Association of health-related quality of life with dual use of prescription and over-the-counter nonsteroidal antiinflammatory drugs




Inadequate prescription therapy pain management, lack of doctor-patient communication about over-the-counter (OTC) medications, and easy accessibility of OTC medications may contribute to patients using more than 1 medication to manage pain. It is well established that taking multiple nonsteroidal antiinflammatory drugs (NSAIDs) can lead to serious gastrointestinal problems. Little is known about whether use of more than 1 NSAID (i.e., dual use) is related to patient self-reported outcomes, specifically health-related quality of life (HRQOL). We hypothesized that dual use of NSAIDs would be associated with reduced HRQOL.


Patients from a managed care organization who filled ≥1 NSAID prescription over a 6-month period were eligible for a telephone interview focusing on NSAID use, which included the Short Form 12 (SF-12) Health Survey. Dual use was defined as taking 2 NSAIDs, either prescription or OTC, at least twice weekly during the past month. A multivariable linear regression model examined the association between dual use and the Physical Component Summary score (PCS-12) from the SF-12.


Dual use was associated with lower PCS-12 scores indicating poorer HRQOL, after controlling for clinical and demographic factors.


Patients may self-manage their pain to improve their daily activities by taking more than 1 NSAID. However, by attempting to obtain symptom relief, patients may be putting themselves at risk for complications. Providers are likely unaware of patients' risk. Future research should evaluate the causal factors contributing to dual use.


Prescription nonsteroidal antiinflammatory drugs (NSAIDs) are used to reduce pain and inflammation, particularly for musculoskeletal conditions such as arthritis. Arthritis affects one-third of all adults (1) and it is expected that by 2020, 60 million Americans will have musculoskeletal conditions and ∼20% will experience resultant limitations in their daily activities (2). Consequently, NSAIDs are one of the most commonly prescribed medications worldwide (3).

In addition to prescription NSAIDs, there are many over-the-counter (OTC) NSAIDs available for self-management of arthritis and pain symptoms (4). Among individuals with arthritis, almost 30% report daily use of OTC NSAIDs (5). In fact, arthritis-related conditions are the leading cause of disability in the US (6). In addition, the availability of OTC analgesics appears to be continually increasing (7).

Given the availability of both prescription and OTC NSAIDs, patients may take both concomitantly (dual use). One potential reason for the therapeutic duplication is for patients to obtain pain relief (8), which may be due to inadequate pain treatment by health care providers. Another potential reason for therapeutic duplication is because patients may be unaware that an OTC NSAID they are taking is in the same therapeutic class as their current prescription NSAID. Patients' unawareness is coupled with a lack of communication between patients and providers about OTC use (9). Thus, providers may also be unaware that the patient is taking an additional NSAID medication.

Unfortunately, it is well established that taking multiple NSAIDs can lead to adverse events, including serious gastrointestinal events (10–12). Thus, patients are at risk for safety problems. Although adverse events are well known, it is not known whether there is a relationship between dual NSAID use and important patient outcomes, such as health-related quality of life (HRQOL). Because dual use of prescription and/or OTC NSAIDs may serve as a marker for inadequate pain control and poor daily functioning, it may be associated with worse HRQOL.

The objective of the current study was to determine the prevalence of dual prescription and/or OTC NSAID use and to evaluate whether dual use was associated with reduced HRQOL. We hypothesized that patients who dually used a prescribed and/or OTC NSAID would have worse physical and mental HRQOL than patients who used a single prescribed NSAID or single OTC NSAID.


Patient selection and eligibility.

Study participants were drawn from a randomized controlled trial that tested an intervention to improve physician adherence to recommended NSAID toxicity monitoring guidelines (13). The patients were all members of a large regional managed care organization in the state of Alabama. A total of 424 patients were eligible for the current study based on the following criteria: 1) the patient filled ≥1 NSAID prescription (traditional NSAID or cyclooxygenase 2 selective inhibitor [COX-2]) between February 2002 and August 2002 according to pharmacy claims, 2) the patient's outpatient medical record was obtained and abstracted, 3) a telephone number was available, and 4) the patient was able to communicate in English. This study was conducted at the University of Alabama at Birmingham (UAB) and was approved by the UAB Institutional Review Board (IRB) and the Durham Veterans Affairs Medical Center IRB.

Survey administration.

