July 2006
Volume 47, Issue 7
Glaucoma  |   July 2006
Views of Glaucoma Patients on Aspects of Their Treatment: An Assessment of Patient Preference by Conjoint Analysis
Author Affiliations
  • Jonathan S. Bhargava
    From the Department of Ophthalmology, Queen’s Medical Centre, Nottingham, United Kingdom; and the
  • Bakula Patel
    Division of Primary Care, University of Nottingham, Nottingham, United Kingdom.
  • Alexander J. E. Foss
    From the Department of Ophthalmology, Queen’s Medical Centre, Nottingham, United Kingdom; and the
  • Anthony J. Avery
    Division of Primary Care, University of Nottingham, Nottingham, United Kingdom.
  • Anthony J. King
    From the Department of Ophthalmology, Queen’s Medical Centre, Nottingham, United Kingdom; and the
Investigative Ophthalmology & Visual Science July 2006, Vol.47, 2885-2888. doi:10.1167/iovs.05-1244
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Jonathan S. Bhargava, Bakula Patel, Alexander J. E. Foss, Anthony J. Avery, Anthony J. King; Views of Glaucoma Patients on Aspects of Their Treatment: An Assessment of Patient Preference by Conjoint Analysis. Invest. Ophthalmol. Vis. Sci. 2006;47(7):2885-2888. doi: 10.1167/iovs.05-1244.

      Download citation file:

      © 2015 Association for Research in Vision and Ophthalmology.

  • Supplements

purpose. To determine by conjoint analysis which factors in the management and treatment of glaucoma were of most importance to patients and to relate these factors to the patient’s clinical glaucoma condition.

methods. An interview-based study was performed. Demographic and visual function data are recorded. Participants completed the Visual Function Questionnaire-25 and ranked 10 hypothetical patient scenarios that contained different risks of moderate visual loss, postoperative complications, long-term blindness, use of topical medication, and glaucoma surgery. Conjoint analysis was performed to determine the relative importance of these factors for individuals and the group as a whole.

results. Eighty-two patients were interviewed from two consultants’ outpatient clinics. Forty-five were male and 37 female. Seventy-nine were white. The most important factors to patients with glaucoma were the risk of moderate visual impairment and the risk of blindness, with an importance of 38% and 27%, respectively. The use of topical medication had an importance of 11%. Proceeding to surgical intervention (trabeculectomy) had an importance of 15%, and the small risk of visual deterioration after surgery (trabeculectomy) had an importance of 9%.

conclusions. To patients, the most important factors regarding glaucoma and its treatment are the risks of moderate visual loss (the ability to continue to drive) and long-term blindness. The treatment methods used are of much less importance.

