Socio-Demographic Differences in Risk Information Seeking Sources for Non-Steroidal Anti-inflammatory Drugs (NSAIDS)

Abstract

Objectives:

Non-steroidal anti-inflammatory drugs (NSAIDs) are frequently prescribed for musculoskeletal pain and inflammatory conditions. A better understanding of patient information seeking behavior can help bridge the gap between patient knowledge and health care resources. This study examines the primary sources of NSAID risk information and the associations with patient socio-demographic factors.

Methods:

A cross-sectional survey analysis of patients on prescription NSAIDs (n=220) seen by primary care physicians in Alabama. Bivariate and multivariable, multinomial logistic regression analyses were conducted to evaluate the associations among primary NSAID risk information sources used with patient socio-demographic factors.

Results:

The primary patient source of information on NSAID risks was physician (57.3%), followed by internet (16.8%), pharmacist (16.4%), and other sources, such as nurses and family/friends (9.6%). Compared to people who use the internet as a primary source of NSAID risk information, patients who were Black/African-American (p=0.002) and 65 years of age or older (p=0.009) were more likely to use a physician. Older patients were also more likely to use a pharmacist (p=0.008) than the internet. In contrast, females (p=0.032) were less likely to use the pharmacist compared to the internet (p=0.032).

Conclusions:

Patients obtain information from a variety of sources, but primarily from health care providers. While the internet is a fast growing source of health information, socio-demographic disparities in internet use for seeking information exist. Health care providers should be aware of their patient preferences for information sources on medication risks to meet the age, race, and gender need differences of all patients.

1. Introduction

In 2014, the Pew Internet and American Life Project reported that 87% of U.S. adults use the internet. Of those who used the internet, 72% searched for health information online and, more specifically, one in three U.S. adults used the internet search for medical-related conditions [1, 2]. In general, the internet is used by patients to seek information to prepare for a doctor’s visit [3], validate or refute health information obtained from other sources [4, 5], communicate via email with clinicians [3], and learn about what other patients have experienced [6]. Patients also use the internet for the purpose of obtaining more information about their specific medical condition [7, 8] such as poisonings [9] regardless of whether there was an intention to obtain a self-cure or not [9], and to more fully understand their medications [6, 8, 10–12] or treatment regimens [13].

Physicians or other health care providers are generally a key resource for information about serious health episodes [1, 2]. However, the Pew study found that second to health care providers is the use of the internet.[1, 2] Patients who use the internet to seek information, rate it as a trusted source ranked ahead of family, friends, and the media [14]. Patients who have chronic ailments are more likely than healthier patients to consult the internet for health information [15], a finding mirrored in a study that matched health status to the use of internet for information seeking behavior [16].

Fundamental to patient empowerment is accessible and accurate health information that is easily read, understood, and actionable. Recognizing the power of the internet, the 2012 Institute of Medicine (IOM) report Best Care at Lower Cost – The Path to Continuously Learning Health Care in America stated that one of the many opportunities that exist for remedying the “wasteful and unsustainable” U.S. health care system is to empower patients. The report concluded that there exists a potential for patients to be active stakeholders and cognizant participants in their health care [17]. The IOM Report also suggested that the power of computers, harnessing information scientifically, mobile computing, the internet, and improving connectivity can lead to better patient-clinician communication [17]. Furthermore, the IOM Report viewed the internet as a powerful force in transforming the nation’s health care system into one that achieves greater value.

Although the internet has been heralded to bring many improvements in patient care, great disparities in the use of the internet exist [18–20]. These disparities are especially pronounced across racial/ethnic [19, 21], age [20], and socioeconomic status groups [22]. Research has examined the larger array of internet information seeking behaviors [23], but have largely been focused on specific language groups [24], age groups [11, 25–27], psychiatric health [6, 10], specific cancers [28–30], physicians seeking information [30], specific care settings (e.g., inpatient or outpatient) [13, 27], specific medical conditions [13, 25], or the credibility and accuracy of medical information available online [7, 9, 12, 32].

Studies that have evaluated online medication information seeking behavior have focused on non-prescription medication [33], psychosocial health [6, 10], and non-adult populations [11]. In contrast, this paper explores online medication risk information sources among medically underserved and vulnerable populations with a specific focus on nonsteroidal anti-inflammatory drugs (NSAIDs). Specifically, we first describe information sources (i.e., the internet, physicians, pharmacists, or other) that patients use when seeking risk information about prescription NSAIDs. A second objective identifies variations in the socio-demographic factors (i.e., race, age, gender, education, income sufficiency, and health literacy) associated with medication risk information seeking from traditional sources compared to the internet.