The survey was administered to the 424 eligible patients during November 2003 using a computer-assisted telephone interview (CATI) protocol. The CATI protocol followed the Behavioral Risk Factor Surveillance System methodology (14), which included calling a respondent up to a total of 15 times during 3 separate calling time frames before declaring that the potential respondent was unavailable to complete the survey.


Using the CATI, information on sociodemographic and socioeconomic characteristics, current and past use of prescription NSAIDs, and current use of OTC NSAIDs was obtained, and the 12-Item Short Form Health Survey (SF-12), version 2 (15) was administered. To obtain information on past prescription NSAID use, all patients' prescribed NSAIDs were captured from the pharmacy database and patients' medical records. To reduce recall bias, patients were asked specifically whether they were currently taking or had previously taken the prescription NSAID(s) captured in the pharmacy database or abstracted from the patient's medical record for arthritis and/or joint pain. After obtaining information on use of specified NSAIDs from pharmacy records and medical records, participants were then asked open-ended questions to elicit whether they were currently taking or had taken over the past year any other prescribed NSAID medications not recorded in the pharmacy or medical record for arthritis and/or joint pain. To obtain information on use of OTC NSAIDs, patients were asked whether they had taken a generic or brand name OTC NSAID for arthritis or joint pain in the past month and, if so, the frequency of use (i.e., 1, 2, or 3 times per day, a couple of times per week, once per week, or a couple of times per month).

The SF-12, version 2 measured generic HRQOL (15). The SF-12 is a well-validated instrument that evaluates a person's health status and calculates 2 summary scores, the Physical Component Summary (PCS-12) and Mental Component Summary (MCS-12). For the general US population, the mean ± SD for the PCS-12 and MCS-12 is 50 ± 10.

Medical record abstractions.

Medical chart abstractions were performed to identify patients' medical comorbidities, number of physician visits, and to determine past receipt of prescription NSAIDs for use in subsequent CATI questions. The CATI survey responses for each patient were linked with their medical record data that were collected for the original randomized controlled trial. The medical record data from a 7-month period were abstracted (February 2002 to August 2002) (13). Trained abstractors obtained variables from the medical records. The interrater reliability of abstractors for key variables associated with NSAID use was 97%.

Definitions for dual and non-dual users.

Dual users were defined as patients who 1) reported currently taking a prescription NSAID for arthritis and/or joint pain and over the past month reported taking an OTC NSAID for arthritis and/or joint pain daily (1, 2, or ≥3 times per day) or a couple times per week, or 2) reported currently taking 2 prescription NSAIDs for arthritis and/or joint pain, or 3) over the past month reported taking ≥2 OTC NSAIDs for arthritis and/or joint pain daily (1, 2, or ≥3 times per day) or a couple times per week. All other patients reporting current use of only 1 prescription NSAID or only 1 OTC NSAID daily or a couple times per week were designated as non-dual users.

Statistical analysis.

All analyses were performed using Intercooled STATA 9.2 for Windows (StataCorp, College Station, TX). First, we examined overall univariate population characteristics. Second, we examined the bivariate associations of NSAID dual use with the main independent variables, the PCS-12 and MCS-12. Not finding a significant association between dual use and the MCS-12, we used multivariable linear regression to examine the independent association of dual use with the PCS-12. Because only 32.5% of the original sample was available for the final analysis, we conducted bivariate analyses of patients who were included in the final analysis (n = 138) and those who were excluded (n = 286) using variables that were available from all patients from the medical record abstraction (i.e., age, sex, modified Charlson comorbidity score, and number of office visits) to determine if there was any observable bias between patients included in the final analysis and those excluded from the final analysis. As noted in Table 1, there were no important or significant differences between the 2 groups. Because we found no observable bias, we did not pursue multivariable bias-reduction techniques, such as Heckman modeling (16). For the regression equation, we entered the covariates in one block representing prespecified potentially important confounders. Thus, the regression model adjusted for age, race, sex, marital status, education, income, employment, and modified Charlson comorbidity score. Participants' age, race, sex, income, education, and employment were obtained from responses on the patient survey. The participants' comorbidities were obtained from the medical chart abstraction and a modified Charlson comorbidity index score (17) was calculated using an SAS macro (SAS Institute, Cary, NC), which includes 19 categories of comorbidity using the International Classification of Diseases, Ninth Revision codes that are adjusted based on risk of 1-year mortality.

