SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:

Objective

To describe the health state preferences of patients with osteoarthritis (OA) according to the level of pain and disability and the extent of gastrointestinal side effects from nonsteroidal antiinflammatory drugs (NSAIDs).

Methods

Using combinations of 5 OA health states (4 specifying medication use) and 6 gastrointestinal side effect profiles, we developed 25 scenarios. In an Internet survey, adults with OA evaluated 5 randomly chosen health state–side effect scenarios (in addition to scenarios for congestive heart failure and wearing dentures, as benchmarks). They rated the scenarios on a 0–100 scale, in which 100 corresponds to best imaginable health. Unadjusted mean ratings were calculated using a difference-in-difference approach. A generalized linear model was used to estimate the effects of disease severity and side effect severity on the ratings, after controlling for patient characteristics.

Results

A total of 4,386 respondents whose mean age was 55.3 years, of whom 3,107 (70.8%) were women and 4,007 (91.4%) were white, completed the survey. Mean adjusted ratings for health state–side effect scenarios ranged from 94.9 for the mildest scenario to 25.3 for the most severe scenario. Severity of NSAID side effects had a greater negative influence on the ratings in milder OA states than in more severe OA states. Ratings were lower among men (P < 0.001) and among respondents with OA pain in the previous 24 hours (P < 0.001). Disease severity had a greater effect on ratings than did side effect severity.

Conclusion

Patients consider pain and functional limitations associated with OA to be important determinants of well-being. Future research should attempt to determine whether patients prefer reductions in their OA-related pain and disability over improvements in treatment side effect profiles.

Arthritis and other rheumatic conditions are leading causes of disability in the US, affecting 1 of every 3 noninstitutionalized adults (1, 2). The prevalence of osteoarthritis (OA), the most common form of arthritis, increases with age and is highest among persons ages >45 years (3). With the aging of the US population, the prevalence of OA is expected to increase substantially by 2030 (4).

Pain from OA is commonly treated with nonsteroidal antiinflammatory drugs (NSAIDs), including cyclooxygenase 2 (COX-2) inhibitors. All NSAIDs are associated with side effects. Patients' ratings of OA health states and/or side effects from medications have not been reported widely in the literature, and few studies in any therapeutic area have examined the influence of treatment-related side effects on patients' assessments of health states or health utilities, a concept similar to health-related quality of life (5–8). Researchers generally have studied side effects as separate health states (e.g., dyspepsia) without considering the combined effects of both disease severity and side effect severity (5–7, 9). Because side effects may affect patients' overall well-being substantially, an understanding of health state preferences across levels of OA severity and profiles of treatment-related side effects will allow for better assessment of new and existing OA therapies, and will inform patient and physician decision making (8, 10, 11).

In this study, we sought to describe the health state preferences of patients with OA according to their level of pain and disability and according to the extent of gastrointestinal side effects from NSAIDs.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:

Scenarios

Using the Western Ontario and McMaster Universities Osteoarthritis Index (12), we developed narratives describing 5 OA health states, ranging in severity from none/mild to extreme. The narratives described severity of pain and stiffness, level of functioning and daily activities, and medication use (if any). We paired health state narratives associated with medication use with 1 of 6 narratives describing NSAID-related gastrointestinal side effects, ranging in severity from no side effects to bleeding ulcer requiring surgery, yielding 25 OA health state–side effect scenarios. In consultation with a panel consisting of 2 rheumatologists and an orthopedic surgeon experienced in the treatment of OA, we reviewed and modified the 25 scenarios. The panel also developed 2 additional scenarios, 1 for mild-to-moderate congestive heart failure (CHF) and 1 for wearing dentures. Patients' ratings of these 2 scenarios provided a benchmark against which to compare the ratings of the OA health state–side effect scenarios. All narratives appear in Appendix A. The OA health state–side effect scenarios are grouped by index number (ranging from 1 to 5 [OA1–OA5] for the OA health states and 0 to 5 [SE0–SE5] for the side effect profiles). Higher index numbers indicate greater severity.

Subjects were asked to rate each OA health state–side effect scenario relative to the best and worst health conditions that they could imagine. They were not told specifically to equate death with the worst possible health condition. Respondents rated scenarios 0–100 on a 10-cm vertical visual analog scale (VAS) (13). The VAS was presented as a feeling thermometer, with a top value of 100 defined as the best possible health condition and the bottom value of 0 defined as the worst possible health condition. A similar rating scale was used in the development and evaluation of the McKnee technique, a classification system and direct utility measure of health states associated with knee replacement surgery (14).

To complement the analysis of scenario ratings, the survey contained several questions relating to subjects' experiences with OA. Additional survey data not related to this analysis are reported elsewhere (15, 16). The survey was pilot-tested in a convenience sample of 20 adults. Based on interviews with participants, the physician panel reviewed suggested changes and incorporated them into the final version of the survey.

Survey administration

The survey was adapted for the Internet by Harris Interactive (Rochester, NY), an international market research and consulting firm, and was fielded from April 21 through April 24, 2003. Persons enrolled in Harris Interactive's Chronic Illness Specialty Panel were invited by E-mail to participate in a survey about OA. Invitees were directed to the survey Web site, where they were given a brief description of OA and asked if they had ever been told by a physician or other clinician that they had OA. Participants who reported receiving a diagnosis of OA and who resided in Australia, Canada, the United Kingdom, or the US were eligible to participate after providing their informed consent on-line. (Harris Interactive also mailed a paper copy of the informed consent document to participants.) Eligible panelists who completed the survey received 100 HIPoints, redeemable for gifts on the Harris Interactive Web site (800 points = 5 dollars). The institutional review board of Duke University Medical Center approved the study.