1.1. Conceptual Framework

Non-steroidal anti-inflammatory drugs (NSAIDs) in both prescription and over-the-counter formulations are frequently prescribed for musculoskeletal pain and inflammatory conditions. Because NSAIDs are available as both prescription and over-the-counter (OTC) products, it is commonly assumed that over-the-counter NSAIDs are safer, less toxic, or have fewer risks than prescription NSAIDs. However, all NSAIDs, regardless of prescription or OTC status, are associated with the risk of adverse events related to the gastrointestinal, cardiovascular, and renal systems especially when used chronically or inappropriately. This gap in patient knowledge of NSAIDs and their associated risk is especially apparent among vulnerable populations where health literacy is a challenge [41, 44]. A better understanding of patient use of various types and sources of information about NSAID risks may help bridge the gap between patient knowledge and health care resources.

The evolving use of different medication information sources reflects the growing importance of tailoring health care information for vulnerable population subgroups such as minorities and those with low income, who are disproportionately affected by low health literacy and may have difficulty processing medically-complex information [48]. Individuals in these socioeconomic groups approach health sources of information differently [4, 15, 16]. As such, clinicians who are responsible for their care should be aware of how different socioeconomic groups seek medication information on the internet differently from other groups, so as to be better able to respond to their inquiries about clinical diagnosis, condition, treatment plan, and medication.

Today’s internet-driven prescription medication information searches [34, 35] call for policymakers to better understand the social and demographic antecedents of patient information seeking behavior. Akin to Schmidt and Spreng’s study on external factors affecting consumer pre-purchase information search [36], we argue that perceived ability to search and perceived benefits of searching online are directly linked to prescription medication information seeking behavior [37–40].

Information processing abilities and motivation to process information are prerequisites for actual information seeking behavior [36]. We posit that individuals in vulnerable populations seek medication risk information differently. Vulnerable groups such as females, older, and minorities as well as those with less education, inadequate health literacy, and inadequate income would be less likely to use the internet as their primary source of prescription medication information [4, 15, 16]. In part, this may be explained by the lack of confidence by these groups in their ability to search, socioeconomic barriers, and their perceptions of the value of searching online.

2. Methods

2.1. Study Design

This study uses baseline, cross-sectional data from patients participating in a group randomized clinical trial (i.e., the Alabama Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) Patient Safety Study: Disparities in Risk Awareness and Communication Project) and has previously been described in detail [49]. Data for this study was collected from March 2011 through November 2011. The study was reviewed and approved by the University of Alabama at Birmingham Institutional Review Board.

2.2. Setting and Procedure

Patients were recruited from 41 Alabama physician practices that are geographically representative of the state of Alabama. Patient eligibility was determined by self-report information on patient exit cards left at physician practices that included: 1) being 19 years of age or older; 2) currently taking a prescription NSAID or over the counter ibuprofen/naproxen as recommended by their physician; and 3) consent and willingness to complete a telephone survey that lasted about 30 minutes. There were a total of 373 patients that met study eligibility and were reachable by phone. Of those eligible participants, 259 patients consented to complete the baseline telephone survey, for a participation rate of 69%. The survey was administered using computer-assisted telephone interview protocols, which performed automatic checks for consistency based on logic and answers that were beyond the range of acceptable responses. Patients were compensated for their time with a $20 gift card. The research staff interviewers were trained and certified before they were allowed to begin collecting data. Patients who were not currently taking an NSAID or missing data on any of the independent and dependent variables of interest were excluded from analysis. The final analytic sample included 220 respondents from 39 practices.

2.3. Measures

The primary dependent variable consisted of source of information used by patients for NSAID problems and risks and measured by the question, “Who/what is your main source of information about problems or risks associated with taking prescription NSAIDs?” Patients were asked to select one from a list consisting of physician, nurse, pharmacist, internet, and others. The other category included nurses, healthcare professionals not explicitly listed, family members, friends, print information, or a combination of listed healthcare professionals not explicitly reported as the main source.