Table 1. Clinical characteristics of included and excluded patients taking nonsteroidal antiinflammatory drugs in a regional managed care organization, 2002
 Excluded (n = 286)Included (n = 138)P
  • *

    10-month period.

Age, mean years55.255.90.662
Female sex, %54.655.80.808
No. of physician visits*
Modified Charlson comorbidity index, mean*


A flowchart of the participants that were included in the data analysis is shown in Figure 1. There were 64 participants who could not be reached using the telephone number provided in their medical record, 57 were contacted on numerous occasions but were consistently unavailable to ask for their consent to complete the survey, and 57 were ineligible (e.g., reported not taking an NSAID, language barrier). Of the 246 participants who were contacted and eligible, 64 (26%) refused to participate and 182 completed the survey, resulting in a response rate of 74%. A total of 41 (23%) participants were excluded from the analyses because they reported not currently taking a prescription or OTC NSAID or because they took an OTC NSAID once per week or a couple of times per month. Additionally, 3 participants did not answer the SF-12 items. This left 138 participants for analysis.

Figure 1.

Flow diagram of participants excluded from the study. NSAID = nonsteroidal antiinflammatory drug; OTC = over-the-counter; SF-12 = Short Form 12.

The sociodemographic and disease characteristics of the participants are presented in Table 2. The mean age of the participants was 56 years and a majority of the participants were women. Thirty-six (26%) participants were dual users. Almost 20% of the participants were nonwhite and approximately one-quarter reported that their total household income was less than $25,000. Approximately half of the participants had rheumatoid arthritis or osteoarthritis and almost one-quarter had chronic back pain. There was only one significant difference between dual and non-dual users for demographic and clinical characteristics: non-dual users were significantly more likely (P < 0.001) to have arthritis or osteoarthritis (57%) than dual users (24%).

Table 2. Demographic and clinical characteristics of dual and non-dual NSAID users in a regional managed care organization, 2002 (n = 138)*
  • *

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

  • Reported currently taking a prescription NSAID and an over-the-counter NSAID daily or a couple times per week during the past month or 2 prescription or 2 over-the-counter NSAIDs.

  • 10-month period.

  • §

    Rheumatoid arthritis and osteoarthritis.

Age, mean ± SD years56 ± 12
Female sex77 (56)
Dual NSAID users36 (26)
Married93 (67)
Nonwhite25 (18)
No college education105 (76)
Unemployed68 (49)
Annual household income <$25,00031 (23)
No. of physician visits, mean ± SD3.6 ± 2.9
Modified Charlson comorbidity index, mean ± SD score0.6 ± 0.9
Arthritis§64 (46)
Chronic back pain30 (22)

Frequencies of prescription and OTC NSAID use for dual and non-dual users in the study population are shown in Table 3. Approximately 80% of dual users were designated as such by using a prescription NSAID (including COX-2 specific inhibitors) and an OTC NSAID. The other 20% were designated dual users by taking 2 prescription or 2 OTC NSAIDs. For non-dual users, a vast majority reported current use of a prescription NSAID.

Table 3. Frequencies of prescription and OTC NSAIDs for dual and non-dual users in a regional managed care organization, 2002 (n = 138)*
 No. (%)
  • *

    OTC = over-the-counter; NSAID = nonsteroidal antiinflammatory drug; RX = prescription.

Dual users (n = 36) 
 RX NSAID and OTC NSAID29 (81)
 2 RX NSAIDs4 (11)
 2 OTC NSAIDs3 (8)
Non-dual users (n = 102) 
 RX NSAID86 (84)
 OTC NSAID16 (16)

The unadjusted PCS-12 and MCS-12 summary scores for the dual and non-dual users are presented in Figure 2. Dual users had significantly lower PCS-12 scores than non-dual users (P = 0.05). Dual users had similar mean ± SD PCS-12 scores (35 ± 11.7) compared with participants with rheumatoid arthritis and osteoarthritis in clinical trials (33 ± 9.0 versus 34 ± 9.3) (18). There was no significant difference between the MCS-12 summary scores for dual and non-dual users. The mean ± SD MCS-12 summary scores for dual users (47 ± 9.5) and non-dual users (46 ± 8.4) were also similar to those reported for participants with rheumatoid arthritis (48 ± 11) and osteoarthritis (50 ± 10.5) (18).

Figure 2.