All survey respondents evaluated the CHF and wearing dentures scenarios, the OA1/SE0 (none/mild OA–no side effects) scenario, and 4 additional OA health state–side effect scenarios. The 4 additional scenarios were assigned randomly so that each respondent was presented with 1 scenario in each OA health state. Respondents rated the wearing dentures and CHF scenarios first; they rated the 5 OA health state–side effect scenarios in random order.

We examined the patterns of patients' ratings for logical inconsistencies indicative of misinterpretation of the rating scale end points. For the benchmark health states, consistency was defined as rating CHF greater than or equal to wearing dentures. For the health state–side effect scenarios, consistency was defined as a rating score for OA1/SE0 of less than or equal to the maximum rating of any scenario involving OA5. We conducted all analyses on both sets of ratings to determine the effects of these exclusions.

Statistical analysis

Values for subject characteristics and for ratings of the CHF and denture scenarios are expressed as sample means or proportions, as appropriate. For the OA health state–side effect scenarios, results are expressed as the unadjusted mean ratings, calculated using a difference-in-difference approach as the average difference among respondents between the rating for OA1/SE0 and the rating for each of the other scenarios. For example, for each respondent (designated j) who evaluated scenario OA2/SE1, we computed the rating decrement relative to OA1/SE0 as follows: ([OA1/SE0]j − [OA2/SE1]j). We averaged the differences across the number of patients who evaluated scenario OA2/SE1, and then subtracted the simple average of the rating differentials for scenario OA2/SE1 from the sample mean rating for OA1/SE0, to generate an unadjusted mean rating for the OA2/SE1 (mild health state–side effect) scenario; this process was repeated for all remaining scenarios.

The difference-in-difference approach is preferred over comparisons of sample means because it maintains the relationships between individuals' ratings. We used OA1/SE0 as the comparison because all participants evaluated this scenario. Using analysis of variance, we identified clusters of benchmark health states and OA health state–side effect scenarios with similar ratings. We then used a generalized linear model with an identity link function and an assumption of normal distribution to estimate the impact of disease severity and side effect severity on the mean ratings while controlling for potential confounders such as age, sex, and recent OA pain, and also controlling for the clustering of ratings within patients. We used SAS software, version 9.1 (SAS Institute, Cary, NC) for all statistical analyses.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:

Characteristics of the respondents. Harris Interactive distributed 57,452 invitations by E-mail, and 5,550 subjects (9.7%) responded. Of those who responded, 1,029 (18.5%) did not report having OA. Of the remaining 4,521 participants (81.5%), 135 (2.5%) did not give informed consent. This left 4,386 respondents who were eligible, gave consent, and completed the survey. Survey respondents were similar to nonresponders (data not shown) but were slightly older (55.3 years for responders versus 52.6 years for nonresponders) and were more likely to be of white or Hispanic race/ethnicity (92.4% for responders versus 79.1% for nonresponders). US residents comprised 97.3% of the sample.

Of eligible respondents, 4,029 (91.9%) provided consistent ratings for the CHF and wearing dentures scenarios, and 4,080 (93.0%) provided consistent ratings for the OA health state–side effect scenarios. A total of 172 (3.9%) respondents provided inconsistent ratings across both health states. Excluding the 491 respondents (11.2%) who provided an inconsistent answer on either set of health states left a total of 3,895 wholly consistent responses. Table 1 shows the characteristics of the entire sample and the subset of respondents with consistent rating patterns for both the benchmark health states and the OA health state–side effect scenarios. Unless otherwise indicated, we present results for the benchmark health states based on the 4,029 respondents who provided consistent ratings across the CHF and wearing dentures scenarios. Ratings for the OA health state–side effect scenarios are based on the 4,080 respondents with consistent rating patterns for those scenarios.

Table 1. Characteristics of the respondents*
CharacteristicAll respondents (n = 4,386)Respondents with consistent rating patterns (n = 3,895)Respondents with inconsistent rating patterns (n = 491)
  • *

    Except where indicated otherwise, values are the number (%).

  • P < 0.001 versus respondents with inconsistent rating patterns.

  • P < 0.05 versus respondents with inconsistent rating patterns.

  • §

    Includes income reported as non-US currency.

  • Pain was measured on a scale of 0–10, where 10 indicates unbearable pain.

  • #

    Proportion reporting nausea and vomiting, heartburn, and/or stomach pain at least 1 time up to 3 times per month during the previous 3 months.