The independent variables selected for inclusion in analytical models were based on prior research and included four patient demographic variables, one socio-economic variable, and an estimate of health literacy. Race was dichotomized as Black or White/other. If a patient reported White or a race other than Black, they were included in the White/other group. Gender was defined as male or female. Age was defined as less than 65 years of age or 65 years or older to contrast older who may be at risk for lower health literacy with younger patients. Education was collapsed to high school or below or some college or more. The dichotomous income sufficiency variable was described as a yes response to the question whether the respondent’s income was sufficient to meet their basic needs for food, housing, clothing, and medical care. Because health literacy is complex and time-consuming to measure in practice, simplified surrogate screening measures have been developed and validated against gold standard measures of health literacy (i.e., Short Test of Functional Health Literacy in Adults (S-TOFHLA) and the Rapid Estimate of Adult Literacy in Medicine (REALM)) [45, 46]. For this study, health literacy was estimated using the one-item screening question, “How confident are you completing medical forms by yourself?” and was dichotomized as adequate or less than adequate using previously validated cut points [45, 46]. “Do not know/Not sure” responses were counted as missing data for demographic characteristics and main NSAID risk information source variables.

2.4. Analysis

Univariate statistics were used to describe the study sample and reported medication information sources. Because there were more than two categories of the dependent variable (i.e., four information sources: physicians, pharmacists, internet, and other), multivariable multinomial logistic regression analyses were used to model the relationship between medication information sources and patient demographics, socioeconomic variables, and health literacy. Specifically, all multivariable models included age, gender, race, income, education, and health literacy variables. The multinomial logistic regression model accounted for the clustering of patients within physician practices using generalized linearized latent and mixed modeling. The adjusted odds ratios and 95% confidence intervals for each of the independent variables were estimated with an a-priori alpha level of statistical significance set to 0.05.

3. Results

3.1. Information Seeking Sources

As shown in Table 1, study participants were predominantly female and had adequate health literacy. Slightly more than half of the sample had at least some college education and income sufficient to meet their basic needs. A minority of the sample was 65 years of age or older and Black/African-American. Slightly more than half of the sample reported using physicians as their main source for NSAIDs risk information followed by the internet, pharmacists, and other sources such as nurses, family or friends. With the exception of gender, all other sociodemographic characteristics were associated with sources of NSAID risk information in bivariate analyses. Regardless of main NSAID risk information source, only a small proportion of subjects discussed either prescription or over-the-counter NSAID risks with either a doctor, pharmacist, nurse, or anyone else.

Table 1 —

Descriptive Statistics of Patient Characteristics Overall by Main Nonsteroidal Anti-inflammatory Drug (NSAID) Risk Information Source Used

	Total %	Main NSAID Risk Information Source				p-value
		Physicians %	Pharmacists %	Other1 %	Internet %
n	220	126 (57.3)	36 (16.4)	21 (9.6)	37 (16.8)
Demographic Characteristics
Female	74.6	70.6	86.1	90.5	67.6	0.063
Age 65 years or older	25.0	28.6	33.3	19.1	8.11	0.042
At least some college	53.6	53.2	33.3	81.0	59.5	0.005
Black/African-American	40.9	50.0	38.9	33.3	16.2	0.003
Adequate health literacy	78.6	77.0	66.7	85.7	91.9	0.051
Sufficient income to meet basic needs	54.6	50.8	41.7	66.7	73.0	0.024
Communication Variables
Talked with Doctor About Prescription NSAID Risks2	23.6	24.6	16.7	23.8	27.0	0.915
Talked with Doctor About Over-the-Counter NSAID Risks2	20.0	23.0	13.9	19.1	16.2	0.594
Talked with Pharmacist About	6.8	4.8	8.3	9.5	10.8	0.544
Prescription NSAID Risks2
Talked with Pharmacist About Over-the-Counter NSAID Risks2	3.2	0.8	8.3	4.8	5.4	0.102
Talked with Nurse About Prescription NSAID Risks2	2.3	2.4	2.8	0	2.7	0.971
Talked with Nurse About Over-the-Counter NSAID Risks2	1.8	0.8	2.8	0	5.4	0.263
Talked with Anyone About Prescription NSAID Risks2	10.9	4.0	13.9	38.1	16.2	<0.001
Talked with Anyone About Over-the-Counter NSAID Risks2	5.0	2.4	11.1	9.5	5.4	0.135

Note:

¹The “Other” category included nurses, healthcare professionals not explicitly listed, family members, and friends, print information, or a combination of listed healthcare professionals not explicitly reported as the main source.

²Discussion Occurred within the Previous 12 months.

3.2. Association of Information Seeking Sources with Patient Characteristics

Table 2 displays the multinomial logistic regression results for each of the independent variables with patient NSAID risk information seeking sources. For the dependent variable, physician, pharmacist and other information sources were contrasted with the internet, which served as the reference outcome.

