SEARCH

SEARCH BY CITATION

Keywords:

  • Health disparities;
  • Patient-doctor communication;
  • Nonsteroidal antiinflammatory drugs

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Objective

Nonsteroidal antiinflammatory drugs (NSAIDs) are commonly used and frequently lead to serious adverse events. Little is known about NSAID-related ethnic/racial disparities. We focused on differences in patient NSAID risk awareness, patient-doctor NSAID risk communication, and NSAID risk-avoidance behavior.

Methods

We performed a cross-sectional analysis of survey data from the Alabama NSAID Patient Safety Study. Eligible patients were ≥65 years old and currently taking prescription NSAIDs (Rx NSAIDS). Generalized linear latent and mixed models accounted for nesting of patients within physicians.

Results

Of all 404 participants, 32% were African American and 73% were female. The mean ± SD age was 72.8 ± 7.5 years, and 64% reported an annual household income <$20,000. African American patients were less likely than white patients to recognize any risk associated with over-the-counter (OTC) NSAIDs (13.3% versus 29.3%; P = 0.001) and Rx NSAIDs (31.3% versus 49.6%; P = 0.001), report that their doctor discussed possible NSAID-related gastrointestinal problems (38.0% versus 52.4%; P = 0.007), and take medications to reduce ulcer risk (30.5% versus 50.2%; P = 0.001). Patients with lower income and education reported significantly less risk awareness for OTC and Rx NSAIDs. Racial/ethnic differences persisted after adjusting for multiple confounders.

Conclusion

In this community-based study of low income elderly individuals receiving NSAIDs, we identified important racial/ethnic differences in risk awareness, communication, and behavior. Additional efforts are needed to promote safe NSAID use and reduce ethnic/racial disparities.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Nonsteroidal antiinflammatory drugs (NSAIDs), one of the most commonly prescribed medication classes used to treat inflammatory, arthritic, and musculoskeletal conditions (1, 2), frequently lead to serious adverse events (AEs) (3). Studies have attributed more than 100,000 hospitalizations, 16,500 deaths, and cost in excess of $500 million to gastrointestinal (GI) complications (4, 5). These estimates of morbidity and mortality do not include emerging concerns about adverse cardiovascular events, such as myocardial infarction, hypertension, and kidney disease (6–8). Furthermore, up to 40% of NSAID prescriptions are written for individuals over the age of 60, the age group at highest risk for NSAID-associated AEs (9).

Because NSAIDs frequently lead to preventable AEs, patients need to understand the associated risks, and health care professionals need to effectively communicate risk information. For example, medications used for GI prophylaxis, such as proton-pump inhibitors, may decrease the risk of GI bleeding from traditional NSAIDs (10). However, failed risk communication resulting in poor medication adherence may negate the potential benefit of GI prophylaxis. Katz et al found that the full spectrum of potential adverse reactions to NSAIDs is rarely disclosed (11). Furthermore, risk communication and behavior are complex phenomena that vary with patient and doctor sociodemographic profiles and health beliefs (12).

Given the extensive literature on health disparities by race/ethnicity and socioeconomic status (13), it is reasonable to assume that similar problems exist for NSAID-related risk awareness, patient-doctor risk communication, and risk behavior. Therefore, we hypothesized that such disparities might exist within the population of patients taking NSAIDs. To investigate our hypotheses, we analyzed data from the Alabama NSAID Patient Safety Study, which provided a population of community-based patients who were currently taking NSAIDs.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Participants.

Participants were selected from the practices of 48 primary care physicians participating in the Alabama NSAID Patient Safety Study and completed an hour-long survey that ascertained NSAID use patterns, race/ethnicity, comorbidity, and sociodemographic data. All physicians were in private, community-based practice in Alabama. All patients met the following criteria: 1) established patient of participating primary care physician, 2) age ≥65 years, 3) currently taking prescription (Rx) NSAIDs, 4) willing to provide contact information and informed consent and complete a telephone interview, and 5) self-reported African American or white race/ethnicity. The study was approved by the University of Alabama at Birmingham Institutional Review Board.

Survey administration.

Interested patients completed a screening survey during a regularly scheduled clinic visit with the study physician. Screening surveys were collected by physician office personnel and mailed to the coordinating center. The screening questionnaire ascertained current NSAID use, age, and contact information. All eligible patients were subsequently contacted by telephone and invited to participate in this study, and further data were obtained using computer-assisted telephone interview protocols. Data were entered directly into the computer by the interviewers. The computer software checked for logical consistency and out-of-range errors. Patients completing the survey were given a $20 gift card. Interviewers underwent formal training with certification of competency before beginning data collection.