Physical Component Summary (PCS-12) and Mental Component Summary (MCS-12) scores for dual and non-dual nonsteroidal antiinflammatory drug (NSAID) users from a managed care organization, 2002. Differences between the dual and non-dual NSAID users were evaluated using an independent t-test. * P = 0.05. ** P = 0.88. Solid bars = dual users; shaded bars = non-dual users.

We found no important or significant multicollinearity among all independent variables included in the final regression model evaluating the association between dual NSAID use and the PCS-12. More specifically, the mean variance inflation factor was 1.24 (range 1.08–1.64). Results for the multivariable linear regression analysis for the PCS-12 are presented in Table 4. After controlling for other factors, dual use was a significant independent predictor of lower PCS-12 scores (standardized β = −4.51, P = 0.05). Other significant factors associated with lower PCS-12 scores were being unemployed and having a higher modified Charlson comorbidity index score.

Table 4. Multivariable linear regression analysis with the physical functioning scale of the Short Form 12 as the outcome in a population using analgesic medication from a regional managed care organization, 2002 (n = 126)*
 β (95% CI)
  • *

    95% CI = 95% confidence interval.

  • Reported currently taking a prescription nonsteroidal antiinflammatory drug (NSAID) and an over-the-counter (OTC) NSAID daily or a couple times per week during the past month or 2 prescription or 2 OTC NSAIDs.

  • P < 0.05.

  • §

    P < 0.001.

  • 10-month period.

Dual users (vs. non-dual users)−4.51 (−9.01, −0.002)
Age−0.08 (−0.12, 0.27)
Nonwhite (vs. white)0.80 (−4.43, 6.03)
Female (vs. male)−2.67 (−6.75, 1.41)
Married (vs. not married)0.55 (−3.95, 5.06)
No college education (vs ≥4 years of college)−1.89 (−6.47, 2.69)
Unemployed (vs. employed)−8.12 (−12.71, −3.53)§
Annual household income <$25,000 (vs. income >$25,000)1.13 (−4.66, 6.91)
Modified Charlson comorbidity index−4.94 (−7.13, −2.75)§


In our study of prescription NSAID users from a large regional managed care organization, we found that 26% of the population reported dual use of a prescription and OTC NSAID, or the use of 2 prescription NSAIDs or 2 OTC NSAIDs. Dual use was also found to be an independent factor associated with worse scores on the PCS-12 of the SF-12 after controlling for several patient and disease characteristics. There was no significant univariate relationship between the MCS-12 summary scores for dual and non-dual users.

Little is known about patients who take multiple analgesics, whether prescribed or OTC. Determination of multiple medication use from pharmacy claims or medical records is difficult because OTC medications are typically not available (19–21). We lack well-accepted standardized measures to capture OTC medications. Thus, there are very few studies that have evaluated the use of OTC NSAIDs (22–24). In our study, we surveyed patients by telephone to obtain this important information.

In addition, OTC medication use is often not discussed at clinic visits (9). Taking high doses of NSAIDs, including the use of more than 1 NSAID, is an important patient safety concern (10, 12, 25). The Joint Commission's 2007 National Patient Safety Goal 8 recommends obtaining and keeping a current complete list of a patient's medications to reduce adverse drug–drug interactions (26). Having all medication use available would likely identify patients using ≥2 NSAIDs. The increased awareness may lead to better communication between the patient and provider about the appropriate use of NSAIDs.

Analyses from the Centers for Disease Control and Prevention Third National Health and Nutrition Examination Survey (NHANES-III) revealed that 26% of respondents were dually using aspirin and ibuprofen (23). A national survey conducted by Roper Starch World Proprietary Telephone Research Center in conjunction with the American Gastroenterological Association found that 38% of individuals were dually using a prescription and OTC NSAID (5). We found the same rate (26%) as the NHANES-III study, despite some differences in the definition of dual use between our study and the NHANES-III study.

We did not find an association between the MCS-12 and the dual and non-dual use categories. Our MCS-12 summary scores are similar to those reported in other rheumatoid arthritis and osteoarthritis populations (18), which are similar to the US norm. Although depression and anxiety are a common comorbid condition in patients who have arthritis and/or pain (27, 28), it is plausible that the MCS-12 was not sensitive enough to detect differences in mental health symptoms between the dual and non-dual users in this study. Additionally, physical health problems, such as those measured on the PCS-12 including pain, limitations in daily activities, and perceived current health status, may be more directly related to dual NSAID use than are mental health symptoms. Our small sample size prevented us from analyzing the 8 subscales (e.g., bodily pain, physical functioning) of the PCS-12 and MCS-12 to determine specific physical and mental health associations between the dual and non-dual users (29). Future work will involve the use of a larger sample size to detect differences in the subscales, in addition to the use of more sensitive measures of depression, anxiety, and pain problems.