Age, mean ± SD years55.3 ± 10.854.9 ± 10.758.7 ± 10.7
Male1,279 (29.2)1,113 (28.6)166 (33.8)
Race/ethnicity   
 White4,007 (91.4)3,558 (91.4)449 (91.5)
 African American56 (1.3)49 (1.3)7 (1.4)
 Hispanic48 (1.1)40 (1.0)8 (1.6)
 Other/unknown275 (6.3)248 (6.4)27 (5.5)
Education   
 High school or lower630 (14.4)526 (13.5)104 (21.2)
 Some college1,847 (42.1)1,644 (42.2)203 (41.3)
 College and/or postgraduate1,837 (41.9)1,664 (42.7)173 (35.2)
 Other/unknown72 (1.6)61 (1.6)11 (2.2)
Income   
 <$50,0002,051 (46.8)1,793 (46.0)258 (52.6)
 $50,000 to $100,0001,382 (31.5)1,257 (32.3)125 (25.5)
 >$100,000450 (10.3)411 (10.6)39 (7.9)
 Other/unknown§503 (11.5)434 (11.1)69 (14.1)
Employed2,168 (49.4)1,986 (51.0)182 (37.1)
Recent osteoarthritis pain level, mean ± SD4.9 ± 2.44.8 ± 2.45.7 ± 2.4
Years since osteoarthritis diagnosis, mean ± SD13.5 ± 17.313.5 ± 17.213.6 ± 17.7
Recent gastrointestinal event#3,050 (69.5)2,709 (69.6)341 (69.5)

Ratings of benchmark health states. The summary mean rating for the CHF scenario was 21.3 (95% confidence interval [95% CI] 20.7–21.9), and the summary mean rating for the wearing dentures scenario was 73.0 (95% CI 72.2–73.7). Respondents who actually wore dentures (n = 837) gave wearing dentures a mean rating of 77.5 (95% CI 76.0–79.0), compared with a mean rating of 71.8 (95% CI 70.9–72.6) among those not wearing dentures (P < 0.0001). Respondents with CHF (n = 216) gave the CHF scenario a mean rating of 25.0 (95% CI 22.2–27.8), compared with a mean rating of 21.1 (95% CI 20.5–21.7) among respondents without CHF (P = 0.002).

Ratings of OA health state–side effect scenarios. Table 2 shows the unadjusted mean ratings for the 25 OA health state–side effect scenarios calculated relative to the OA1/SE0 (none/mild OA–no side effects) scenario. Analysis of variance of the mean difference-in-difference ratings revealed clusters of OA health state–side effect scenarios and benchmark health states (i.e., wearing dentures and CHF) that were rated similarly (Tables 2 and 3 and Appendix B). With the exception of OA2/SE0 and OA3/SE0, ratings for the 5 OA health states without side effects were significantly different from each other (P < 0.05). The greatest amount of clustering occurred with the most severe OA health state–side effect scenarios. Respondents with consistent rating patterns for the OA health state–side effect scenarios rated CHF similarly to the ratings of several scenarios involving OA4 and OA5 (Table 3).

Table 2. Difference-in-difference mean scenario ratings*
SE profileOA health state
OA1OA2OA3OA4OA5
  • *

    Values are the difference-in-difference mean rating (95% confidence interval [95% CI]) except for OA1/SE0, which serves as referent for clustering analysis of variance. P < 0.05 for most interactions among osteoarthritis/side effect (OA/SE) scenarios. Difference-in-difference mean ratings were 71.2 (95% CI 70.4–70.2) for the denture scenario and 22.2 (95% CI 21.5–23.0) for the congestive heart failure scenario. Results do not include the 306 respondents with inconsistent rating patterns across OA health state–side effect scenarios.

  • OA health states are as follows: OA1 = none/mild; OA2 = mild; OA3 = moderate; OA4 = severe; OA5 = extreme.

  • SE profiles are as follows: SE0 = no side effects; SE1 = mild side effects, treatable with over-the-counter medication; SE2 = mild side effects, treatable with prescription medication; SE3 = ulcer, treatable with prescription medication; SE4 = bleeding ulcer, treatable with prescription medication; SE5 = bleeding ulcer requiring surgery.

SE094.5 (94.1–94.9)61.6 (59.3–63.9)51.8 (39.2–42.8)29.4 (27.0–31.8)18.1 (16.2–20.0)
SE157.0 (55.4–58.5)47.0 (39.2–42.8)28.6 (26.9–30.3)19.9 (18.4–21.4)
SE254.8 (53.2–56.5)44.5 (43.0–46.0)29.4 (27.8–31.0)17.5 (16.1–18.9)
SE350.5 (48.8–52.2)42.1 (40.6–43.7)25.6 (24.1–27.2)20.8 (19.2–22.3)
SE447.9 (46.2–49.6)40.0 (38.4–41.6)24.8 (23.3–26.3)18.0 (16.5–19.4)
SE541.0 (39.2–42.8)35.4 (33.7–37.0)21.1 (19.6–22.6)16.9 (15.4–18.4)
Table 3. Clustering of osteoarthritis (OA) health state–side effect (SE) scenarios and benchmark health states*
State/profileNo. of respondersTukey grouping
123
  • *

    Clusters were identified by Tukey groupings, unadjusted for patient-level characteristics. The Tukey groupings indicate similar mean difference-in-difference ratings (those with the same letter are not significantly different). CHF = congestive heart failure.

  • Benchmark health state.