Of all 556 eligible patients, 478 were contacted. Of the 478 contacted, 55 refused to be interviewed and 13 terminated the interview early or had missing data. Therefore, the overall response rate was 73.7% (410 of 556), the cooperation rate was 85.8% (410 of 478), the contact rate was 86.0% (478 of 556), and the refusal rate was 9.9% (55 of 556) (3). Additional patients who indicated that their race/ethnicity was not African American or white (n = 6) were also excluded, resulting in a sample size of 404 for this study.

Variable definitions.

All variables were derived from patient self-report. The main study end points were yes/no responses to interview questions focusing on 1) over-the-counter (OTC) NSAID risk awareness (“Do you know of any problems or risks connected with taking over-the-counter NSAIDs?”), 2) Rx NSAID risk awareness (“Do you know of any problems or risks connected with taking prescription NSAIDS?”), 3) patient-doctor communication about potential GI risk from NSAID use (“Did your doctor talk with you about the risk of stomach or intestinal problems that could be related to your prescription NSAID medication?”), and 4) risk-avoiding behavior by taking GI prophylaxis (“Have you taken any medicines to protect your stomach, such as Prevacid, Prilosec, Pepcid, Zantac, or Tagamet in the last 4 weeks?”). Socioeconomic position was represented by education and income. For the main analyses, total annual household income was entered as a dichotomous variable indicating values <$20,000 or ≥$20,000. Education was entered as a dichotomous variable indicating whether the patient had at least a 2-year college degree.

Main statistical analyses.

First, we examined univariate and summary statistics for all study variables. Second, we compared clinical and socioeconomic characteristics of African American and white participants using the chi-square test for categorical variables and the t-test for continuous variables. Third, we examined each study end point separately by race/ethnicity, education, and annual household income. Fourth, we calculated unadjusted and adjusted odds ratios from separate multivariable regression models for each study end point.

Before building the multivariable models, we examined potential independent variables for multicollinearity (14). Among all independent variables included in the final models, we found an average variance inflation factor of 1.08, suggesting an absence of important multicollinearity. Likewise, principal components analysis with variance proportion decomposition did not suggest important multicollinearity.

All models accounted for the clustering of patients (level 1) within physicians (level 2) and were derived using the generalized linear latent and mixed model procedure implemented by Stata (StataCorp, College Station, TX) (15). Maximum likelihood estimation was based on adaptive quadrature (16). We calculated unadjusted odds ratios to quantify the association of race/ethnicity, education, and income with each end point from regression models containing only a single independent variable. Next, we added age, sex, clinical variables, and socioeconomic variables to each regression model. Because our goal was to estimate the independent effect of race/ethnicity rather than to develop predictive models, we entered covariates as a block instead of using a stepwise variable selection procedure. We examined interactions of race/ethnicity, income, and education with each end point by stratified bivariate comparisons and by entering product terms in the regression equations. To examine the influence of extreme observations, we calculated deleted standardized Pearson's residuals (level 1), deleted deviance residuals (level 1), deleted standardized residuals from empirical Bayesian estimates (level 2), and Cook's distance (level 2). Models were examined both with and without extreme observations, revealing no important changes in the magnitude or significance of the major findings.

For the full multivariable models, we calculated the residual intraclass correlation coefficient (ICC) to represent the proportion of total variability attributable to the physician level (level 2) after adjusting for patient-level covariates. The ICC calculation assumed an unobserved, latent continuous outcome (17). Bias-corrected and accelerated confidence intervals for each ICC were based on ∼2,000 nonparametric bootstrap replications with resampling at the cluster level (18, 19).

Secondary statistical analyses.