Table 2 —

Multinomial, Multivariable¹ Logistic Regression Results of the Relationship between Main Nonsteroidal Anti-inflammatory Drug (NSAID) Risk Information Sources and Patient Characteristics (n= 220)

Variable	Physician vs. Internet AOR (CI, p-value)2	Pharmacist vs. Internet AOR (CI, p-value)2	Other3 vs. Internet AOR (CI, p-value)
Female (vs. male)	0.79 (0.32, 1.93) p = 0.599	0.24 (0.07, 0.88) p = 0.032	0.20 (0.0.04, 1.05) p = 0.057
Age >=65 (vs. <55)	5.52 (1.52, 20.00) p = 0.009	6.97 (1.65, 29.40) p = 0.008	3.54 (0.67, 18.72) p = 0.137
Black/African-American (vs. White/Other)	4.62 (1.72, 12,37) p = 0.002	2.21 (0.67, 7.24) p = 0.192	2.56 (0.67, 9.68) p = 0.167
Education  Some College (vs. high school or below)	1.09 (0.48, 2.46) p = 0.838	0.47 (0.17, 1.30) p = 0.14	3.37 (0.90, 12.55) p = 0.071
Adequate health literacy (Less than adequate health literacy)	0.41 (0.10, 1.68) p = 0.216	0.25 (0.05, 1.21) p = 0.085	0.35 (0.05, 2.31) p = 0.275
Sufficient income to meet basic needs	0.61 (0.24, 1.53) P = 0.293	0.44 (0.14, 1.34) p = 0.147	1.08 (0.30, 3.85) p = 0.91

Note:

¹Includes age, gender, race, income, education, and health literacy variables.

²AOR: Adjusted Odds Ratio; CI: Confidence Interval.

³The “Other” category included nurses, healthcare professionals not explicitly listed, family members, and friends, print information, or a combination of listed healthcare professionals not explicitly reported as the main source.

We found support for the gender, age, and race hypotheses in multivariable models, depending on the outcome contrasts. When contrasting physicians as a source of NSAID risk information with the internet, patients who were 65 years of age or older were significantly more likely to use their physician compared to those less than 65 years of age (Adjusted Odds Ratio (AOR) = 5.52, p = 0.009). As well, compared to White patients, Black/African-American patients were also more likely to use their physician as a source of NSAID risk information (AOR=4.62, p=0.002). Gender, education, income sufficiency, and estimated health literacy were not associated with the use of physician as a source of NSAID risk information compared to the internet.

When contrasting pharmacists as a source of NSAID risk information with the internet, patients who were 65 years of age or older were significantly more likely to use their pharmacist compared to those less than 65 years of age (AOR = 6.97, p=0.008). In contrast, compared to male patients, females were significantly less likely to use their pharmacist as a source of NSAID risk information (AOR=0.24, p=0.032). Race, education, income sufficiency, and estimated health literacy were not associated with the use of pharmacist as a source of NSAID risk information compared to the internet. No relationships were observed for any sociodemographic variables and the choice of other sources of NSAID risk information when contrasted with the internet.

4. Discussion

The 2012 Institute of Medicine report Best Care at Lower Cost – The Path to Continuously Learning Health Care in America called for patients to be active stakeholders and cognizant participants in their health care [17]. Actively using the internet as a primary source of risk and side effect information about NSAIDs is one example of greater patient involvement in their care. Harnessing the power of the internet to satisfy patient needs for information can lead to better patient-clinician communication [17]. This study modeled demographic (race/ethnicity, age, and gender), socio-economic (education and income), and health literacy factors associated with patient medication information seeking behavior for NSAIDs risks or problems, with a special focus on internet users as the primary reference group. In recent years, national policies and initiatives have been implemented to enhance patients’ use of electronic health information. However, disparities across socio-demographic groups exist [42].

Our findings suggest that Black/African-American patients tend to rely more on gathering information about the side effects and risks of their NSAIDs prescriptions from their physicians rather than the internet. This finding has important implications for the State of Alabama, where there is a sizeable population of non-Hispanic Black patients. Consequently, physicians who care for Black/African-American patients should not rely on their patients to seek risk information elsewhere as they rely more on the personal encounter. Time should be taken to discuss NSAID risk information during an office visit. If there is not sufficient time during the visit, other active patient educational strategies that inform patients about the potential risks associated with NSAIDs use should be considered such as those from The Alliance for Rational Use of NSAIDs [47].

Our findings also strongly suggest that persons 65 years of age or older are more likely to use physicians and pharmacists for NSAID risk information as opposed to the internet. Not having been raised in a “tech culture”, the elderly are less likely to use electronic sources for health information. On the other hand, females were less likely to rely on their pharmacist compared to the internet as their primary source of NSAID risk information.

