Medication-Related Predictors of Health-Related Quality of Life in Glaucoma Patients Enrolled in a Medicare Health Maintenance Organization Rajesh Balkrishnan, PhD,’ J. Brent Bond, MD,* Wesley G. Byerly, PharmD? Fabian T. Camacho, MS,3 and Roger T. Anderson, PhD3 ‘Division of Management and Policy Sciences, University of TexasSchool of Public Health, Houston, Texas,and2Department of Ophthalmology and 3Department of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, North Carolina
ABSTRACT Background: Glaucoma is an important public health concern in the United States, particularly among older adults (aged 265 years). Pharmacologic therapy for glaucoma consists mainly of topical eye drops containing beta-blockers or prostaglandin analogs. Objective: The goal of this study was to assess the associations between factors of topical medication use (self-reported medication compliance, belief in benefit of medication use, usage difficulty, usage assistance, and complexity of medication regimen) and health-related quality of life (HRQOL) in a cross-sectional population of older patients with glaucoma. Methods: A self-administered, 48question survey soliciting information on medication-taking behaviors, treatmentrelated factors, and HRQOL was mailed to members of a Medicare health maintenance organization who were aged 265 years and had primary open-angle glaucoma. Two mailings were conducted 4 months apart; the second was sent to members whose responses to the first mailing had not yet been received. The 12-Item Short-Form Health Survey (SF-12) and the 25-Item National Eye Institute Visual Function Questionnaire (VFQ-25) were used to assessHRQOL. Other questions addressed perceptions of eye drop use in these patients. Multiple regression techniques were used to analyze associations between medicationrelated factors and HRQOL in this population.
Results: The questionnaire was mailed to 589 patients; 375 responded (218 in the first mailing and 157 in the second mailing). A total of 358 responses were complete and analyzable (effective response rate, 62%). After controlling for the effects of other confounders, we found that self-reported difficulty in using eye drops was strongly associated with decreased HRQOL (11.5% in VFQ-25 total score and 8.4% in SF-12 mental health score, P< 0.05). Other medicationrelated factors that were examined were not significantly associated with changes in HRQOL. Conclusion: Based on our findings, patients aged 265 years with glaucoma were likely to have significant comorbidity, which affected both visual and general health and well-being perception. Additionally, a significant proportion of these patients reported difficulty with use of topical medication, which was independently associated with a significant decrease in HRQOL. Care of older patients with glaucoma should incorporate strategies to minimize the difficulty associated with medication use. (Am J Geriatr Pharmacother. 2003;1:75-81) Copyright 0 2003 Excerpta Medica, Inc. Key words: glaucoma, elderly, health-related quality of life, questionnaire, managed care, topical medication, patient outcome.
INTRODUCTION Glaucoma is an important public health concern in the United States, with recent estimates finding the prevalence of primary open-angle glaucoma rising from 4.6% to 13.8% in the Medicare population from 1991 to 1999, and the incidence varying from 4.6% to 7.8% during that period. 1,2 Pharmacologic therapy for glaucoma
consists mainly of topical eye drops, containing betablockers or prostaglandin analogs, usually requiring continuous application over extended periods of time.3 Although the clinical efficacy of glaucoma medications Accepted for publication September 29, 2003. Copyright
0 2003 Excerpta
in lowering intraocular pressure and preserving sight has been documented,4 little evidence exists regarding the impact that factors related to their use have on patient health outcomes, such as health-related quality of life (HRQOL)? A literature review using the MEDLINE database for 1966 through 2001, using the search terms &aucoma, treatment, outcomes, quality of life, and medications, found a few studies that assessedthe impact of glaucoma treatment on HRQOL.6-11 Perfetti et al6 conducted a survey in 332 patients with glaucoma (mean age, 65 years) to evaluate patient HRQOL. They found that 28% of the sample experienced worsening of HRQOL after glaucoma diagnosis, and 19% had judged their HRQOL to be seriously compromised. A medicationdosing schedule requiring multiple doses per day was a major reason for treatment dislike. A study by Sherwood et al* found that complexity of therapy, among other factors, was associated with poorer HRQOL. Few studies have examined behavioral mechanisms that may affect patients’ perceptions and outcomes related to medication use in glaucoma. One theoretical model for a study assessing the behavioral outcomes associated with a health behavior (ie, taking medicinal treatment for management of a medical condition) draws on the Health Belief Model.12 A person’s motivations to undertake a health behavior (in this case, medication use) can be separated into 3 main categories: individual perceptions (eg, severity of disease, importance of health, effectiveness of medication), modifying behaviors (eg, demographic behaviors, perceived threat), and likelihood of action (eg, adherence to medication regimen). Reviews of empiric work on treatment compliance support the theory that regimen complexity, side effects, behavioral cues in the physical and social environments, and patient education affect compliance rates in older adults.13T14 A study by Gurwitz et al l5 of the New Jersey Medicaid program clarified some glaucoma medication noncompliance issues. Factors associated with noncompliance included using glaucoma medication requiring >2 administrations per day, as well as the presence of multiple medications in the patient’s drug regimen. Drawing on classic models of health behavior and these empiric works, one could hypothesize that ease of medication use, simple medication regimens, adherence to medication therapy, belief in benefits of medication use, and independent ability to administer glaucoma medication would be associated with improved patient outcomes, such as HRQOL, in older adults with glaucoma using