We performed a subanalysis including only the 24 physicians who had data for both African American and white patients. For each physician, we determined the proportion of African American and white patients with a positive response on each of the 4 dependent variables. Next, for each physician, we calculated the difference in performance for African American and white patients on each dependent variable. Finally, we reported overall differences between African American and white patients for each dependent variable by averaging across physicians, recognizing that the small subsample size resulted in loss of power.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Among all 404 study patients from 40 Alabama counties, 32% were African American, and the rest were white. The mean ± SD number of African American patients per physician practice was 2.7 ± 4.2 (range 0–19). For all patients, the mean ± SD age was 72.8 ± 7.5 years, with no significant difference by race/ethnicity. Compared with white patients, African American patients were almost twice as likely to report diabetes; otherwise, comorbidity was approximately balanced by race/ethnicity (Table 1). Alcohol and tobacco use did not differ significantly by race/ethnicity. There were significant differences in socioeconomic status by race/ethnicity, with African American patients more likely to have an annual household income <$20,000 and less likely to have completed at least 2 years of college.

Table 1. Patient characteristics by race/ethnicity (Alabama NSAID Patient Safety Study, 2006)*
 African American (n = 129)White (n = 275)P
  • *

    Values are the percentage unless otherwise indicated. NSAID = nonsteroidal antiinflammatory drug.

Age, mean years72.972.70.793
Female sex79.571.60.094
Comorbidity   
 Arthritis92.293.10.760
 Heart failure14.016.00.595
 Diabetes41.921.00.001
 Gastrointestinal bleed20.714.70.150
 Liver disease2.32.90.737
 Kidney disease2.34.00.391
 Mean no. of comorbidities (maximum 6)1.71.60.467
Medications in the past month   
 Prednisone2.34.70.249
 Warfarin2.39.80.007
Health behavior in the past month   
 Alcohol13.217.10.315
 Cigarettes7.810.90.322
Socioeconomic   
 Annual household income <$20,00078.356.80.001
 Not filled medication in past year because of cost41.719.50.001
 Inadequate income to meet basic needs53.617.30.001
 Completed at least 2 years of college24.237.80.008
Insurance   
 Medicare91.388.40.357
 Commercial55.082.50.001
 Medicaid16.35.80.001
Rural residence25.821.20.312

African American patients were less likely than white patients to 1) recognize any risk associated with OTC and Rx NSAIDs, 2) report that their physician discussed possible NSAID-related GI problems, and 3) take GI prophylaxis (Figure 1). Patients with an annual household income <$20,000 and those with <2 years of college experience were less likely to be aware of OTC and Rx NSAID risk (Figures 2 and 3). There were no significant differences in patient-reported risk communication with the physician or use of GI prophylaxis by income or education.

thumbnail image

Figure 1. Patient-reported nonsteroidal antiinflammatory drug (NSAID) risk awareness, patient-doctor NSAID risk communication, and NSAID risk behavior by race/ethnicity. All data are from a telephone survey of 404 patients who were currently taking NSAIDs (Alabama NSAID Patient Safety Study, 2006). † Medication for gastrointestinal (GI) prophylaxes included prescription (Rx) and over-the-counter (OTC) H2 histamine blockers and proton-pump inhibitors. MD = physician.

Download figure to PowerPoint

thumbnail image

Figure 2. Patient-reported nonsteroidal antiinflammatory drug (NSAID) risk awareness, patient-doctor NSAID risk communication, and NSAID risk behavior by annual household income. All data are from a telephone survey of 404 patients who were currently taking NSAIDs (Alabama NSAID Patient Safety Study, 2006). † Medication for gastrointestinal (GI) prophylaxes included prescription (Rx) and over-the-counter (OTC) H2 histamine blockers and proton-pump inhibitors. MD = physician.

Download figure to PowerPoint

thumbnail image

Figure 3. Patient-reported nonsteroidal antiinflammatory drug (NSAID) risk awareness, patient-doctor NSAID risk communication, and NSAID risk behavior by education. All data are from a telephone survey of 404 patients who were currently taking NSAIDs (Alabama NSAID Patient Safety Study, 2006). † Medication for gastrointestinal (GI) prophylaxes included prescription (Rx) and over-the-counter (OTC) H2 histamine blockers and proton-pump inhibitors. MD = physician.

Download figure to PowerPoint

From the multivariable models in Table 2, African American race/ethnicity was significantly associated with lower odds of 1) risk awareness for both OTC and Rx NSAIDs, 2) reporting risk communication with the physician, and 3) engaging in risk-avoiding behavior with GI prophylaxis, even after adjustment for age, sex, clinical variables, and socioeconomic variables. From the unadjusted models, low income was associated with lower risk awareness for OTC and Rx NSAIDs, and low education was associated with lower risk awareness for OTC NSAIDs. After adjustment for race/ethnicity and other clinical variables, the associations of income and education with the dependent variables were no longer significant. ICCs revealed that only a small fraction of the variability in the dependent variables occurred at the physician level (level 2). All interaction terms from the multivariable models were not significant.

