing value on various eye drop characteristics. A better understanding of patient preference could lead to better compliance.
THE USE of eye drops to lower intraocular pressure (IOP) is the mainstay of glaucoma treatment. This therapy is profitable for pharmaceutical companies but a major expense for patients and payers. The past several years have seen the development of several new but relatively costly agents for lowering the IOP. Although these drugs have the advantages of less frequent administration and fewer systemic adverse effects, they may cause greater local adverse effects. Choosing between these various agents can be a challenging task for both physicians and patients. Although we do not have precise knowledge of how certain drugs are chosen vs others, eye drop characteristics that may determine their use include potency, cost, ease of use, and adverse effects, both real and perceived.
Inadequate compliance with topical glaucoma medications has been repeatedly documented in the literature.1- 6 Comfort, convenience, and absence of adverse effects may influence compliance with a particular eye drop. Therefore, an eye drop's characteristics may partly determine compliance with a therapeutic regimen and hence its effectiveness.
The recent literature on economic evaluation in health and health care has shown increasing interest in the use of willingness-to-pay (WTP) questionnaires as a measure of health benefits. Willingness to pay is a contingent valuation and involves using a hypothetical survey to directly ask individuals the maximum amount they are willing to pay for the commodity in question.7 Although these types of studies have been performed to assess the relative values of angina treatment, 8 radiographic contrast dye, 9 and the treatment of otitis media, 10 to the best of our knowledge they have not been used for ophthalmic interventions.
We thought that a WTP questionnaire might be an appropriate means to determine what aspects of eye drop therapy are most important to our patients. We therefore performed a pilot study evaluating the relative values of various medications in specific situations.
We recruited subjects from glaucoma practices at the Wilmer Eye Institute at Johns Hopkins University (Baltimore, Md), the Krieger Eye Institute at Sinai Hospital of Baltimore, and 2 private practices between February 2000 and May 2001. All adult patients receiving follow-up for glaucoma or suspicion of glaucoma were eligible. We excluded patients not fluent in English or judged not to be mentally able to complete the study. We explained the study protocol, which had been reviewed and approved by the institutional review board governing each center, to each patient. Each participant gave both written and verbal informed consent to the interviewer. We provided no reimbursement to the subjects for their time. Only the interviewer was able to match a subject with his or her responses.
We abstracted the following information from the subject's medical record: treating physician, birth date, race, sex, home zip code, Snellen visual acuity measurement, IOP measurement, lens status (clear lens, cataract, or lens implant), cup-disc ratio, current ocular medications, and ocular surgical history.
All interviews were conducted by the same experienced interviewer (R.B.M.). The interview lasted approximately 30 minutes. In a face-to-face interview directly preceding the WTP questionnaire, we asked the subjects whether they had systemic hypertension, diabetes mellitus, cardiac or pulmonary problems, or arthritis. They were also asked about family history of glaucoma, formal education, and employment status.
Subjects were asked to rate their vision on a scale of 0 (blindness) to 100 (ideal vision). We used a visual analog scale known as a feeling thermometer.11
In the first part of the interview, we asked the subjects about the number of medications they used for their general health, the number of eye drops they used, and how long they had been using eye drops. We also asked subjects whether they had insurance to cover the cost of their medications and, if so, the amount of their copayment. In the second part of the interview, subjects were asked about ocular symptoms such as blurred vision and tearing and whether they thought that these symptoms were related to their eye drops. They were also asked questions regarding how often they forgot to administer their systemic medications or eye drops, whether the cost of their medications was a burden, and how much they worried about adverse effects of their systemic or ocular medications.
We then asked the subjects a series of questions in the following general format: Imagine that you are paying $50 a month (cash out-of-pocket) for an eye drop that results in a particular symptom (eg, blurred vision for 2 hours). Suppose that the same medication was formulated so that it did not result in this specific symptom. All other things being equal, how much would you be willing to pay for the eye drop that did not produce that symptom? We asked each question a second time with the assumption that the subject had a prescription plan for which there was a $10 copayment per bottle of eye drops, and we asked the subject how much above and beyond the copayment he or she would be willing to pay to avoid this particular symptom. The question about sexual performance was asked only of the last 81 subjects.