topical medication therapy. 12-1s However, our literature search found little empiric evidence regarding pharmacologic therapy-related correlates of HRQOL in patients with glaucoma. One cross-sectional study16 in a Canadian population of 235 patients with chronic openangle glaucoma found that health status was correlated with the visual functioning of the patient as well as the number of oral medications the patient was taking. Our search did not find any study that examined the relationship between issues related to topical medication use specifically related to glaucoma and health outcomes in patients. The study of topical medication-related issues may be of particular importance because all medications used to treat this condition are available in the form of topical eye drops. The current study was designed to assessthe associations between factors of topical medication use (self-reported medication compliance, belief in benefit of medication use, usage difficulty, usage assistance, and complexity of medication regimen) and HRQOL in a cross-sectional population of older patients with glaucoma. MATERIALS AND METHODS Study Population The target population consisted of patients aged 265 years with primary open-angle glaucoma enrolled in a Medicare health maintenance organization (HMO) in the southeastern United States (Qualchoice of North Carolina, Winston-Salem, North Carolina). The population was restricted to this age group because complete information was available only on these patients as a part of a larger risk-adjustment study.17 This managed-care plan has been operational since late 1996 and is the sole insurer of medical care to the members (lock-in risk/benefit plan). The time frame for examining the claims data was October 1, 1996 (time of initiation of Medicare HMO), through September 30, 2000 (date on which study was closed to identify potential subjects for mailed survey). Patients were identified from the database using International Classification of Diseases, Ninth Revision, Clinical Modijkation (ICD-9-CM),ls codes for glaucoma, as well as receipt of medications for glaucoma (identified by a therapeutic category index developed for use within the HMO pharmacy database). The availability of a substantial discount for prescription refills after the monthly coverage limit was exhausted enabled us to reliably capture the prescriptions. However, prescription data were not available for all survey respondents because some were not enrolled in the prescription benefit plan. The study was restricted to J
ambulatory patients who were still enrolled in the Medicare HMO as of January 1, 2000. A mailed questionnaire and cover letter (both approved by the regional Health Care Financing Administration office [Raleigh, North Carolina] and the institutional review board [Winston-Salem, North Carolina]) were sent to study-eligible members in March 2000. The questionnaire was again sent in July 2000 to members whose responses had not yet been received. Study Instrument The primary purpose of the questionnaire mailing was to obtain descriptive information on HRQOL in the enrolled population to serve as a basis for the design of a potential glaucoma care management program. Therefore, to maximize response rates in a population with some level of visual impairment, questions other than the HRQOL component (37 items) were kept to a minimum ( 11 items). The 48-item questionnaire comprised the following instruments: (1) the 25-Item National Eye Institute Visual Function Questionnaire (VFQ-25), an instrument to assess visual functioning found to be reliable and valid across several common eye diseasesi9; (2) the 12-Item Short-Form Health Survey (SF-12), a measure of general health status19; and (3) an eye medication treatment questionnaire component that included 6 questions assessing the perceptions associated with eye drop use in glaucoma patients. The latter instrument was developed from summarizing anecdotal patient communications with an ophthalmologist and a review of the treatment satisfaction literature. The VFQ-25 consists of 12 subscales: self-rated general health, overall vision, neat-vision, distance vision, social hmctioning, vision-specific dependency, vision-specific role difficulties, vision-specific mental health, driving, color vision, peripheral vision, and ocular pain. The mean of the summary scores on each of these 12 subscales is used to create a composite score from 0 to 100, with lower scores indicating worse self-reported visual functioning. Internal consistency estimates indicate that the subscales of the VFQ-25 are reliable.‘” The validity of the VFQ-25 was suggested by high correlations between the short- and long-form versions of the measure. Additionally, between-group differences were observed in scores for persons with different eye diseases of varying severity. l9 Because the article describing the psychometrics of the VFQ-25 was still in press at the time that the current study was conducted, a prepublication draft was obtained from the lead author of that article.
