Factors Influencing Judgments to Consult Prescription Monitoring Programs: A Factorial Survey Experiment

Abstract

Background:

Prescription monitoring programs (PMPs) can provide healthcare professionals with valuable information. However, few studies have explored providers’ decision-making for accessing PMPs.

Purpose:

This study aimed to identify provider characteristics and situational factors most influencing perceived importance of consulting the PMP for patients in a simulated context.

Design:

The study used a cross-sectional factorial survey.

Methods:

Community Pharmacists, Advanced Practice Registered Nurse (APRN) and Physicians were recruited by mail with a link to the online survey. The survey consisted of demographic questions, eight randomly generated vignettes, and one ranked item. The vignettes described a hypothetical prescription using eight experimental variables whose levels were randomly varied. Respondents evaluated each vignette for importance to access the PMP. Analyses used linear mixed-effects models in R (version 3.5.0).

Results:

A total of 138 responses were available for multilevel analysis. Women, physicians, and APRNs rated it more important to consult the PMP for a given prescription compared with men and pharmacists. Accessing PMP was perceived as more important with cash payments, quantity dispensed, suspicion for misuse, hydromorphone, and oxycodone prescriptions, and headache. Advancing age, postoperative pain, and anxiety or sleep indications were associated were associated with less importance.

Conclusions:

Age, indication for prescribing, misuse, and payment mode each independently had greater importance to providers in accessing the PMP. This was the first study to isolate the influence of different controlled substances on how important it was to consult the PMP.

Clinical Implications:

There is a need for targeted education and developing competencies to increase accessing PMP for safety reasons and not just misuse or diversion, and to raise awareness for monitoring weaker opioids and benzodiazepines.

The opioid epidemic has been declared a public health emergency (U.S. Department of Health & Human Services, 2017). According to the Centers for Disease Control and Prevention (CDC, 2017), 130 Americans die, on average, every day from an opioid overdose. Members of the National Institute on Drug Abuse have stated that the increase in the availability of prescription opioids for the treatment of pain has contributed to the rise in opioid misuse and diversion (Compton & Volkow, 2006). Consequently, cases of prescription opioid overdose have increased and now account for about 40% of opioid-related deaths (Seth, Scholl, Rudd, & Bacon, 2018). In 2016 alone, nearly 17,000 people died in the United States from overdoses involving prescription opioids, a fivefold increase since 1999 (Seth et al., 2018).

Multiple national initiatives have been undertaken to control the opioid crisis. One such initiative is the implementation of Prescription Drug Monitoring Programs (abbreviated PDMP or PMP). PMPs are state maintained electronic databases that collect data on dispensed controlled substance prescriptions. The information collected includes demographic details about each patient and their controlled substance prescription history. The prescription data generally includes the medication, quantity, written and fill dates, prescriber, and dispensing pharmacy. Healthcare professionals are encouraged to use the database to evaluate patient history for patterns that may indicate potential misuse and dangerous medication combinations. There has been evidence that PMPs have been successful in reducing inappropriate prescribing behavior, multiple prescriber use, and overdose deaths (Haffajee et al., 2018; Lin, Wang, Boyd, Simoni-Wastila, & Buu, 2018; Pardo, 2017; Patrick, Fry, Jones, & Buntin, 2016; Winstanley et al., 2018).

Currently, although almost all states have an operational PMP, each has requirements regarding access permission, drugs monitored, mandatory registration and use, and update frequency. In some states including Iowa, PMP use is voluntary, which means providers use their professional judgment to determine if accessing the PMP is warranted in a specific context (The Prescription Drug Monitoring Program Training and Technical Assistance Center at Brandeis University, 2018). Healthcare providers have multiple reasons for consulting the PMP during the prescribing or dispensing process, and no two patient situations are exactly alike. There may be concerns about appropriate dosing, the frequency of controlled substance dispensing, and the use of multiple prescribers and pharmacies (Fleming et al., 2014; Hildebran et al., 2014; St Marie, 2014). Patient behaviors like requesting specific drugs or demanding higher doses and frequencies, requesting early refills, reporting multiple allergies, and reporting a lack of efficacy for certain opioids also may prompt a provider to consult the PMP (National Association of Boards of Pharmacy, 2015). Some providers access the PMPs for every patient while others access it only with new patients, new prescription opioids, or patients they suspect of misuse or abuse (Irvine et al., 2014).