Finally, we asked the subjects how much they paid on average for a haircut, how much they spent per person when they went out to dinner, and what their household income was. We used this information plus the median income of households in the subject's ZIP code to estimate the subject's income and wealth.
To check the validity of the WTP instrument for assessing patient preference, at the end the subjects were asked to rank 7 attributes of eye drop therapy, assigning a number 1 to the most important, number 2 to the second most important, and so on. If the interviewer noticed an obvious disparity between the WTP value ascribed by the subject to an eye drop attribute and the ranking assigned to that attribute, this was brought to the subject's attention, and he or she was offered the opportunity to repeat the ranking.
We calculated descriptive statistics to explore both the characteristics of the study subjects and the distribution of responses to the questions. We used the χ2 statistic to compare the frequency of adverse effects between subjects who did and did not use eye drops. We constructed graphs and tables to visually display the data.
Our general approach was to use a 2-part model12 to determine factors associated with a subject's WTP for a particular characteristic of an eye drop. In the first part, we used a logistic regression model to determine whether a subject was willing to pay extra for an eye drop that did not possess an undesirable characteristic. We then restricted our attention to those subjects who were willing to pay more and estimated the average amount extra that they would be willing to pay using a log-normal model. The log-normal model was appropriate because the distribution of the extra cost willing to be paid was positively skewed (ie, a few subjects were willing to pay large extra costs, shifting the tail of the distribution toward more positive values).
Univariate analyses were performed using 2-part models (as described previously) to determine if key demographic variables including age, race, income, education, sex, and the burden of medication cost were associated with a subject's response to questions involving his or her WTP to obtain or avoid eye drop characteristics. These characteristics included frequency of administration, 1 bottle vs 2 separate bottles, lack of blurring or stinging, taste, inhibition of sexual performance, and brand vs generic. Multivariate analyses were adjusted for key demographic variables associated with these characteristics.
Multivariate analyses were performed to determine if various factors were associated with a subject's responses to the WTP questions. The factors related to patient attitude toward medication included frequency of forgetting to take medications and frequency of thinking about adverse effects of medications(categorized as "sometimes, " meaning every day, every other day, once a week, or once a month, and "almost never/never"). Factors related to vision status were also treated as categorical variables, with the following categories:"log MAR better" (≤0, 0-0.2, or ≥0.3), "log MAR worse" (≤0.3, 0.30-0.9, or ≥1.00), and "linear rating" (≤50, 50-75, 76-89, or ≥90). Factors related to medication use were categorized as follows: number of general medications(0, 1-3, and ≥4), number of eye drops (0, 1-2 bottles, or ≥3 bottles), and duration of eye drop use (<5 years, 5-10 years, or >10 years).
We compared the WTP for subjects who reported adverse effects of blurred vision, tearing, stinging, a bad taste in the mouth, and fatigue or drowsiness either none of the time or a little of the time with those who experienced these adverse effects some of the time, most of the time, or all of the time. In addition, for subjects who reported experiencing these adverse effects, we compared the WTP for those who indicated that the eye drops were related to (or caused) these effects with subjects who indicated that the eye drops were not related or who were unsure.
We interviewed 230 subjects. Their mean age was 66 years, 106 (46%) were men, and 75 (33%) were African American (Table 1). Of these, 111 (48%) reported a family history of glaucoma. Many subjects reported systemic hypertension (52%) and arthritis (46%); 82% had at least a high school education, and most (72%) were either retired or not working.
The median income of the study subjects, estimated from the zip codes of their mailing addresses, was $41 300 (Table 2). The mean amount that subjects would pay for a haircut was $15.01, and the mean amount for dinner in a restaurant was $23.15. Most subjects reported that the cost of oral medications and eye drops represented either no financial burden or a small financial burden. Only 39 subjects (17%) stated that they paid the entire cost of their medications.
Subjects' assessment of their vision on a scale of 0 for total blindness to 100 for ideal vision yielded a mean of 72.1 and a median of 80 (Table 3). The median Snellen visual acuity measurements were approximately 20/25 in the better eye and 20/40 in the worse eye. The median cup-disc ratio was 0.70 OD and 0.80 OS.