Open-angle glaucoma is a common condition worldwide, with a prevalence of 1% to 2% in white patients older than 40 years. 1 Epidemiologic studies have shown that glaucoma is responsible for 20% of cases of blindness in patients in an Afro-Caribbean population 2 and 6% of patients in a predominantly white population 3 who are defined as blind by World Health Organization criteria. In the United Kingdom, patients with glaucoma account for 34% of patients referred by optometrists to ophthalmology outpatient clinics. 4  
At present, the only treatable risk factor for open-angle glaucoma is raised intraocular pressure (IOP), and, in nearly all patients, treatment revolves around lowering IOP. Conventional treatment is with topical medication or surgery (a trabeculectomy). Topical treatment may continue for the remainder of the patient’s life. This may cause considerable problems for elderly patients who may have problems with administering the drops, and such a difficulty may affect compliance. There may be side effects from drops, in the form of stinging, transient blurred vision, and, occasionally, systemic side effects. However, it is generally believed that most patients are happy to use eye drops in preference to an operation. The standard surgical approach necessitates a trabeculectomy. As with all operations, there is a failure rate, and complications include a risk of visual loss at the time of surgery or in the postoperative period; however, 84% of patients remain drop free 12 months after the procedure. 5  
Recently, there have been several studies examining the different available treatment options. In terms of efficacy of treatment, the Collaborative Initial Glaucoma Treatment Study (CIGTS) 6 found that, at 5 years, a group treated with trabeculectomy and a group treated with eye drops demonstrated very similar visual field outcomes. In addition, they found that the impact on quality of life was similar in the two groups after 5 years. 7  
Although we have evidence to show that trabeculectomy and topical treatment are equally efficacious and that there is little difference in impact on quality of life, to our knowledge there have been no reports published to examine the patient’s preference for these treatments. One way of ascertaining patient preferences is by a tradeoff using conjoint analysis, which is a technique that has been used in market research for many years and has recently been applied to ophthalmology 8 in particular and to healthcare research 9 10 11 12 13 in general. The full-concept method of analysis generates several profiles or scenarios with all the factors of interest represented. It allows the relative importance of different factors to be assessed 14 and shows what features individuals are prepared to trade to obtain what they think is most important. It allows utility scores (also called part worths) to be generated that are analogous to regression coefficients for each of the factors and can be used to find the relative importance of each factor. 
It is therefore a quantitative technique that requires the subjects to make decisions and rank a series of scenarios. The scenarios consist of issues that are relevant to the decision maker. In our case, scenarios were based on the results of a pilot study. The subject had to evaluate the scenarios and had to make trade-offs to rank the scenarios. Two additional scenarios, termed holdouts, were used to asses the validity. These holdouts are generated from another random plan and not the experimental orthogonal plan. The analysis allowed us to discover the relative importance of each factor for individuals but, more important, to help to identify trends in a group. Conjoint analysis generates a utility score that is a numeric value of how desirable or undesirable a patient values the aspects presented in the scenario. A mean “utility score” will then show which factors the whole group finds desirable. 
The purpose of this study was to elucidate the factors that are important to patients with glaucoma in their management and prognosis. This information would greatly help when counseling patients on their course of treatment and would offer us an insight into aspects of management considered important by patients. 
A pilot study was performed to determine which factors were important to patients with glaucoma. A list of factors that were thought to be important was compiled by the ophthalmic team (JSB, AJK, AJEF). Ten patients were interviewed and were asked about their glaucoma and their perceptions of it, and from these interviews, we identified that the major concern was risk of visual loss. These were to be traded against the use of medication, undergoing surgery and the risk of early visual loss (a recognized complication of surgery). This led to five factors:
    Topical drop treatment (two levels, yes or no).
    Having a trabeculectomy (two levels, yes or no).
    Sudden loss of vision as a consequence of starting treatment (two levels, small [defined as <1%] increased risk versus no increased risk).
    Moderate visual loss, defined as being ineligible to hold a U.K. drivers license (three levels 5%, 10%, or 20%) at 10 years.