The SF-12 questionnaire, a valid and reliable measure of general health status, contains questions related to the effect of disease on physical health, mental health, and social health, as well as on health perceptions. The scores on 8 subscales are collated to produce summary scores for physical components (PCS) and mental components (MCS)?O The eye medication treatment questionnaire assessed self-reported adherence to eye medications (range: never to all of the time), use of multiple medications (range: 0 to >5), difficulty in using eye drops (range: never to all of the time), reasons for difficulty in using eye drops (choices: multiple administration, physical inability, irritation and side effects, other [detailed by the responder]), assistance needed to use eye drops (range: never to all of the time), and patient’s perception that use of eye medication will reduce need for surgery (range: strongly disagree to strongly agree). The instruments used in the current study (with the exception of the 6-item treatment component) have been used successfully in a self-administration setting.21 In addition, the questionnaire included 5 questions about comorbid conditions, falls, living arrangements, and assistance required to complete the questionnaire. These questions were chosen based on previous work by the authors in older adultsF2 as well as clinical judgment of risk factors that may be of particular importance to patients with visual impairments. Statistical Analysis Summary scores were calculated for the VFQ-25 and the SF-12 (PCS and MCS). The responses to the 6 items in the treatment questionnaire were converted into categorical (yes/no) responses for the analyses, based on extreme cutoff points. Bivariate statistical analysis (analysis of variance) was used to determine relationships between hypothesized study factors and HRQOL scores (ie, VFQ-25, SF-12 PCS, and SF-12 MCS). Variables that were significantly associated with 21 HRQOL score were included in the multivariate regression analysis. Multiple ordinary least squares regression analysis was used to determine strengths of association of the HRQOL score (ie, VFQ-25 total score, and SF-12 PCS and MCS scores) predictors.23 Because the HRQOL scores were skewed, we applied natural logarithmic transformation to approximate normal distribution; in addition, methods proposed by Halvorsen and Palmquist17 with modifications by Kennedy24 were used to approximate our regression parameter estimate (B). Regressions were checked for violations of the Gaussian
assumptions.25>26 All statistical analyses were conducted using Stata statistical software, version 6.0 (Stata Corporation, College Station, Texas). RESULTS The questionnaire was mailed to 589 patients; 375 responded, 218 in the first mailing and 157 in the second mailing. Twenty-four questionnaires were returned as undeliverable or stating that the enrollee had died. A total of 358 responses (95% of responders) were complete and analyzable: 210 from the first mailing and 148 from the second mailing (effective response rate of survey, 62%). There were no statistically significant differences in patient characteristics and responses across the 2 mailings. Table I outlines the demographic, medication-related, and HRQOL outcome variables in the study sample. Sixty-nine percent of the 358-patient study sample were women (247 patients), and the overall mean (SD) age was 75.8 (7.3) years. The mean (SD) scores on the SF-12 PCS, SF-12 MCS, and VFQ-25 were 39.3 (12.0), 52.2 (9.9), and 75.5 (18.9), respectively (ranges on all scores: o-100). Thirty-three percent of respondents (118 patients) reported living alone, 40% (143 patients) reported having a fall in the last year, and 20% (72 patients) reported believing that their health had significantly worsened in the previous year. Fourteen percent of patients (50 patients) reported difficulty self-administering eye drops, and 17% (61 patients) reported needing assistance to administer eye drops. Eighty-two percent of the surveyed population (293 patients) described their medication use