Of our subjects, 169 (77%) were using eye drops to lower their IOP. The median number of eye drops used by each subject was 2, with a range of 1 to 7. Subjects administering eye drops had used them for a mean duration of 9.4 years, with a range of 0 to 46 years. β-blockers and prostaglandins were the most commonly used eye drops (Figure 1).
The frequency with which subjects reported symptoms possibly related to the use of eye drops was generally low (Figure 2). Fatigue, blurred vision, and tearing were more frequently reported than stinging or bad taste. According to the χ2 test, the frequency of these symptoms did not differ statistically between the 169 subjects using eye drops and the 61 subjects not using eye drops. The proportion of subjects reporting a symptom who attributed that symptom to the use of eye drops was low, ranging from 7 (4%) of 169 for fatigue to 40 (24%) of 169 for stinging.
Of the subjects using eye drops, 60% (101/169) reported that they never considered the potential adverse effects of their eye drops; of the subjects taking oral systemic medications, 39% (69/178) reported that they never considered the potential adverse effects of these medications (Figure 3). Eighty-five percent (144/169) and 87% (155/179), respectively, reported never or almost never forgetting to take their eye drops or systemic medications.
Subjects were asked if they attributed various symptoms to their eye drops (Table 4). The odds ratios(ORs) of attributing blurred vision to eye drop use were highest and statistically significant for subjects using α-agonists (OR = 3.22; 95% confidence interval [CI], 1.20-8.61) and topical carbonic anhydrase inhibitors (CAIs)(OR = 3.00; 95% CI, 1.13-7.95) and lower for the use of β-blockers (OR= 2.39; 95% CI, 0.74-7.77) and prostaglandins (OR = 1.98; 95% CI, 0.74-5.32). Although not reaching statistical significance, the OR of attributing tearing to eye drop use was higher for subjects using topical CAIs (OR = 2.77; 95% CI, 0.99-7.75) and α-agonists (OR = 2.43; 95% CI, 0.77-7.67) than for those using β-blockers (OR = 1.32; 95% CI, 0.45-3.89) and prostaglandins(OR = 0.82; 95% CI, 0.30-2.21). The OR of attributing stinging to eye drop use was statistically significant only for subjects using topical CAIs (OR= 2.73; 95% CI, 1.03-7.29). The OR of attributing bad taste in the mouth to an eye drop was 10.07 (95% CI, 3.12-32.47) for subjects using topical CAIs and not statistically significant for any other eye drop. The OR of attributing fatigue or drowsiness to eye drops was statistically significant only for subjects using α-agonists (OR = 6.63; 95% CI, 1.55-28.36). Because multiple comparisons were made, those found to be statistically significant at P<.05 represent only moderate evidence of an association.
When subjects were told that an eye drop with a potentially unfavorable characteristic cost $50 and asked how much they would be willing to pay for an eye drop without that effect, the responses varied from a mean of $54.13(median, $50) to obtain a brand name as opposed to a generic form of the same drug, to a mean of $70.61 (median, $60) for an eye drop that did not produce blurring (Table 5). Valued almost as much as avoiding blurring were avoiding drowsiness (mean, $68.52), inhibition of sexual performance (mean, $68.28), and a bad taste (mean, $65.76). The percentage of subjects who would pay anything higher than $50 ranged from a low of 26% for switching from generic to brand name to a high of 85% to avoid blurring. The percentage of subjects who were willing to pay anything more than $50 generally correlated with the mean amount extra that subjects were willing to spend, being highest for avoiding blurring (85%) and drowsiness(83%).
We asked 2 related questions concerning preferences for generic vs brand-name medications. In the first scenario, the brand name cost $50, and we asked subjects how much cheaper a generic drug would have to cost to make them switch. The mean price that would induce subjects to switch from the brand name to the generic was $27.33. However, if the question was phrased differently and the subject was already taking a generic medication at $50 and was asked how much more he or she would pay for the brand-name drug, only 26% of subjects were willing to pay more.