Our finding that dual users had lower PCS-12 summary scores than non-dual users may have resulted from patients seeking relief from inadequate clinical pain management (30–32). Inadequate pain management is highlighted by the American Pain Society's attempts to make health care professionals more aware of the importance of assessing and managing pain by declaring that pain should be viewed as the fifth vital sign (33). It is possible that participants dually using NSAIDs in the current study had not received adequate pain relief from their first NSAID, and therefore began using an additional NSAID to better control their pain. Due to concerns about recall bias, our survey did not ask patients about the dose or frequency of use of their prescription NSAID or OTC analgesic medication. Thus, it is unknown whether dual users were receiving the maximum therapeutic dose for their prescription or OTC NSAID medication.

It is possible that our current lower PCS-12 score results in dual NSAID users are confounded by patients' level of pain. In other words, dual use could be a marker for higher levels of pain. Unfortunately, we are not able to confirm with the current data set whether the current dual NSAID users indeed have more pain than non-dual NSAID users. If our supposition is correct, then research is needed about how to better educate and enable patients to discuss their level of pain with their doctors and how to encourage health professionals to ask about patients' pain at each visit. Furthermore, if dual NSAID use is a marker for higher pain scores, then it becomes even more important for providers to address patients' symptoms, in addition to assessing whether they are using NSAIDs unsafely.

Despite the important findings in this study, there are limitations. First, the study was cross-sectional and therefore the results can only establish associations rather than causality between dual use and HRQOL. Considering the paucity of research conducted in dual NSAID use, it is appropriate to begin research in evaluating associations between dual use and important patient-based outcomes such as HRQOL. Second, there is not an agreed upon standard for defining dual NSAID use. We used a concrete, conservative definition of dual use, including self-report of current prescription use and at least twice per week use of an OTC NSAID. Third, this study was conducted with patients in a managed care organization and may not be generalizable to other populations. However, our prevalence rates of dual use were consistent with larger studies that have been conducted with random samples of the US population (5, 23). Furthermore, future research will be conducted with NSAID users who are not part of managed care organizations to determine whether our current findings can be replicated, and extended or accounted for by other outcome measures, such as pain intensity and interference. We may not have captured all relevant factors associated with dual use, such as perceived control over arthritis and pain symptoms. Finally, although only ∼43% of eligible patients completed the survey, we found no differences between patients included and those excluded from our final analysis, suggesting no observable response bias.

Our study had several strengths. This is the first study to look beyond potential patient characteristics associated with dual use and instead the association with HRQOL. We used several sources of data to ascertain NSAID use. The pharmacy data, in addition to the medical records, were used to identify all potential prescribed NSAIDs that the patients had potentially taken over the past year, so that specific NSAIDs could be mentioned by name to promote patient recall. The telephone survey captured whether patients were currently taking the prescription NSAIDs identified in the pharmacy and medical record data. We were able to reduce recall bias by asking patients explicitly whether they had taken in the past or were currently taking the prescribed NSAID medications. In addition, we were able to obtain patients' use of OTC NSAIDs, which is not often collected in research studies.

In conclusion, to our knowledge this is the first study to establish an association between dual NSAID use and physical HRQOL. Dual users may take multiple NSAIDs to reduce their pain, and may knowingly or unknowingly do this despite NSAID safety risks. Future research should focus on establishing causal factors of dual NSAID use and evaluating the best methods to identify dual users, who may be at a higher risk for NSAID-related toxicity. If inadequate pain management is a contributing factor, then identifying more appropriate pain management strategies that are less likely to cause adverse events is needed. Adequate pain management may have the potential to reduce dual use, improve patient symptoms, including physical functioning, and reduce patient safety problems.


Dr. Kovac 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. Kovac, Saag, Allison.

Acquisition of data. Kovac, Curtis.

Analysis and interpretation of data. Kovac, Saag, Curtis, Allison.

Manuscript preparation. Kovac, Curtis, Allison.

Statistical analysis. Kovac, Allison.