Table 2. Unadjusted and adjusted odds ratios (95% confidence intervals) for patient-reported NSAID risk awareness, communication, and behavior by race/ethnicity, income, and college education (Alabama NSAID Patient Safety Study, 2006) (n = 404)*
 African AmericanIncomeCollege educationICC
Estimate95% CI
  • *

    Values are the odds ratio (95% CI) unless otherwise indicated. Estimates were derived from generalized linear latent and mixed models accounting for the clustering of patients (level 1) within physicians (level 2). Separate models were developed for each outcome (understand OTC NSAID risk, understand Rx NSAID risk, physician discussed NSAID GI risk, and taking GI prophylaxis). Unadjusted models included only a single independent variable. Independent variables for adjusted models included race, age, sex, use of prednisone, use of warfarin, comorbidity count, income <$20,000, at least 2-year college degree, and rural residence. ICCs assumed an underlying latent continuous outcome; bias-corrected and accelerated confidence intervals were generated from 2,000 nonparametric bootstrap replications. NSAID = nonsteroidal antiinflammatory drug; ICC = intraclass correlation coefficient; 95% CI = 95% confidence interval; OTC = over-the-counter; Rx = prescription; GI = gastrointestinal; MD = physician.

Understand OTC NSAID risk     
 Unadjusted0.35 (0.19–0.65)0.52 (0.31–0.87)2.05 (1.26–3.34)  
 Adjusted0.42 (0.21–0.85)0.69 (0.37–1.28)1.56 (0.84–2.88)0.060.00–0.12
Understand Rx NSAID risk     
 Unadjusted0.51 (0.30–0.84)0.50 (0.31–0.80)1.46 (0.93–2.30)  
 Adjusted0.50 (0.28–0.88)0.52 (0.30–0.89)1.41 (0.82–2.40)0.030.00–0.04
MD discussed NSAID GI risk     
 Unadjusted0.52 (0.32–0.84)0.76 (0.49–1.20)1.11 (0.72–1.70)  
 Adjusted0.51 (0.29–0.89)0.89 (0.54–1.50)1.28 (0.76–2.15)0.030.00–0.06
Taking GI prophylaxis     
 Unadjusted0.43 (0.27–0.68)0.82 (0.51–1.31)1.00 (0.65–1.54)  
 Adjusted0.43 (0.26–0.73)0.95 (0.57–1.58)1.13 (0.67–1.88)0.010.00–0.04

Each of the 4 end points by income and education stratified by race/ethnicity are presented in Table 3. For white patients, higher income was associated with increased risk awareness for OTC NSAIDs and Rx NSAIDs. A similar nonsignificant trend was seen for the association between education and risk awareness for OTC and Rx NSAIDs in white patients. For African American patients, there were no significant differences in risk awareness for Rx NSAIDs by income. However, African American patients with greater education reported more risk awareness for OTC NSAIDs, with a similar, but nonsignificant trend for Rx NSAIDs.

Table 3. Patient-reported NSAID risk awareness, communication, and behavior: unadjusted percentage for income and education, stratified by race/ethnicity (Alabama NSAID Patient Safety Study, 2006)*
 Annual household incomeCollege education
<$20,000 (n = 226)≥$20,000 (n = 128)PNone (n = 262)≥2 years (n = 132)P
  • *

    See Table 2 for definitions.

Understand OTC NSAID risk      
 African American14.412.00.7558.630.00.003
 White22.235.90.02026.334.70.148
Understand Rx NSAID risk      
 African American31.532.00.95930.140.00.315
 White40.761.20.00246.154.90.162
MD discussed GI NSAID risk      
 African American40.036.00.71738.336.70.873
 White48.556.30.23351.853.90.733
Taking GI prophylaxis      
 African American30.036.00.56729.930.00.982
 White51.551.50.99851.848.00.551

Among the physicians who had data for both African American and white patients (n = 24), we observed findings similar to those reported for the main analyses. For these physicians, the mean absolute positive response rates for white patients were 20% greater for OTC NSAID risk awareness, 8% greater for discussion of Rx NSAID risk awareness, 25% greater for discussion of NSAID risk with the physician, and 15% greater for use of GI prophylaxis.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