Professional judgment is a hallmark of being a licensed health care provider. Such discretion, however, is not without its potential limitations. In recent years, implicit bias has received attention owing to its potential impact on healthcare disparities. Implicit bias refers to unconscious attitudes or beliefs that can unintentionally impact behaviors and decisions (Blair, Steiner, & Havranek, 2011). Research suggests that implicit biases related to race, gender, age, and other factors can unconsciously affect the decision-making of health professionals (FitzGerald & Hurst, 2017; Maina, Belton, Ginzberg, Singh, & Johnson, 2018). Research has highlighted the impact of implicit bias in the treatment of conditions including mental illness, disability, obesity, and HIV status (Chapman, Kaatz, & Carnes, 2013; Maina et al., 2018; McHugh et al., 2013; Pokela, Bell, Lihavainen, Sulkava, & Hartikainen, 2010; Sadowski, Kee, VanderWeele, & Buchanan, 2009) and in the treatment of pain among African-Americans (Meghani, Byun, & Gallagher, 2012; Sabin & Greenwald, 2012). To the best of our knowledge, no research has specifically investigated implicit bias related to PMP consultation.

Consulting the PMP before writing or dispensing every controlled substance prescription may reduce some elements of bias, but this may not be feasible. Studies have reported significant barriers to using PMP databases, including difficulties with accessing and navigating the databases, insufficient time, and lack of knowledge of its use (Blum, Nelson, & Hoffman, 2016; Deyo et al., 2015; Garcia, Rosen, Finley, & Potter, 2017). For example, a study of community pharmacists in Kentucky found those working in chain pharmacies often did not have PMP access because there was no external internet access (Wixson, Blumenschein, Goodin, Talbert, & Freeman, 2015). In separate studies, pharmacists and nurses expressed challenges with using the PMP in practice and concerns about scrutiny by licensing board, legal ramifications, and relationship tensions with prescribers and patients that may accompany controlled substance medication monitoring. (Christianson, Driscoll, & Hull, 2018; Hartung et al., 2018; Norwood & Wright, 2016).

Given the role of professional judgment with voluntary access of the PMP, there is a need to systematically characterize the factors that influence a provider’s decision-making related to PMP consultation, including the potential role of implicit biases. Quantifying how factors independently influence decision-making may provide useful information applicable to developing best practices for PMP use, designing continuing education, identifying barriers, and otherwise improve PMP use. The purpose of this study is to: 1) identify provider characteristics that most influence perceived importance of consulting the PMP for patients in a simulated context; and 2) identify situational factors that most influence perceived importance to consult the PMP.

Methods

Design

A cross-sectional factorial survey was conducted to evaluate the decision-making process of multiple healthcare providers surrounding PMP use. Factorial surveys combine the convenience of a survey with the strengths of random assignment associated with an experimental design (Auspurg & Hinz, 2015; Rossi & Nock, 1982). Factorial surveys employ vignettes (short narrative accounts that describe hypothetical situations or patient cases) in which several variables presumed to influence decision-making are systematically incorporated. This approach has the unique ability to simultaneously measure the independent effects of respondents and situational factors without the need to give participants every possible combination of variables. Also, the multidimensionality of vignettes provides a useful context for testing unconscious bias, because respondents are always encountering a mix of information in each case, effectively shrouding whether any given response is a sign of bias. This method makes participants responses less susceptible to social desirability bias (Auspurg & Hinz, 2015), a type of self-report response bias in which individuals underreport socially undesirable behaviors and over-report desirable ones (Phillips & Clancy, 1972).