    Risk of blindness (of 2% or 5%) at 10 years.
With these factor levels, it is possible to generate 48 different scenarios (2 × 2 × 2 × 3 × 2 = 48), and rather than ask respondents to rank all 48 scenarios, we used a factor design (termed “orthoplan” in SPSS 14 ; SPSS, Chicago, IL) to generate randomly an “orthogonal array” of eight scenarios. In performing this, it is assumed that interactions between the different factors are negligible. An example of one scenario is shown in Figure 1
In addition, two holdout scenarios were generated. These were generated from another random plan and not the experimental orthogonal plan, to assess validity of the estimated utilities. 
The protocol for the participation of patients in the study adhered to the guidelines set forth in the Declaration of Helsinki. We interviewed 82 patients selected from two consultants’ glaucoma clinics between May 2003 and April 2004. The glaucoma clinic from which the patients were recruited is operated in a state-funded National Health Service Hospital. Glaucoma treatment is free for less-affluent patients, whereas the more affluent have to pay a standard prescription charge only, and therefore receiving treatment is not affected by ability to pay. All citizens of the United Kingdom have equal access to these facilities. 
Every fifth patient attending the clinic when interviewers were present was asked to participate. Seven patients declined to be interviewed. Two of the authors, one a general practitioner and the other an ophthalmologist, conducted the interviews (JB, BP). All the patients were required to have glaucoma that was being treated and to be able to read English. Patient information sheets were given to the participants. The interview had three stages:
    Collection of demographic and visual disease data.
    Completion of the visual function questionnaire (VFQ)-25. The VFQ-25 was designed by the National Eye Institute to assess visual impairment. It has been shown to be a reliable reflection of visual disability. 15
    Ranking of the 10 scenario cards.
Data Entry and Analysis
The demographic, medication, visual field, and ranking data were entered into a database (Access; Microsoft, Redmond, WA) and then transferred to analysis software (SPSS; SPSS, Inc., London, UK). The conjoint analysis was performed using the “conjoint” procedure in SPSS categories. This procedure takes the ranking of the different scenarios for each participant, and through a series of linear regressions, generates utility scores for each factor level. Each factor was specified as linear. SPSS calculates a regression coefficient for each factor and the utility scores are the product of the coefficients times the factor level. The distribution of the regresssion coefficients are plotted in Figure 2
The relative importance of each factor can also be expressed in percentage terms. This is done by taking the range of utility scores for any factor (highest minus lowest) and dividing it by the sum of all the utility ranges, and multiplying by one hundred. 
Ethical Approval
Ethical approval was obtained from the Trust and regional Ethics committees. 
Eighty-nine patients were approached and 82 agreed to take part in the study. The study population consisted of 45 men and 37 women aged between 31 and 93 years (mean, 69 years; median, 71 years). Seventy-nine were white, two were Afro-Caribbean, and one was Indian. Sixty-one had primary open-angle glaucoma, 13 normal-tension glaucoma, 5 narrow-angle glaucoma, and 3 secondary open-angle glaucoma. The average ocular pressure at diagnosis was 26 mm Hg right and 24 mm Hg left and the average pressure at the time of the study was 17 mm Hg right and 18 mm Hg left. There average score on the VFQ-25 was 0.82. The duration of glaucoma, treatment, and driving status are summarized in Table 1 . Within this sample, patients who were older were less likely to drive (P < 0.003) and the men were more likely to drive than were the women (P < 0.0001). The men had had glaucoma longer than the women in this sample (P < 0.03). The lower the number of drops used, the more likely that a β-blocker was used (P < 0.005). 
The outcomes of conjoint analysis are summarized in Table 2 . Internal-validity checks showed that the observed rankings for the eight packages and the utility scores from the conjoint analysis correlated highly (Pearson’s R = 0.964; P = 0.0001), and the two holdouts correlated perfectly, with the ranking predicted by the conjoint analysis (Kendall’s τ = 1.0). 
With the utility scores used as the dependant variables, multivariate analysis was performed with stepwise linear regression (P for entry < 0.05 and for removal >0.1). None of the demographic or clinical variables predicted were usable to predict the utilities scores for medical or for surgical treatment. The only associations are summarized in Table 3 . Note that the utilities scores are negative, showing lower values with increasing disability. Those who had reduced vision as determined by the VF-Q25 were less concerned about moderate visual loss than those with high scores (presumably they considered that they had already suffered moderate loss), whereas those who were taking medication seemed less concerned about early visual loss, possibly reflecting that they were not at risk. 