In our population of mainly elderly NSAID users with lower socioeconomic status, we found important disparities in risk awareness, communication, and behavior by race/ethnicity, income, and education. Patients who were African American, had an annual household income <$20,000, and had <2 years of college education were significantly less likely to recognize any risk associated with OTC and Rx NSAIDs. These vulnerable patients were also less likely to report that their physician discussed NSAID risk and less likely to take GI prophylaxis. This finding persisted even after multivariate adjustment for income and education and other important medical and demographic factors. Although we had limited power to examine interactions between race/ethnicity, income, and socioeconomic status, greater education appeared to be more tightly linked to improved risk awareness in African American patients than in white patients. Even among white patients, rates of risk awareness, communication, and behavior showed important room for improvement.

The low ICCs revealed that most of the variability in risk awareness, communication, and behavior resulted from factors at the patient level rather than the physician level, with race/ethnicity being the most powerful patient-level explanatory variable. Subanalyses of physicians with data for both African American and white patients suggested that the observed disparities did not result from most African American patients seeing a few physicians with distinctive practice patterns.

Our findings stand in contrast to the magnitude and trends of many other disparities reported in the current literature. Although African American patients receive pervasively lower quality of care and have worse outcomes for a wide array of conditions, many disparities have started to narrow. For example, the 2005 National Healthcare Disparities Report published by the Agency for Healthcare Research and Quality found that for African Americans, care is improving on 58% of 46 core quality measures and 100% of 13 core access measures (20). Although we did not examine changes over time, we found absolute disadvantages in NSAID risk awareness, communication, and behavior for African American patients, ranging from 14% to 16%.

Results from a 2002 public opinion survey by the National Consumer League (NCL) about OTC analgesic use were similar to our findings but did not address disparities by race/ethnicity, income, or education (21). This survey found that patients often did not read medication labels properly, often took more than the recommended dose, and did not fully understand how other medications and/or alcohol might lead to potential adverse reactions. Furthermore, the NCL found that few adults taking OTC NSAIDs could identify side effects associated with the stomach (21%), liver (21%), or kidney (12%). Only 39% reported having discussed interactions with other OTC or prescription medications with a health professional, and 29% talked with their physician about the risk of GI bleeding or ulcers or the potential for liver or kidney damage.

Given the magnitude, consequences, and preventable nature of NSAID AEs, we urgently need better interventions to promote safer use. One potential solution may be improved doctor-patient communication (22). In time-constrained outpatient visits for patients with multiple comorbidities, competing clinical demands may forestall risk communication, especially when such activities are not directly reimbursed (23). Lack of trust in the medical system may also lead a patient to not disclose OTC NSAID use, and building a relationship of trust is especially important for minority patients (24).

Alternatively, physicians may discuss NSAID risk, but patients may not adequately process the message. Ashton et al proposed that such discordant risk communication occurs when the patient and physician do not share health beliefs or explanatory illness models (25). According to the Ashton Racial and Ethnic Patient-Physician Communication Model, the patient's cultural background and the institutional context of health care delivery exert profound influences on patient-doctor communication. For example, in our study, patients may assume that OTC medications are completely safe; however, it is obvious to the physician that ibuprofen carries the same risk whether purchased with or without a prescription. Up to 40% of patients receiving a prescription NSAID concomitantly take an OTC NSAID (26), but physicians frequently do not inquire about OTC medication use during routine office visits (27).

In addition to doctor-patient communication, communication with the pharmacist at the point of sale might also improve risk awareness and behavior (28, 29). Pharmacists, who often have access to the patient's complete medical record, might easily detect unsafe prescribing or therapeutic duplication, especially if automated software were in place. Additional point-of-sale strategies include placing OTC NSAID medications behind the counter to promote interaction with the pharmacist and implementing an easy-to-understand visual coding scheme with different colors representing medications with similar side-effect profiles, thus giving the consumer important information at the time of sale (30).

Public awareness campaigns may also lead to improved NSAID risk awareness and behavior. Direct-to-consumer advertising has been highly correlated with patient prescription requests and physicians' NSAID prescribing patterns (31) and, perhaps, could be converted into a vehicle for the public good (32, 33). Likewise, other interventions to increase health literacy may actually reduce disparities. In support of this hypothesis, our stratified analysis suggested a disproportionate benefit of education for African American patients.