OA1/SE04,386 A
Dentures4,386 B
OA2/SE0383 C
   C
OA2/SE1827DC
  DC
OA2/SE2740DC
  DC
OA3/SE0409DC
  D
OA2/SE3818DE
   E
OA3/SE1808FE
  FE
OA2/SE4815FE
  F
OA3/SE2824F
  F
OA3/SE4762F
  F
OA3/SE3802FG
  FG
OA2/SE5803FG
   G
OA3/SE5781HG
  H
OA4/SE1812HI
  HI
OA4/SE2846HIJ
   IJ
OA4/SE0394KIJ
  KIJ
OA4/SE4801KIJ
  KIJ
OA4/SE3746KIJ
  K J
OA4/SE5787KLJ
  KLJ
CHF4,386KLJ
  KL
OA5/SE1769KL
  KL
OA5/SE4825KL
  KL
OA5/SE3763KL
   L
OA5/SE5806ML
  ML
OA5/SE2784ML
  M
OA5/SE0439M

Appendix B gives the results of the generalized linear model of OA health state–side effect scenario ratings. OA2 through OA5 were statistically significantly different from OA1 (P < 0.001 for all comparisons), and SE1 through SE5 were statistically significantly different from SE0 (P ≤ 0.001 for all comparisons, except P = 0.04 for the comparison with SE1). All of the OA health state–side effect interactions were statistically significant (P < 0.05) except OA3/SE1, OA3/SE2, OA3/SE3, OA3/SE4, and OA4/SE1. Male sex (P < 0.001) and more severe recent (within the previous 24 hours) OA pain (P < 0.001) were associated with lower ratings. Age, education, race/ethnicity, income, years since OA diagnosis, recent gastrointestinal events, and employment status were not associated with the ratings of health state–side effect scenarios.

Adjustment for patient-level covariates (Table 4) increased the mean ratings for all scenarios and altered the relative rankings of some scenarios, but these adjustments did not eliminate the inconsistencies shown in Table 2. Inconsistencies were again clustered around the more severe OA health state–side effect scenarios.

Table 4. Mean scenario ratings after controlling for patient characteristics*
SE profileOA health state
OA1OA2OA3OA4OA5
  • *

    Values are the mean adjusted rating (95% confidence interval). Results do not include the 306 respondents with inconsistent rating patterns across osteoarthritis (OA) health state–side effect (SE) scenarios. Respondent characteristics included in the generalized linear model of ratings included age, sex, race, education, income, recent OA pain, recent gastrointestinal event, years since OA diagnosis, and employment status.

  • OA health states are as follows: OA1 = none/mild; OA2 = mild; OA3 = moderate; OA4 = severe; OA5 = extreme.

  • SE profiles are as follows: SE0 = no side effects, SE1 = mild side effects, treatable with over-the-counter medication; SE2 = mild side effects, treatable with prescription medication; SE3 = ulcer, treatable with prescription medication; SE4 = bleeding ulcer, treatable with prescription medication; SE5 = bleeding ulcer requiring surgery.

SE094.9 (90.4–99.4)64.6 (57.9–71.2)56.3 (49.8–62.9)35.5 (28.8–42.2)25.6 (19.1–32.2)
SE161.9 (52.7–71.1)52.3 (39.5–65.1)36.4 (23.4–49.3)28.9 (16.2–41.6)
SE259.3 (50.0–68.6)50.0 (37.2–62.9)35.4 (22.4–48.5)25.9 (13.1–38.7)
SE355.5 (46.3–64.8)47.5 (34.7–60.3)33.1 (20.1–46.1)28.2 (15.4–40.9)
SE452.4 (43.2–61.7)46.6 (33.8–59.4)32.5 (19.5–45.6)27.8 (15.0–40.6)
SE546.6 (37.3–55.9)41.8 (28.9–54.7)29.3 (16.2–42.4)25.3 (12.4–38.1)

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:

Our analysis confirms the expectation that pain and functional limitations adversely affect patient well-being and suggest that OA, in its most severe forms, is associated with as substantial a decrement to quality of life as that associated with CHF, a serious chronic condition with a poor prognosis. Furthermore, side effects of therapies for OA have a negative influence on health status that is nonadditive and depends on the severity of the disease. First, increasing medication-related side effects have an increasingly negative effect on patients' health ratings, given OA severity. Second, the adverse impact of side effects is lower when OA pain is severe and higher when OA pain is milder. These findings have important implications for the economic evaluation of new and existing therapies for OA, as well as for patient care.

The literature assessing patient preferences across OA health states is sparse. Most studies have focused on the evaluation of pain in a single joint, a limited number of health states, or a combination of these in the context of a particular drug therapy (5, 11, 17). To our knowledge, this is the first general, patient-based study to focus on OA and to assess multiple aspects of patients' experiences. The combined assessment of OA and side-effect severity should facilitate the economic evaluation of OA therapies. It is important to note that the use of health-state ratings, or utilities, that do not incorporate both disease and treatment effects on quality of life can yield biased results. Models that do not consider the sometimes-substantial negative effects of side effects on health-related quality of life will overestimate the benefits of treatment for patients with mild disease. Similarly, models that combine separate assessments of OA severity and side effects will likely underestimate the benefit of therapy for those with more severe disease. In combination, these biases could result in an inefficient use of resources for the treatment of OA that reduces the overall quality of life.

Clinical practice guidelines suggest that mild OA should be managed through diet and exercise (18, 19). However, when pharmacologic intervention is unavoidable, it is recommended that physicians begin with the least expensive drug with the lowest risk of side effects. Our findings reinforce these guidelines, in that side effects had a greater impact on patients' evaluations of mild OA health states compared with their evaluations of health states with more severe disease. This may be particularly important given the recent controversy involving the risks associated with use of COX-2 inhibitors. Conversely, patients with severe disease may be more willing to accept the risk of side effects that accompanies pharmacologic and surgical treatments. Physicians' understanding of how patients with either less severe disease or more serious disease value tradeoffs between pain relief and treatment-related side effects should facilitate doctor–patient communication and appropriate therapeutic selection.