The main conclusion of this study is that the chief concern of patients with glaucoma is their visual outcome and that the method of treatment is of much less importance to them. In particular, there was no clear preference for medical or surgical treatment, and a small risk of visual loss (described as <1%) was not a major factor in influencing their decision. The perceived risk of future visual problems was the major concern, accounting for two thirds of the importance of the factors presented to them. 
Existing disability or age in particular did not influence their choice. 
Conjoint analysis is a powerful technique for eliciting patient preferences for healthcare. This study had a good response rate (82/89), and the model developed showed high levels of internal validity. 
All the participants came from one locality and were predominantly white. All could speak English. The most frequent diagnosis was primary open-angle glaucoma, which is the commonest form of glaucoma. However, this narrow geographic and demographic range means that the findings may have limited applicability to other populations and should be extrapolated with caution. 
Glaucoma is a disease that primarily affects the peripheral vision (loss of central vision occurring late in the disease process). Loss of peripheral vision has been shown to have a negative impact on one’s visual quality of life. Jampel et al. 16 found a moderate correlation between VFQ-25 score and Esterman visual field. Sumi et al. 17 found that retinal sensitivity in the lower hemifield within 5° of fixation and visual acuity in the better eye significantly correlated with the patient’s visual disability. An Italian study found a significant correlation between mean deviation and a questionnaire-based visual impairment assessment. 18  
The clinical scenarios in this study were generated by software and their complexity is limited by the number of factors included in the scenario generation, which is limited to ensure the feasibility of the study. The reduction of real clinical scenarios to formulated scenarios excludes some of the undoubted complexity of true clinical situations, but the tradeoff is between the complexity and inconvenience of treatment and the quality of the visual outcome. There is a large body of literature supporting the use of conjoint analysis in market research and healthcare. 8 13 19 20 21 In addition, the scenario choices were based on an initial pilot study that identified factors important to patients. 
Two observations are worthy of note. The first is that the difference in the risk of blindness at 10 years was only 3% (the choices were 2% or 5%) and at least one of the authors (AJEF) suspected that many would rather not have lifelong drug treatment for such a small difference in risk. However, the results make it clear that even such a small reduction in absolute risk was much more important to patients than the inconvenience of taking treatment. 
The second observation is that one of the outcomes was immediate loss of vision, whereas the other two outcomes were phrased as being a long-term risk (10 years). One might have suspected that patient age would have made a significant difference in how they viewed the relative importance of these risks, but despite a good age range within the study population, we could not demonstrate such an effect. Clearly, elderly people think that they have a future, and their perceptions should be remembered when deciding about treatment. 
It is clear that the loss of vision is perceived as a threat by patients, and the conclusion of this study is that patients are concerned with their visual outcome and not their method of treatment. Patients are prepared to have the inconvenience of lifelong treatment for even a small absolute reduction of a late adverse effect of visual loss or blindness and that this fear of late loss of vision is independent of age. The low patient preference to treatment modality means that physician preference can play a larger role. 
Figure 1.
An example of a scenario. This was generated randomly to include one factor level for each of the factors assessed in the conjoint model. The accompanying explanation made clear that “small ” in this context meant less than 1% risk of blindness and being unable to drive was within 10 years.
Figure 1.
An example of a scenario. This was generated randomly to include one factor level for each of the factors assessed in the conjoint model. The accompanying explanation made clear that “small ” in this context meant less than 1% risk of blindness and being unable to drive was within 10 years.
Figure 2.
Demonstrating the individual patient regression coefficients for “risk of blindness,” “surgical treatment,” “moderate visual impairment,” and “postoperative visual loss.” A regression coefficient of 0 indicates indifference to the factor. A negative regression coefficient indicates that the factor is less desirable.
Figure 2.