as regular (indicating adherence to prescribed regimen), 27% (97 patients) reported using 25 different total medications daily, and 59% (2 11 patients) reported believing that their use of topical eye drops reduced the need for surgery. The results of the regression analyses are outlined in Table II. The inclusion of significant predictors from the bivariate models in the final regression models resulted in a variance prediction rate of 27% for the VFQ-25 and SF-12 PCS scores, and 34% for the SF-12 MCS scores. Difficulty in eye drop use was associated with an 11.5% decrease in VFQ-25 score (P < 0.05) and a 8.4% decrease in the SF-12 MCS (P <:0.01) but was not associated with any changes in the SF- 12 PCS scores. Taking 25 medications daily was not associated with changes in any of the HRQOL scores in the multivariate analyses. Other factors that were significantly associated with decreased HRQOL outcomes were increased number of comorbidities and perception of health worsening in last
Table I. Demographics and health-related qualityof-life (HRQOL) data for 358 members of a southeastern US Medicare health maintenance organization who were aged 265 years, had primary open-angle glaucoma, and responded to a mailed questionnaire. Characteristic Demographics Age. Y” Mean (SD) Range Sex, no. (%I Women Men HRQOL No. of comorbidities,” mean (SD) Living alone,* no. (o/o) Fall in previous year,” no. (%I Belief of health worsening in previous year,* no. (%) Taking >5 medications daily,” no. (%) Regular use of eye drops,+ % Assistance needed in eye drop use, % Difficulty in self-administering eye drops,* % Belief that medication use reduces need for surgery, % SF -12 PCS score,* mean (SD) SF -12 MCS score,* mean (SD) VFQ-25 total score+*
75.8 (7.3) 65-89 247 (69) 111 (31) 2.3 (2.1) 118 (33) 143 (40) 72 97 293 61 50
(20) (27) (82) (17) (14)
211 39.3 52.2 75.5
(59) (12.0) (9.9) (18.9)
SF-12 = 12Item Short-Form Health Survey; PCS = physical components subscale; MCS = mental components subscale; VFQ-25 = 25-Item National Eye Institute Visual Function Questionnaire. *P < 0.05, for association with 21 health-related quality-of-life score, using bivariate analysis of variance. +Adherent to prescribed regimen. *Based on a score of 0 to 100.
year (both P 4 0.01). Having fallen in the last year was significantly associated with decreased SF- 12 MCS (P < O.Ol), whereas findings of increased age were mixed when linear and exponential effects were included in the model. DISCUSSION As determined by our literature review, this is one of the first empiric explorations of the relationships between medication-related factors and associated HRQOL in glaucoma patients receiving Medicare benefits. The results of this exploratory study underscored several
Table II. Results of multiple regression analysis” estimates of predictors of health-related quality of life among 358 members of a southeastern US Medicare health maintenance organization who were aged 265 years, had primary open-angle glaucoma, and responded to a mailed questionnaire. Dependent
VFQ-25 Total Score (natural logarithm)+
&e, Y Age squared Comorbidities, no. Falls in last year Perception of health worsening Taking 25 medications daily Difficulty in eye drop use Constant VFQ-25 = 254tem ponents subscale; *Parameter estimates +R2 = 0.27. *R2 = 0.34. §P < 0.05, Student rrP < 0.01, Student
in last year
0.110’ -o.ooo795 -0.04111 -0.016 -0.2111 0.050 -0.1205 0.32
(0.051) (0.00034) (0.0110) (0.040) (0.067) (0.037) (0.057) (1.93)
SF-12 PCS Score (natural logarithm)+ 0.110 -0.0007611 -0.050~~ -0.047 -0.2211 -0.031 -0.030 -0.20
National Eye Institute Visual Function Questionnaire; SF -12 = 12.Item MCS = mental components subscale. Cp) are presented with robust SEs indicated in parentheses.
t test assessing t test assessing
SF-12 MCS Score (natural logarithm)*
(0.039)11 (0.00025) (0.0075) (0.031) (0.042) (0.038) (0.045) (1.51) Health
0.031 (0.027) -0.00019 (0.00017) -0.02011 (0.0057) -0.07411 (0.023) -0.1911 (0.036) -0.000083 (0.024) -0.08711 (0.036) 2.8211 (1.07) Survey;
PCS = physical
p = 0. p = 0.