We empirically tested our finding that side effects had a greater effect in the milder OA health states by modeling side effect profiles as a continuous variable in the generalized linear model (data available from the authors). In that specification, the effect of side effects on rating scores was less notable in the more severe OA health states. We also re-estimated the ratings model using a log transformation, which increased the range of ratings by a few points but did not change the relationships between health states or the magnitudes of the parameter estimates for other explanatory values. Because the log-based generalized linear model requires a transformation of parameters, we present the more straightforward model in the current report.

Disease severity appeared to have a greater effect on ratings than did side effect severity, but we cannot conclude that patients value disease severity more than side effect severity, because these were not compared directly on the same scale. Patients may have interpreted side effects as controllable or curable in the short term or as single-occurrence events, as in the case of bleeding ulcer, rather than chronic, incurable impairments due to OA. However, the substantial effect of pain and disability on scenario ratings, even in milder OA health states, is consistent with the observations in other research studies that have examined patients' valuations of pain and disability (20).

We used a rating scale to measure patients' valuations of health states. The rating scale is simpler and more straightforward than other methods of eliciting health-related quality of life or health state preferences, such as the time tradeoff and standard gamble methods (21). Rating scales have been shown to outperform other utility measures in differentiating among alternative health states (22–24). Although ratings are not used as a measure under conditions of uncertainty, other investigators have reported them to be a reliable and consistent method for assessment of quality of life, noting that rating-scale values tend to be lower than those elicited by the standard gamble and other utility measures (25–33). In a study of patients with chronic musculoskeletal pain, a rating scale was preferred to a standard gamble instrument as a means of assessing the impact of interventions on health outcomes, because it had better construct validity and ability to detect differences in health status (33).

Respondents' ratings were related inversely to recent OA pain, a marker of severity (34). In our sample, respondents evaluated their recent OA pain as 4.9 on a scale from 0 to 10. Levels of OA severity were well-represented among participants, except the more extreme values, and most respondents appeared to have at least moderate OA pain. Future studies should strive to recruit patient cohorts that represent the spectrum of OA severity.

Our study has some limitations. First, we relied on self-reported OA diagnosis. The respondent ratings may include individuals whose symptoms matched the brief description of OA that we used but which were, in fact, not caused by OA. We would not expect this to seriously bias the results. Individuals who have neither a diagnosis of OA nor symptoms consistent with OA may also have participated. Because pain was negatively related to ratings in the multivariable regression analysis, we would expect the inclusion of non-OA patients who have no OA-related symptoms to bias the health-state ratings upward.

Second, our sample may not be representative of the general population of patients with OA. Sex and race distributions were similar to those reported in the 2000 Medical Expenditure Panel Survey (MEPS) (35), in which 96.6% of patients with OA were white, 2.7% were African American, and 71.0% were women (using national population weights). In our survey, 92.4% of respondents were white or Hispanic, 1.3% were African American, and 70.8% were women. Our findings should be applicable to a large proportion of patients, because OA is more prevalent among women. The proportions of African American patients in the MEPS and in our sample are small. Future studies to ascertain the impact of disease and side effect profiles in African American patients would be of interest. Furthermore, our sample appears to be relatively well-educated and reported relatively high incomes.

Respondents in our sample were younger than respondents in the MEPS (mean 55.3 years versus 63.1 years). This may be due to a greater degree of survivor bias in our survey or could be a reflection of the younger age of Internet users (36). Because age did not affect unadjusted difference-in-difference ratings (data not shown) and was not a significant predictor of rating in the multivariable analysis, we believe that the relatively young age of our sample was not a significant source of bias.

The overall response to the initial survey invitation was low; 7.6% of persons who were sent an invitation to participate completed the survey during the 4-day survey period. Harris Interactive and others report a range of Internet survey response rates from 7% to 40%, depending on the topic and length of the survey, the content and targeting of the initial contact, the number of contacts, and the incentives (37, 38). The low response rate to our survey is not surprising, because the first contact with potential respondents was not targeted to persons with OA. In addition, the informed consent process (which is unusual for a Harris Interactive survey) may have confused participants or otherwise discouraged participation. One option to improve the overall response rate would be to conduct a screening survey to identify potential study patients and then target the full survey to those who report a diagnosis or other characteristic of interest. This option, however, could be prohibitively expensive in many cases. It also does not solve the underlying bias problem completely, because the total number of “true” potential respondents would remain unknown. This area of survey methodology should be explored further.

Finally, we eliminated 7–8% of responses that appeared to be inconsistent with the rating scale. In general, this exclusion had the predictable effect of increasing ratings for less severe states and decreasing ratings for more severe health states, and of reducing the amount of nontransitivity in the more severe OA health state–side effect scenarios. A multivariable logistic regression analysis suggested that younger respondents, women, respondents with higher education, and respondents with lower ratings of recent OA pain were more likely to report consistent rating patterns (data not shown). The degree to which inconsistency in rating patterns was indicative of a simple reversal of the rating scale end points rather than a deeper lack of understanding about the rating exercise is unclear. Our Internet-based survey also excluded individuals without access to a computer, a fact reflected in the relatively high education and income levels reported by survey participants. Ratings tended to increase with higher education and income, although these characteristics were not significant predictors in our analysis. This trend could have occurred for a variety of reasons. Individuals with these characteristics may be employed in occupations that are less affected by OA. Income and education could also be correlated with improved access to medical treatments and other factors that can reduce the impact of OA on quality of life.