When the same questions assumed that the subject had a $10 copayment for medications (instead of $50 out-of-pocket), the percentage of subjects willing to pay more and the amount more that they were willing to pay for eye drop characteristics were similar (Table 6) for the $10 copayment and $50 out-of-pocket scenarios. Subjects were willing to increase their copayment to $23.42 to avoid inhibiting sexual function and to $22.32 and $22.03, respectively, to avoid blurred vision and drowsiness. Subjects were least willing to increase their copayment to reduce dosing from 3 times a day to twice a day ($15.02) and for having 2 medications in 1 bottle as opposed to 2 different bottles ($16.54).
The subject's educational level was moderately associated with being willing to pay more than $50 and, if so, the additional amount that subject was willing to pay for an eye drop used with less frequency, that came in 1 bottle, and that did not cause blurred vision, drowsiness, stinging or tearing, a bad taste, or inhibition of sexual performance (data not shown). The subject's income level was moderately associated with the additional amount that he or she was willing to pay for an eye drop used with less frequency, that came in 1 bottle, and that did not cause blurred vision, stinging or tearing, drowsiness, or a bad taste. Evidence also suggested that the burden of cost for general medications was related to the WTP for an eye drop used only once a day and that did not cause stinging, tearing, or drowsiness. We found no associations between WTP and age, race, sex, or the burden that taking eye drops placed on the subject. Multivariate analyses were adjusted for educational level, income, and age, factors that we believed might be important confounders even though they did not show significant associations.
The visual acuity of the subjects seemed to influence their WTP for brand name as opposed to generic eye drops. Subjects with a visual acuity of 20/20 or better in their better-seeing eye were more likely to pay more for a brand-name eye drop than subjects with visual acuities of 20/25 to 20/40(OR = 3.2; 95% CI, 1.51-6.67) or worse than 20/40 (OR = 2.7; 95% CI, 1.06-6.67). Subjects with a visual acuity of 20/40 or better in their worse-seeing eye were more likely to pay more for a brand-name eye drop than subjects with a visual acuity of 20/40 to 20/200 (OR = 2.8; 95% CI, 1.30-6.25). In general, subjects with better vision valued brand name more than those with poorer vision.
For most comparisons, there was no relationship between the use of a particular class of eye drop and the WTP to avoid certain eye drop characteristics. The only statistically significant relationships were for subjects taking topical prostaglandins or topical CAIs. Subjects taking prostaglandins were less likely (OR = 0.51; 95% CI, 0.30-0.89; P = .02) to pay more for an eye drop bottle containing 2 medications to lower the IOP(as opposed to 2 bottles, each containing a different medication) than those not taking a prostaglandin. Subjects taking a topical CAI were less willing to pay extra for an eye drop used only once daily (OR = 0.51; 95% CI, 0.26-0.97; P = .04) or that did not cause stinging or tearing (OR= 0.51; 95% CI, 0.26-0.98; P = .04).
We explored whether there was a relationship between symptoms that subjects reported and their WTP for an eye drop that did not produce such symptoms(Table 7) and found no relationship in our data. We also examined the subset of subjects who not only reported a symptom but also attributed that symptom to an eye drop. Those subjects who both experienced and attributed their blurred vision to the use of eye drops were more than 3 times as likely to pay more for an eye drop that would not result in blurred vision compared with subjects who did not attribute their blurred vision to the use of eye drops. No other relationships were detected between symptoms attributed to eye drops and WTP for an eye drop that did not produce such symptoms.
Specific subject attributes that did not affect WTP for any of the eye drop characteristics studied included the following: self-reported frequency of forgetting to take either systemic or ocular medications, self-reported frequency of thinking about adverse effects of medications, number of eye drops, duration of eye drop use, and number of medications for general health.
As an alternative method for assessing subject preference for eye drop characteristics and to check the validity of the WTP approach, we compared the likelihood that a subject would pay more than $50 in the WTP analysis with his or her ranking of the importance of the characteristic in relationship to the other characteristics. Four of the 230 subjects reordered their list of attributes when gross disparities between their WTP responses and their ranking were pointed out to them. We compared the odds of being willing to pay extra for a particular attribute of an eye drop for subjects who ranked that attribute as 1 of the 3 most important with those who ranked it as fourth to seventh in importance (Table 8).For 5 of the 7 attributes, excluding blurred vision and stinging or tearing, the ORs strongly suggested a correlation between the 2 ways of expressing preference.