Demonstrating the individual patient regression coefficients for “risk of blindness,” “surgical treatment,” “moderate visual impairment,” and “postoperative visual loss.” A regression coefficient of 0 indicates indifference to the factor. A negative regression coefficient indicates that the factor is less desirable.
Table 1.
Summary of Disease Duration, Treatment Received, and Driving Status of the Patients in the Study Group
Table 1.
Summary of Disease Duration, Treatment Received, and Driving Status of the Patients in the Study Group
Clinical Details Number Percentage
Duration of glaucoma
 <1 y 13 16
 1–4 y 20 24
 5–9 y 16 20
 ≥10 y 33 40
Number treated with drops 72 88
Number with trabeculectomy 36 44
Number with side effects from drops 14 17
Driving status
 Current 38 46
 Ex-drivers 40 49
 Stopped due to eyesight 11 13
Table 2.
Outcomes of Conjoint Analysis
Table 2.
Outcomes of Conjoint Analysis
Factor Factor Level Mean Utility Mean Importance (%)
Risk of being unable to drive 5% −1.5055 38.60
10% −3.0111
20% −4.5166
Risk of blindness 2% −1.9390 26.88
5% −3.8780
Preference for trabeculectomy Yes −0.2866 14.85
No −0.5732
Preference for topical treatment None −0.1463 10.77
Drops −0.2927
Risk of early visual loss Normal −0.4573 8.90
Increased −0.9146
Table 3.
Outcomes of Stepwise Linear Regression Analysis
Table 3.
Outcomes of Stepwise Linear Regression Analysis
Variable B SE Beta Beta T P
Early visual loss
 Number of drops used −0.42 0.16 −2.9 −2.6 0.01
 Number of trabeculectomies 0.75 0.37 0.22 2.0 <0.05
Moderate visual loss
 VF-Q25 −0.016 0.0049 −0.35 −3.29 <0.002
TuckMW, CrickRP. The age distribution of primary open angle glaucoma. Ophthalmic Epidemiol. 1998;5:173–183. [CrossRef] [PubMed]
LeskeMC, WuSY, HymanL, NemesureB, HennisA, SchachatAP. Four-year incidence of visual impairment: Barbados Incidence Study of Eye Diseases. Ophthalmology. 2004;111:118–124. [CrossRef] [PubMed]
ForanS, WangJJ, MitchellP. Causes of visual impairment in two older population cross-sections: the Blue Mountains Eye Study. Ophthalmic Epidemiol. 2003;10:215–225. [CrossRef] [PubMed]
VernonSA. The changing pattern of glaucoma referrals by optometrists. Eye. 1998;12:854–857. [CrossRef] [PubMed]
EdmundsB, ThompsonJR, SalmonJF, WormaldRP. The National Survey of Trabeculectomy. II. Variations in operative technique and outcome. Eye. 2001;15:441–448. [CrossRef] [PubMed]
LichterPR, MuschDC, GillespieBW, et al. Interim clinical outcomes in the Collaborative Initial Glaucoma Treatment Study comparing initial treatment randomized to medications or surgery. Ophthalmology. 2001;108:1943–1953. [CrossRef] [PubMed]
JanzNK, WrenPA, LichterPR, et al. The Collaborative Initial Glaucoma Treatment Study: interim quality of life findings after initial medical or surgical treatment of glaucoma. Ophthalmology. 2001;108:1954–1965. [CrossRef] [PubMed]
RossMA, AveryAJ, FossAJ. Views of older people on cataract surgery options: an assessment of preferences by conjoint analysis. Qual Saf Health Care. 2003;12:13–17. [CrossRef] [PubMed]
RyanM, ScottDA, ReevesC, et al. Eliciting public preferences for healthcare: a systematic review of techniques. Health Technol Assess. 2001;5:1–186.
Van der PolM, CairnsJ. establishing patient preferences for blood transfusion support: an application of conjoint analysis. J Health Serv Res Policy. 1998;3:794–799.
RyanM. Using conjoint analysis to take account of patient preferences and go beyond health outcomes: an application to in vitro fertilisation. Soc Sci Med. 1999;48:535–546. [CrossRef] [PubMed]
SinghJ, CuttlerL, ShinM, SilversJB, NeuhauserD. Medical decision-making and the patient: understanding preference patterns for growth hormone therapy using conjoint analysis. Med Care. 1998;36(8 suppl)AS31–AS45. [PubMed]
RyanM, FarrarS. Using conjoint analysis to elicit preferences for health care. BMJ. 2000;320:1530–1533. [CrossRef] [PubMed]
SPSS Categories 6.1. 1994;SPSS, Inc. Chicago.
MangioneCM, LeePP, GutierrezPR, SpritzerK, BerryS, HaysRD. Development of the 25-item National Eye Institute Visual Function Questionnaire. Arch Ophthalmol. 2001;119:1050–1058. [CrossRef] [PubMed]
JampelHD, SchwartzA, PollackI, AbramsD, WeissH, MillerR. Glaucoma patients’ assessment of their visual function and quality of life. J Glaucoma. 2002;11:154–163. [CrossRef] [PubMed]
SumiI, ShiratoS, MatsumotoS, AraieM. The relationship between visual disability and visual field in patients with glaucoma. Ophthalmology. 2003;110:332–339. [CrossRef] [PubMed]
IesterM, ZingirianM. Quality of life in patients with early, moderate and advanced glaucoma. Eye. 2002;16:44–49. [CrossRef] [PubMed]
RyanM, McIntoshE, ShackleyP. Methodological issues in the application of conjoint analysis in health care. Health Econ. 1998;7:373–378. [CrossRef] [PubMed]
RyanM, McIntoshE, ShackleyP. Using conjoint analysis to elicit the views of health service users: an application to the patient health card. Health Expect. 1998;1:117–129. [CrossRef] [PubMed]
RyanM, HughesJ. Using conjoint analysis to assess women’s preferences for miscarriage management. Health Econ. 1997;6:261–273. [CrossRef] [PubMed]
Copyright 2006 The Association for Research in Vision and Ophthalmology, Inc.

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.