The Web-based format of our survey provided a timely assessment of health status and needs across a large number of patients. Understanding patients' experiences with OA and its treatment is important for planning health services and programs to prevent or address OA-related disability. Although we found that OA severity had a larger impact on health-state ratings than did side effect severity, it is difficult to infer how patients would value OA treatments that vary by disease and side effect severity, because we did not ask this question directly. Future research should attempt to determine whether patients prefer reductions in OA- related pain and disability to improvements in treatment side effect profiles.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:

We thank Kevin J. Anstrom, PhD, and Kevin P. Weinfurt, PhD, for helpful comments throughout the study, and Damon Seils for editorial assistance and manuscript preparation.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:
  • 1
    Centers for Disease Control and Prevention. Prevalence of disabilities and associated health conditions among adults: United States, 1999. MMWR Morb Mortal Wkly Rep 2001; 50: 1205.
  • 2
    Centers for Disease Control and Prevention. Prevalence of self-reported arthritis or chronic joint symptoms among adults: United States, 2001. MMWR Morb Mortal Wkly Rep 2002; 51: 94850.
  • 3
    Brownson RC, Remington PL, Davis JR. Chronic disease epidemiology and control. Washington, DC: American Public Health Association; 1998.
  • 4
    Centers for Disease Control and Prevention. Public health and aging: projected prevalence of self-reported arthritis or chronic joint symptoms among persons aged >65 years: United States, 2005-2030. MMWR Morb Mortal Wkly Rep 2003; 52: 48991.
  • 5
    Groeneveld PW, Lieu TA, Fendrick AM, Hurley LB, Ackerson LM, Levin TR, et al. Quality of life measurement clarifies the cost-effectiveness of Helicobacter pylori eradication in peptic ulcer disease and uninvestigated dyspepsia. Am J Gastroenterol 2001; 96: 33847.
  • 6
    Lawrence WF, Fryback DG, Martin PA, Klein R, Klein BE. Health status and hypertension: a population-based study. J Clin Epidemiol 1996; 49: 123945.
  • 7
    Maetzel A, Krahn M, Naglie G. The cost effectiveness of rofecoxib and celecoxib in patients with osteoarthritis or rheumatoid arthritis. Arthritis Rheum 2003; 49: 28392.
  • 8
    Tosteson AN, Gabriel SE, Kneeland TS, Moncur MM, Manganiello PD, Schiff I, et al. Has the impact of hormone replacement therapy on health-related quality of life been undervalued? J Womens Health Gend Based Med 2000; 9: 11930.
  • 9
    Tilden DP, Schulz M, Davey P, Noble I. A utility valuation study assessing treatment outcomes in osteoarthritis. International Society for Pharmacoeconomics and Outcomes Research 8th International Meeting; 2003 May 1821; Arlington, VA.
  • 10
    Panel on Cost Effectiveness in Health and Medicine. Cost Effectiveness in Health and Medicine. New York: Oxford University Press; 1996.
  • 11
    Ernst ME, Iyer SS, Doucette WR. Drug-related problems and quality of life in arthritis and low back pain sufferers. Value Health 2003; 6: 518.
  • 12
    Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988; 15: 183340.
  • 13
    Stephenson NL, Herman JA. Pain measurement: a comparison using horizontal and vertical visual analogue scales. Appl Nurs Res 2000; 13: 1578.
  • 14
    Bennett KJ, Torrance GW, Moran L, Smith F, Goldsmith CH. Health state utilities in knee replacement surgery: the development and evaluation of McKnee. J Rheumatol 1997; 24: 1796805.
  • 15
    Chang J, Kauf T, Reed S, Friedman J, Omar M, Kahler K, et al. Productivity loss as an indicator of unmet needs for osteoarthritis patients. American College of Rheumatology/Association of Rheumatology Health Professionals 67th Annual Scientific Meeting; 2003 October 2428; Orlando, FL.
  • 16
    Chang J, Weinfurt K, Kauf TL, Curtis LH, Reed SD, Friedman JY, et al. Quantitative and qualitative effects of test accuracy on willingness to pay. 25th Annual Meeting of the Society for Medical Decision Making; 2003 October 1922; Chicago, IL.
  • 17
    Groessl EJ, Kaplan RM, Cronan TA. Quality of well-being in older people with osteoarthritis. Arthritis Rheum 2003; 49: 238.
  • 18
    American College of Rheumatology Subcommittee on Osteoarthritis Guidelines. Recommendations for the medical management of osteoarthritis of the hip and knee: 2000 update. Arthritis Rheum 2000; 43: 190515.
  • 19
    Jordan KM, Arden NK, Doherty M, Bannwarth B, Bijlsma JW, Dieppe P, et al. EULAR Recommendations 2003: an evidence based approach to the management of knee osteoarthritis: report of a task force of the Standing Committee for International Clinical Studies Including Therapeutic Trials (ESCISIT). Ann Rheum Dis 2003; 62: 114555.
  • 20
    Solomon NA, Glick HA, Russo CJ, Lee J, Schulman KA. Patient preferences for stroke outcomes. Stroke 1994; 25: 17215.
  • 21
    Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB. Recommendations of the panel on cost-effectiveness in health and medicine. JAMA 1996; 276: 12538.
  • 22
    Giesler RB, Ashton CM, Brody B, Byrne MM, Cook K, Geraci JM, et al. Assessing the performance of utility techniques in the absence of a gold standard. Med Care 1999; 37: 5808.
  • 23
    Torrance GW. Preferences for health outcomes and cost-utility analysis. Am J Manag Care 1997; 3Suppl:S8–20.
  • 24
    Torrance G. An interview on utility measurement: interview by Bernie O'Brien. J Rheumatol 1995; 22: 12002.
  • 25
    Badia X, Monserrat S, Roset M, Herdman M. Feasibility, validity and test-retest reliability of scaling methods for health states: the visual analogue scale and the time trade-off. Qual Life Res 1999; 8: 30310.
  • 26
    Bakker C, Rutten M, Santen-Hoeufft M, Bolwijn P, van Doorslaer E, Bennett K, et al. Patient utilities in fibromyalgia and the association with other outcome measures. J Rheumatol 1995; 22: 153643.
  • 27
    Coons SJ, Rao S, Keininger DL, Hays RD. A comparative review of generic quality-of-life instruments. Pharmacoeconomics 2000; 17: 1335.
  • 28
    Gallagher EJ, Bijur PE, Latimer C, Silver W. Reliability and validity of a visual analog scale for acute abdominal pain in the ED. Am J Emerg Med 2002;20: 28790.
  • 29
    Joyce CR, Zutshi DW, Hrubes V, Mason RM. Comparison of fixed interval and visual analogue scales for rating chronic pain. Eur J Clin Pharmacol 1975; 8: 41520.
  • 30
    Shikiar R, Bresnahan BW, Stone SP, Thompson C, Koo J, Revicki DA. Validity and reliability of patient reported outcomes used in psoriasis: results from two randomized clinical trials. Health Qual Life Outcomes 2003; 1: 53.
  • 31
    Stavem K. Quality of life in epilepsy: comparison of four preference measures. Epilepsy Res 1998; 29: 2019.
  • 32
    Peeters HR, LJongen-avrencic M, Bakker CH, Vreugdenhil G, Breedveld FC, Swaak AJ. Recombinant human erythropoietin improves health-related quality of life in patients with rheumatoid arthritis and anaemia of chronic disease; utility measures correlate strongly with disease activity measures. Rheumatol Int 1999; 18: 2016.
  • 33
    Goossens ME, Vlaeyen JW, Rutten-van Molken MP, van der Linden SM. Patient utilities in chronic musculoskeletal pain: how useful is the standard gamble method? Pain 1999; 80: 36575.
  • 34
    Farrar JT, Young JP Jr, LaMoreaux L, Werth JL, Poole RM. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 2001; 94: 14958.
  • 35
    Agency for Healthcare Research and Quality. Medical expenditure panel survey. URL: http://www.meps.ahrq.gov/whatis.htm.
  • 36
    Morrell RW, Mayhorn CB, Bennett J. A survey of World Wide Web use in middle-aged and older adults. Hum Factors 2000; 42: 17582.
  • 37
    Cook C, Heath F, Thompson RL. A meta-analysis of response rates in web- or internet-based surveys. Educ Psychol Meas 2004; 60: 82136.
  • 38
    Harris Interactive. An interview with our panel expert. URL: http://www.harrisinteractive.com/tech/HI_Panel_Overview.pdf.