Several limitations should be noted. First, all analyses were cross-sectional, limiting inference of cause and effect. Second, we did not observe clinical end points, so it is difficult to know how many participants experienced AEs, such as GI bleeding or cardiovascular morbidity. Third, we could not determine if patients were taking GI medications for prophylaxis versus diagnosed GI disease. However, our multivariable models included adjustment for a history of GI bleeding. In addition, we were able to use a rich data set to adjust for several factors associated with AEs (such as age, use of prednisone, use of warfarin, and comorbidity) (34). Fourth, we did not assess duration of NSAID treatment, which could potentially influence risk awareness and communication. It is reasonable to assume that patients with more recent NSAID prescriptions would report less risk awareness and communication, thus biasing the overall end points downward. However, because our study focused on racial/ethnic disparities, significant bias of the main findings toward the null would require that these associations operate differently in African Americans and whites. Fifth, we could not ensure that study patients were representative of the patient population in each practice, but the overall response rate among the patients who agreed to participate was high. Sixth, our dichotomous dependent variable did not capture the full range of complexity associated with patient-doctor communication; even so, the finding that most African American patients reported no NSAID-related risk communication is alarming and should motivate interventions to reduce observed disparities. Future work is needed to develop more comprehensive instruments for measuring NSAID and other therapeutic risk communication. Finally, our study population only included 2 racial/ethnic groups from a single state. However, our work was conducted within a patient population of low income and low education, which may increase generalizability to other similarly vulnerable populations.

NSAID toxicity exerts a tremendous health burden on the US population, with elderly persons and those with increased comorbidity being at greatest risk. Our study suggests that African Americans, the poor, and those without higher education are even more vulnerable. Given these results, we urgently need additional research on interventions that can effectively promote safer NSAID prescribing and use.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Dr. Fry had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Ray, Cobaugh, Saag, Curtis, Allison.

Acquisition of data. Ray, Allison.

Analysis and interpretation of data. Fry, Ray, Weissman, Kiefe, Shewchuk, Saag, Curtis, Allison.

Manuscript preparation. Fry, Ray, Cobaugh, Weissman, Kiefe, Shewchuk, Allison.

Statistical analysis. Fry, Weissman, Kiefe, Shewchuk, Allison.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