aspects of medicine-taking behaviors and health perceptions of glaucoma patients aged 265 years enrolled in a Medicare HMO. First, these patients reported varying degrees of difficulty in using their eye drops as well as difficulty adhering to their prescribed eye drop medication regimens. Second, the health perception and visual HRQOL scores of these participants were lower than national estimates of older adult populations.19320 In bivariate analyses, the only medication-related factors that were significantly associated with HRQOL were the use of 25 medications daily and difficulty in use of eye drops for glaucoma (P < 0.05 and P < 0.01, respectively). In the multivariate analyses, difficulty with eye drop use remained the only medication-related factor significantly predictive of lower VFQ-25 and SF-12 MCS scores (both P < 0.01). As expected, increased comorbidity and significant health worsening in the last year were also strongly associated with poorer HRQOL in glaucoma patients aged 265 years (both P < 0.01 for all 3 measures). Empiric studies of predictors of HRQOL could potentially be important in designing effective, evidencebased disease management interventions for older adults with glaucoma, especially interventions designed to improve patient outcomes through prophylactic medica-
tion use. Glaucoma treatment requires continuous rather than symptomatic pharmacotherapy, and regular medication use is vital to the success of glaucoma management. To design educational interventions to better manage glaucoma in older adults, it must be known to what extent patient medication-taking characteristics influence patient outcomes. Strategies and issues to alleviate difficulty with medication use in older patients with glaucoma need further investigation. These include compliance aids2’ (eg, Easidrop [Vidcom Marketing Ltd., London, United Kingdom], Autodrop [Owen Mumford Ltd., Woodstock, United Kingdom], OptiCare [OptiCare Health Systems, Inc., Waterbury, Connecticut]) and adherence-enhancing educational interventions, such as patient reminder letters and educational video programs. The difficulties associated with medication use in older patients with glaucoma could have several causes.5J6 First, medicinal treatment for glaucoma requires longterm compliance with medication instillation on a daily basis. Second, the medication formulation represents an obstacle to treatment (topical application to the eye is difficult for many older adults). Third, many older adults have considerable comorbidity, such as arthritis, which impairs their ability to depress the applicators of the eye
drops. As previous studies in other conditions such as asthma have demonstrated,28l29 complex chronic medication regimens often result in poor compliance in older adults, potentially leading to adverse health outcomes and poor perception of health. The current study should be considered only an initial exploration of these issues and, due to certain study limitations, caution should be exercised in interpreting our findings. We were limited to self-reported information available in the questionnaire for examining our associations. We also relied on the ICD-9-CM codes and electronically recorded medication use data to determine our study population. The study was framed using a behavioral outcome pathway that excluded clinical efficacy of medications because of design limitations. Using our cross-sectional design, it was difficult to examine history of medication use and actual physician prescriptions. The survey design also limited the scope of questioning to maintain a reasonable survey length. Consequently, we were unable to obtain detailed information on all types of medications being used. The cross-sectional nature of the study also prevented a detailed examination of outcomes related to difficulty with eye drops and poor self-reported adherence, such as discontinuation. The study population was limited to ambulatory Medicare HMO-enrolled adults aged 265 years with glaucoma, who may not be representative of adults with glaucoma in general. Also, this study is subject to potential nonresponse bias (62% effective response rate). However, this response rate is typical of mailed questionnaires.** In addition, the 6-item patient-perception component has not been officially validated for glaucoma and, therefore, could be subject to some criticism. Finally, one cannot infer causation from any crosssectional study design. However, these limitations should not undermine the trends observed in the questionnaire responses. CONCLUSIONS This cross-sectional survey study found that a significant proportion of older adults with glaucoma experienced difficulty in using their topical medications for glaucoma, and this difficulty was, in turn, associated with lower HRQOL scores. Based on our findings, patients aged 265 years with glaucoma were likely to have significant comorbidity, which affected both visual and general health and well-being perception. Strategies and issues to alleviate difficulty with medication use in older glaucoma patients, including compliance aids or adherence-enhancing