Appenxix A:

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:
Table  . AAPPENDIX A. HEALTH STATE AND SIDE EFFECT PROFILE NARRATIVES*
 Description
  • *

    OA = osteoarthritis; SE = side effect; NA = not applicable.

OA health state 
 OA1: None/mildYou experience no or minor pain in your knees, hips, or hands.
 You do not take any medication for your pain.
 You do not experience stiffness (a sensation of decreased ease in moving a joint).
 You have no limitation in your daily activity.
 OA2: MildYou experience mild pain in your knees, hips, or hands.
 You take over-the-counter medication for your pain sometimes.
 You experience mild stiffness (a sensation of decreased ease in moving a joint) when you first wake up in the morning, and it lessens after you have been sitting, lying down or resting later in the day.
 You cannot perform heavy household chores without difficulty.
 OA3: ModerateYou experience moderate pain in your knees, hips, or hands.
 You take prescription medication for your pain sometimes.
 You experience moderate stiffness (a sensation of decreased ease in moving a joint) when you first wake up in the morning, and it lessens after you have been sitting, lying down or resting later in the day.
 You cannot walk up or down the stairs without difficulty.
 OA4: SevereYou experience severe pain in your knees, hips, or hands.
 You take prescription medication for your pain regularly.
 You require a cane to help you get around.
 You experience severe stiffness (a sensation of decreased ease in moving a joint) when you first wake up in the morning, and it lessens after you have been sitting, lying down or resting later in the day.
 You cannot sit down or get up from chairs without difficulty.
 OA5: ExtremeYou experience extreme pain in your knees, hips, or hands.
 You take prescription medication for your pain regularly, and your doctor has suggested joint replacement for your condition.
 You require a wheel chair or someone to help you move around.
 You experience extreme stiffness (a sensation of decreased ease in moving a joint) when you first wake up in the morning, and it lessens after you have been sitting, lying down or resting later in the day.
 You cannot rest in bed without difficulty.
Side effect profile
 SE0: No side effectsNA
 SE1: Mild, treatable with over-the-counter  medicationYou suffer from general stomach upset, which consists of mild heartburn, nausea, and loss of appetite, caused by your osteoarthritis medication. These symptoms were bothersome enough that you consulted with your doctor, who suggested treatment with over-the-counter medication.
 SE2: Mild, treatable with prescription  medicationYou suffer from general stomach upset, which consists of mild heartburn, nausea, and loss of appetite, caused by your osteoarthritis medication. These symptoms were bothersome enough that you consulted with your doctor, who suggested treatment with prescription medication.
 SE3: Ulcer, treatable with prescription  medicationYou suffer from stomach upset, which consists of mild heartburn, nausea, vomiting, and loss of appetite, caused by your osteoarthritis medication. These symptoms were severe enough that, when you consulted with your doctor, he/she performed a procedure and discovered a stomach ulcer. The ulcer is not bleeding and can be treated with prescription medication.
 SE4: Bleeding ulcer, treatable with  prescription medicationYou suffer from stomach upset, which consists of mild heartburn, nausea, vomiting, and loss of appetite, caused by your osteoarthritis medication. These symptoms were severe enough that, when you consulted with your doctor, he/she performed a procedure and discovered a bleeding stomach ulcer. The ulcer can be treated with prescription medication.
 SE5: Bleeding ulcer requiring surgeryYou suffer from stomach upset, which consists of mild heartburn, nausea, vomiting, and loss of appetite, caused by your osteoarthritis medication. These symptoms were severe enough that, when you consulted with your doctor, he/she performed a procedure and discovered a bleeding stomach ulcer which needs to be treated with surgery.
Benchmark health state 
 Wearing denturesYou must wear dentures to eat and speak clearly.
 You experience minor inconveniences like having to clean the dentures regularly.
 Congestive heart failureYou have swelling in your ankles, legs or abdomen.
 You experience shortness of breath when you exercise.
 You feel tired and sometimes dizzy after walking up the stairs.
 You experience chest pain after carrying heavy objects.
 You need to take medicine regularly.