We acknowledge Cynthia Johnson, Linda Jones, Damien Larkin, and Carolyn Hollinghead for their help with data collection, data entry, and statistical analyses.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  • 1
    Paulose-Ram R, Hirsch R, Dillon C, Losonczy K, Cooper M, Ostchega Y. Prescription and non-prescription analgesic use among the US adult population: results from the third National Health and Nutrition Examination Survey (NHANES III). Pharmacoepidemiol Drug Saf 2003; 12: 31526.
  • 2
    Wilcox CM, Cryer B, Triadafilopoulos G. Patterns of use and public perception of over-the-counter pain relievers: focus on nonsteroidal antiinflammatory drugs. J Rheumatol 2005; 32: 221824.
  • 3
    The American Association for Public Opinion Research standard definitions: final dispositions of case codes and outcome rates for surveys. 4th ed. Lenexa (KS): AAPOR; 2006.
  • 4
    Chan AT, Manson JE, Albert CM, Chae CU, Rexrode KM, Curhan GC, et al. Nonsteroidal antiinflammatory drugs, acetaminophen, and the risk of cardiovascular events. Circulation 2006; 113: 157887.
  • 5
    Singh G. Recent considerations in nonsteroidal anti-inflammatory drug gastropathy. Am J Med 1998; 105: S318.
  • 6
    Epstein M. Non-steroidal anti-inflammatory drugs and the continuum of renal dysfunction. J Hypertens 2002; 20 Suppl 6: S1723.
  • 7
    Gaziano JM. Nonnarcotic analgesics and hypertension. Am J Cardiol 2006; 97: 106.
  • 8
    Hermann M, Ruschitzka F. Coxibs, non-steroidal anti-inflammatory drugs and cardiovascular risk [published erratum appears in Intern J Med 2006;36:799]. Intern Med J 2006; 36: 30819.
  • 9
    Gabriel SE, Jaakkimainen L, Bombardier C. Risk for serious gastrointestinal complications related to use of nonsteroidal anti-inflammatory drugs: a meta-analysis. Ann Intern Med 1991; 115: 78796.
  • 10
    Naesdal J, Brown K. NSAID-associated adverse effects and acid control aids to prevent them: a review of current treatment options. Drug Saf 2006; 29: 11932.
  • 11
    Katz JN, Daltroy LH, Brennan TA, Liang MH. Informed consent and the prescription of nonsteroidal antiinflammatory drugs. Arthritis Rheum 1992; 35: 125763.
  • 12
    McComas KA. Defining moments in risk communication research: 1996–2005. J Health Commun 2006; 11: 7591.
  • 13
    Groman R, Ginsburg J, and the American College of Physicians. Racial and ethnic disparities in health care: a position paper of the American College of Physicians. Ann Intern Med 2004; 141: 22632.
  • 14
    Belsley DA, Kuh E, Welsch RE. Regression diagnostics: identifying influential data and sources of collinearity. New York: John Wiley & Sons; 1980.
  • 15
    Rabe-Hesketh S, Skrondal A. Multilevel and longitudinal modeling using Stata. College Station (TX): Stata; 2005.
  • 16
    Rabe-Hesketh S, Skrondal A, Pickles A. Reliable estimation of generalized linear mixed models using adaptive quadrature. Stata J 2002; 2: 121.
  • 17
    Snijders TA, Bosker RJ. Multilevel analysis: an introduction to basic and advanced multilevel modeling. Thousand Oaks (CA): Sage; 1999.
  • 18
    Davison AC, Hinkley DV. Bootstrap methods and their application. Cambridge: Cambridge University; 1997.
  • 19
    Efron B, Tibshirani RJ. An introduction to the bootstrap. New York: Chapman & Hall; 1993.
  • 20
    Agency for Healthcare Research and Quality. 2005 National Healthcare Disparities Report. Rockville (MD): AHQR; 2005
  • 21
    National Consumers League Survey: over-the-counter pain medications. URL: http://www.nclnet.org/otcpain/index.htm. 2003.
  • 22
    Edwards A, Elwyn G, Wood F, Atwell C, Prior L, Houston H. Shared decision making and risk communication in practice: a qualitative study of GPs' experiences. Br J Gen Pract 2005; 55: 613.
  • 23
    Boyd CM, Darer J, Boult C, Fried LP, Boult L, Wu AW. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA 2005; 294: 71624.
  • 24
    Boulware LE, Cooper LA, Ratner LE, LaVeist TA, Powe NR. Race and trust in the health care system. Public Health Rep 2003; 118: 35865.
  • 25
    Ashton CM, Haidet P, Paterniti DA, Collins TC, Gordon HS, O'Malley K, et al. Racial and ethnic disparities in the use of health services: bias, preferences, or poor communication? J Gen Intern Med 2003; 18: 14652.
  • 26
    Pincus T, Swearingen C, Cummins P, Callahan LF. Preference for nonsteroidal antiinflammatory drugs versus acetaminophen and concomitant use of both types of drugs in patients with osteoarthritis. J Rheumatol 2000; 27: 10207.
  • 27
    Sleath B, Rubin RH, Campbell W, Gwyther L, Clark T. Physician-patient communication about over-the-counter medications. Soc Sci Med 2001; 53: 35769.
  • 28
    De Young M. Research on the effects of pharmacist-patient communication in institutions and ambulatory care sites, 1969–1994. Am J Health Sys Pharm 1996; 53: 127791.
  • 29
    Schommer JC, Wiederholt JB. A field investigation of participant and environment effects on pharmacist-patient communication in community pharmacies. Med Care 1995; 33: 56784.
  • 30
    Allison JJ, Saag KG. Unbalanced regulation of over-the-counter analgesics: the lesser of two evils? Pharmacoepidemiol Drug Saf 2006; 15: 4101.
  • 31
    Spence MM, Teleki SS, Cheetham TC, Schweitzer SO, Millares M. Direct-to-consumer advertising of COX-2 inhibitors: effect on appropriateness of prescribing. Med Care Res Rev 2005; 62: 54459.
  • 32
    Hollon MF. Direct-to-consumer advertising: a haphazard approach to health promotion. JAMA 2005; 293: 20303.
  • 33
    Mebane FE. The importance of news media in pharmaceutical risk communication: proceedings of a workshop. Pharmacoepidemiol Drug Saf 2005; 14: 297306.
  • 34
    Singh G, Triadafilopoulos G. Epidemiology of NSAID induced gastrointestinal complications. J Rheumatol 1999; 56: 1824.