educational interventions, should be investigated. To confirm and build on our findings, future studies are required in larger populations with stronger experimental designs to jointly examine mechanisms to reduce the difficulty of medication use and to improve therapeutic efficacy in the treatment of older adults with glaucoma. ACKNOWLEDGMENT This study was funded by a grant from Pharmacia Corporation (now part of Pfizer Inc), New York, New York. REFERENCES 1. Lee PP, Feldman ZW, Osterman J, et al. Longitudinal prevalence of major eye diseases. Arch Ophthalmol. 2003; 121:1303-1310. 2. Sloan FA, Brown DS, Carlisle ES, et al. Estimates of incidence rates with longitudinal data. Arch Ophthalmol. 2003;121:1462-1468. 3. Linden C, Alm A. Prostaglandin analogues in the treatment of glaucoma. DrugsA&qg. 1999;14:387-398. 4. The Advanced Glaucoma Intervention Study (AGIS): 7. The relationship between control of intraocular pressure and visual field deterioration. The AGIS Investigators. Am J Ophtbalmol. 2000;130:429440. 5. Lee BL, Wilson MR. Health-related quality of life in patients with cataract and glaucoma. J Glaucoma. 2000;9: 87-94. 6. Perfetti S, Varotto A, Massagrandi S, et al. Glaucoma and quality of the life. Acta Ophthalmol Stand Suppl. 1998;76: 52. 7. Gutierrez I’, Wilson MR, Johnson C, et al. Influence of glaucomatous visual field loss on health-related quality of life. Arch Opbtbalmol. 1997;115:777-784. 8. Sherwood MB, Garcia-Siekavizza A, Meltzer MI, et al. Glaucoma’s impact on quality of life and its relation to clinical indicators. A pilot study. Opbtbalmology. 1998; 105:561-566. 9. Wilson MR, Coleman AL, You F, et al. Functional status and well being in patients with glaucoma as measured by the Medical Outcomes Study Short Form-36 questionnaire. Opbtbalmology. 1998;105:2112-2116. 10. Parrish RK II, Gedde SJ, Scott IU, et al. Visual function and quality of life among patients with glaucoma. Arch Opbtbalmol. 1997;115:1447-1455. 11. Mills RI? Correlation of quality of life with clinical symptoms and signs at the time of glaucoma diagnosis. Tram Am Opbtbalmol Sot. 1998;96:753-812. 12. Rosenstock IM. Why people use health services. Milbank Mem Fund Q 1966;44(Suppl):94-127. J
13. Anderson RT, Ory M, Cohen S, McBride JS. Issues of aging and adherence to health interventions. Control Clin Trials. 2000;21:17(Supp1 5):171S-183s. 14. Park DC, Jones TR. Medication adherence and aging. In: Fisk AD, Roger WA, eds. Handbook of Human Factors and the Older Adult. San Diego, Calif: Academic Press; 1997: 257-288.
15. Gurwitz JH, Glynn RJ, Monane M, et al. Treatment for glaucoma: Adherence by the elderly. Am J Public Health. 1993;83:711-716. 16. Montemayor F, Sibley LM, Courtright I’, Mikelberg FS. Contribution of multiple glaucoma medications to visual function and quality of life in patients with glaucoma. Can J Ophthalmol. 2001;36:385-390. 17. Halvorsen R, Palmquist R. The interpretation of dummy variables in semilogarithmic equations. Am Econ Rev. 1980; 70:474475.
18. International Classification of Diseases, Ninth Revision, Clinical Modification. Los Angeles, Calif: Practice Management Information Corporation; 1995. 19. Mangione CM, Lee PP, Gutierrez P, et al. Development of the 25-item National Eye Institute Visual Function Questionnaire (VFQ-25). Arch Ophthalmol. 2001;119: 1050-1058. 20. Ware JE, Kosinski M, Keller SD. SF-12: How to Score the SF-12 Physical and Mental Summary Scales.Boston, Mass: Health Institute, New England Medical Center; 1995.
21. Schiffman RM, Christianson MD, Jacobsen G, et al. Reliability and validity of the Ocular Surface Disease Index. Arch Opbtbalmol. 2000;118:615-621. 22. Balkrishnan R, Anderson RT. Predictive power of a riskassessment questionnaire across different disease states: Results in an elderly managed-care enrolled population. Am JManag Care. 2001;7:145-153. 23. Greene WH. Econometric Analysis. 3rd ed. Upper Saddle River, NJ: Prentice Hall; 1997. 24. Kennedy P. Estimation with correctly interpreted dummy variables in semilogarithmic equations. Am Econ Rev. 1981; 71:801. 25. Cook D, Weisberg S. An Introduction to Regression Graphics. New York: Wiley; 1994. 26. White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasricity. Econometrica. 1980;48:817-838. 27. Rivers PH. Compliance aids-Do they work? Drugs Aging. 1992;2:103-111. 28. Balkrishnan R, Christensen DB. Inhaled corticosteroid nonadherence and immediate avoidable medical events in older adults with chronic pulmonary ailments. Asthma. 2000;37:511-517. 29. Balkrishnan R, Christensen DB. Inhaled corticosteroid use and associated outcomes in elderly patients with moderate to severe chronic pulmonary ailments. Clin They. 2000;22: 452469.
Address correspondence to: Rajesh Balkrishnan, PhD, Associate Professor of Management and Policy Sciences, University of Texas School of Public Health RASE331, 1200 Herman Pressler, Houston, TX 77030. E-mail: [email protected]