The aforementioned findings about primary sources of NSAID risk information have important implications for large sectors of the population. First and foremost, there is still an important need to develop and test interventions that enhance the health care provider-patient encounter to ensure essential information is communicated to patients in a time-constrained environment.

Additional research should also focus on how to increase the proportions of minorities, elderly, and males seeking prescription medication information online and to most effectively use that information. One potential place of intervention is through public library systems. Public libraries are located throughout advantaged and disadvantaged communities, serve as a communal gathering place, have access to the internet and trained professionals who can teach patrons how to use the internet. In collaboration with health care professionals, programs could be developed through libraries to enhance internet use among the disadvantaged groups observed in this research.

Future research could focus on the social dynamics at play in the case of patient-provider interactions. In particular, of worthy examination is whether there is a difference in health outcomes when patients seek health information from different sources. To elaborate, patients from lower socio-economic groups might differ in health outcomes when compared to their more affluent peers precisely because of the source of their health information (health providers or the internet). In addition, one could suggest a training program through community libraries and senior centers on the use of the MedlinePlus system [50]. MedlinePlus is one trusted and comprehensive internet source produced by the National Library of Medicine that should be recommended as a starting point for obtaining accurate and dependable information [50].

The health care provider-patient encounter and use of the internet are not mutually exclusive and must work together. Given that only a minority of patients discussed NSAID risks with some health care provider in the past 12 months, those who rely heavily on the internet for medication and health information need their physician, pharmacist or other health care provider to confirm, refute, or clarify their understanding of information sought online. This is especially true since the internet may contain conflicting, false, and misinformation. Further, health care providers must apply what patients have learned to the specific patient condition and health history. Research studying how physicians interface with highly health literate and internet savvy patients is also necessary.

Our study has a number of potential limitations that should be noted. First, given that the data is cross-sectional, only correlation inferences should be drawn, not causality. Second, our study specifically examined primary sources of prescription NSAIDs risk information. Although is not likely to be different, caution should be exercised when generalizing to other forms of health care information seeking behavior as there might be other factors influencing those searches. Third, although the State of Alabama has a large proportion of individuals from minority, lower income, and vulnerable populations, it may not be representative of other geographic locations. Finally, as with any study, there may be missing variable bias although the variables used were derived from the literature. Factors that have the potential to influence patient characteristics and are inherently difficult to measure, such as culture, religion, and trust in information sources were data were not collected. Despite these limitations, we believe our research provides important insight for physicians and health systems planners on how vulnerable patient populations seek information about NSAIDs online.

Our findings that identify race, age, and gender differences for primary NSAID risk information sources is significant to health care providers and policymakers. Both stakeholders will need to consider these differences to improve NSAID risk communication. In the seminal 2001 Institute of Medicine report Crossing the Quality Chasm, the authors stated that among other ideals, health care in the twenty-first century should be patient-centered. Defining this, they specified that patient-centered care is health care that is responsive to individual patient preferences, needs, and values [43]. Given this main finding, we should expect health care providers to be attentive to potential racial differences between Black/African American and White patients in information seeking behaviors, and tailor their communication to address these differences. These same considerations apply to the older compared to younger patients as well as female patients compared to males. Tailoring health care communication approaches should be encouraged by policymakers.

In conclusion, the findings from this study present important implications for health care providers who are responsible for the health care of vulnerable and minority patients prescribed NSAID medication. Since there are well-known, significant side-effects and risks of NSAIDs, health care providers should pay extra attention to ensure that these concerns are understood by all. To effectively resolve these differences, there needs to be interventions at the provider and patient level. The form of these interventions to optimize health care provider patient behavior should be the focus of future research.

Highlights.

• Patients use internet search for medication risk information as well as consulting with their providers and family/friends.

• Patient race, age and gender are factors related to medication risk information seeking behavior.

• Minorities and older patients are more likely to rely on their physician for medication risk information than on the internet.

• Men are more likely to rely on their pharmacist for medication risk information than on the internet, compared to women

Summary Points.

Knowing facts:

• Patients use internet to obtain health information, and communicate with their physicians.

• Physicians or other health care providers are generally a key resource for information about serious health episodes.

Points added from this study:

• Our study explores online medication risk information sources among medically underserved and vulnerable populations with a specific focus on nonsteroidal anti-inflammatory drugs (NSAIDs).

• Patients use internet search for medication risk information as well as consulting with their providers and family/friends.

• Actively using the internet as a primary source of medication risk information illustrates greater patient involvement.

• Patients search the internet for medication risks information but disparities across sociodemographic groups exist.

• Patient race, age and gender are factors related to where to get medication information seeking behavior.