Our WTP analysis suggests that subjects place a higher value on certain attributes of eye drops than others. Subjects valued the fact that an eye drop medication did not produce blurring of vision, drowsiness, or inhibition of sexual performance more than they valued once-a-day use or that 2 eye drop medications could be administered from 1 bottle rather than 2.
Subjects with a higher level of education were more likely to pay more for desirable attributes and, if so, to pay greater amounts. More educated individuals may place greater value on convenience and absence of adverse effects, but it is also possible that these individuals possessed a better understanding of the study.
Subjects with higher incomes, as assessed by median income within a ZIP code area, generally were not willing to pay more; however, those who would pay more were willing to spend greater amounts. One can speculate that the value of each extra dollar was less important to subjects with higher incomes. No other characteristics besides education and income consistently influenced subjects' preferences.
Most subjects did not feel that their symptoms were related to the use of eye drops and were more likely to consider the possibility of adverse effects from oral medications. Perhaps physicians and patients tend to downplay the fact that eye drops for glaucoma are real medications, and patients may not be aware that eye drops can cause adverse effects. There was a tendency for subjects to attribute blurred vision and fatigue to α-agonists, and blurred vision, stinging, and bad taste to topical CAIs. However, for subjects who attributed a symptom to their eye drops, blurred vision was the only symptom they were willing to pay more to avoid.
One limitation of a WTP analysis is that it is entirely hypothetical. Regardless of the subject's response to the instrument, at the conclusion of the interview, the subject has not yet had to spend any of his or her money. What individuals state that they would do and what they would actually do in particular circumstances may be quite different. An alternative study design might involve observing patients' behavior if given such choices in a pharmacy, but such a study would be difficult to perform.
Another limitation of our study is that the answers provided by the subjects can be heavily influenced by the question and the interviewer. For instance, the question about blurred vision asked only about 1 duration of blurred vision, 2 hours, and did not take into account shorter durations of the symptom or how severe it was. Likewise, the question about inhibition of sexual performance was extremely vague, and its specifics were left to the imagination of the subject. A female interviewer asked this question, and subjects may have responded differently if the interviewer had been male. However, it is not clear that more specific or open-ended questions would necessarily have been more informative.
Our experience with addressing the question of patient preference for generic vs brand-name medications highlights the particular difficulties in this area. If the question assumed that the subject was already taking a generic medication, most subjects were highly reluctant to pay anything additional for the similar brand-name medication. On the other hand, if the subject was already using the brand-name medication, the generic medication had to be significantly less expensive to trigger a switch. One can speculate that patients who are tolerating a generic drug assume that a brand-name substitute will not be any more effective or better tolerated, whereas those who are already using a brand name do not want to "rock the boat" for potential savings.
We designed the instrument to explore scenarios in which the subject either paid entirely out-of-pocket for eye drops or had a $10 copayment that could be increased to obtain a more desirable eye drop. There were no substantive differences in the responses to these scenarios. Not surprisingly, the absolute amount subjects would pay extra was greater from a $50 starting point, whereas the percentage increase was greater from the $10 starting point.
There are numerous reports in the literature about inadequate compliance with glaucoma medications.1,2 The information that we obtained concerning patients' reporting of compliance with using their eye drops must be interpreted in the context of earlier studies. We found that 85% of subjects reported never or almost never forgetting to take their eye drops, whereas Kass et al13 found that a much higher proportion of patients fail to consistently use their eye drops when assessed with a compliance monitor. Although the search to develop eye drops that are convenient to use with minimal adverse effects is important, it is justified only if it results in better patient compliance with therapy. Future research should focus on whether we can improve patient compliance by developing eye drops that have more desirable characteristics.
Corresponding author and reprints: Henry D. Jampel, MD, MHS, Maumenee B-117, 600 N Wolfe St, Baltimore, MD 21287-9205 (e-mail: email@example.com).
Submitted for publication April 20, 2002; final revision received November 5, 2002; accepted December 6, 2002.