Appenxix B:

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Appenxix A:
  9. Appenxix B:
Table  . RESULTS OF GENERALIZED LINEAR MODEL ESTIMATION OF HEALTH STATE–SIDE EFFECT SCENARIO RATINGS
ParameterEstimate (95% confidence interval)P
  • *

    Compared with Hispanic.

  • Compared with college or postgraduate.

  • Compared with OA1. Osteoarthritis (OA) health states are as follows: OA1 = none/mild; OA2 = mild; OA3 = moderate; OA4 = severe; OA5 = extreme.

  • §

    Compared with SE0. Side effect (SE) profiles are as follows: SE0 = no side effects; SE1 = mild side effects, treatable with over-the-counter medication; SE2 = mild side effects, treatable with prescription medication; SE3 = ulcer, treatable with prescription medication; SE4 = bleeding ulcer, treatable with prescription medication; SE5 = bleeding ulcer requiring surgery.

Intercept94.90 (90.42, 99.38)<0.001
Male−2.04 (−2.93, −1.15)<0.001
Age−0.01 (−0.05, 0.03)0.51
Recent OA pain level−0.71 (−0.89, −0.53)<0.001
No recent gastrointestinal event−0.20 (−1.05, 0.66)0.65
Years since OA diagnosis−0.006 (−0.03, 0.02)0.60
Employed0.15 (−0.69, 1.00)0.72
Race/ethnicity*  
 White−0.12 (−3.78, 3.54)0.95
 African American−1.17 (−6.29, 3.94)0.65
 Other/unknown−0.34 (−4.44, 3.76)0.87
Education  
 High school or lower−0.74 (−2.00, 0.53)0.25
 Some college−0.18 (−1.05, 0.69)0.69
 Other/unknown−3.37 (−6.92, 0.18)0.06
Income  
 <$50,0000.26 (−1.09, 1.60)0.71
 $50,000 to $100,0000.33 (−1.07, 1.73)0.64
 > $100,0001.08 (−0.63, 2.78)0.22
OA health state  
 OA2−30.34 (−32.51, −28.18)<0.001
 OA3−38.55 (−40.62, −36.48)<0.001
 OA4−59.43 (−61.64, −57.21)<0.001
 OA5−69.27 (−71.34, −67.20)<0.001
Side effect profile§  
 SE1−2.66 (−5.22, −0.10)0.04
 SE2−5.24 (−7.89, −2.59)0.00
 SE3−9.01 (−11.61, −6.42)<0.001
 SE4−12.12 (−14.70, −9.53)<0.001
 SE5−17.93 (−20.57, −15.29)<0.001
OA–side effect scenario  
 OA3/SE1−1.42 (−5.09, 2.26)0.45
 OA3/SE2−1.08 (−4.79, 2.63)0.57
 OA3/SE30.18 (−3.45, 3.82)0.92
 OA3/SE42.36 (−1.31, 6.02)0.21
 OA3/SE53.38 (−0.33, 7.09)0.07
 OA4/SE13.55 (−0.18, 7.28)0.06
 OA4/SE25.21 (1.50, 8.92)0.01
 OA4/SE36.61 (2.92, 10.31)<0.001
 OA4/SE49.19 (5.46, 12.92)<0.001
 OA4/SE511.78 (8.03, 15.54)<0.001
 OA5/SE15.96 (2.37, 9.56)<0.01
 OA5/SE25.51 (1.88, 9.13)<0.01
 OA5/SE311.54 (7.91, 15.18)<0.001
 OA5/SE414.31 (10.67, 17.94)<0.001
 OA5/SE517.56 (13.92, 21.23)<0.001