Participants

The study sample consisted of advanced practice registered nurses (APRNs), pharmacists, physicians and physician assistants (PAs) who had controlled substance registrations with the Iowa Board of Pharmacy. Mailing lists of providers were obtained from the Iowa Board of Pharmacy. An invitation letter was sent to providers by mail with an explanation of the study and an easily entered web link for them to access the survey. There were 752 letters initially sent to pharmacists, and 1,155 letters sent to a combination of APRNs, physicians, and PAs. One reminder postcard also was sent 4 weeks later. A notice about the study also was put in the daily newsletter for the University Hospital. The University Institutional Review Board approved the study. Participants read the elements of consent before starting the survey.

Data Collection

The survey was developed and conducted using Qualtrics^® online software platform. The web link to the survey was provided in the mailed letter. The survey remained active for three months following the mailings from May 10, 2018, to August 09, 2018. Participation in the survey was anonymous, voluntary, and no compensation was provided.

Survey Instrument

The survey questionnaire consisted of a demographic section, a vignette section, and a rank order item. The demographic characteristics collected include age group, gender, profession (physician, APRN, pharmacist, PA) and work status (full time/part-time).

The vignette section contained eight randomly generated vignettes, each describing a hypothetical patient case scenario. It is generally recommended that the number of vignettes presented to each respondent roughly corresponds to the number of variables used to construct the vignettes (Hox, Kreft, & Hermkens, 1991). Also, factorial surveys are generally analyzed using multilevel models. For such models, Hox et al. (2010) provided a general recommendation that approximately 100 respondents evaluating 10 vignettes each should provide reasonable parameter estimates. Taking into account these suggestions and a precautionary measure to avoid cognitive overload, eight vignettes were used. Participants were asked to imagine that they were on a task force evaluating the appropriateness of PMP use in their community. They were informed that the vignettes presented to them represent situations where area providers did not check the PMP. After evaluating each vignette, respondents reported their perceived importance of checking the PMP on a 0-10 point semantic differential scale (0=Not at all Important to 10=Extremely Important). These instructions were used to make the vignette evaluation task applicable across clinics and pharmacies. Also, certain environmental and relationship variables can be challenging to model in vignettes (Witry & Doucette, 2015).

Lastly, a rank order question was presented to providers to assess their perspectives surrounding prescription opioid use disorder. Concerns related to opioid prescribing stems from their potential for misuse and diversion along with safety risk particularly in patients concurrently taking benzodiazepines (Hurstak et al., 2017; Lipari & Hughes, 2017; Sun et al., 2017). Providers were asked to rank these concerns (drug misuse, drug diversion, and patient safety) in the order they considered to be the most important reason for consulting the PMP. The survey instrument was pilot tested for clarity with a small group of APRNs and pharmacists before disseminating.

Vignette Design

The study vignettes described hypothetical patient prescriptions constructed using eight experimental variables (Fig.1). The variables were patient age, race, prescribed drug, quantity, directions, indications, prescriber, and payment mode. Patients’ race was manipulated using White and African-American sounding names selected from relevant literature on discrimination (Bertrand & Mullainathan, 2004; Levitt & Dubner, 2005). The values or levels of the variables used to construct the vignettes are described in Table 1.

Figure 1.

Vignette skeleton and example. PRN = Pro Re Nata (As needed); APRN = Advanced Practice Registered Nurse.

Table 1.

Vignette Variables and Their Values

Variables	Variable type	Variable levels
Age	Continuous variable	Range: 18-81
Race	Dichotomous variable Reference: African-American	White and African-American sounding first namesa. Last names were generic.
Drug	Categorical variable with 10 values. Reference: Acetaminophen/Codeine 300/30	Acetaminophen/Codeine 300/30, Oxycodone 5 mg, Oxycodone/APAP 5/325, Hydrocodone/APAP 5/325, Hydromorphone 2mg, Alprazolam 0.5mg, Alprazolam 2mg, Oxycodone (ER) 10 mg, Oxycodone (ER) 40mg, Tramadol 50mg
Quantity	Continuous variable	5, 10, 15, 20, 30, 45, 60, 90
Indications	Categorical variable with 6 values Reference: pain	Pain, headache, post-operative pain, tooth pain, back pain, anxiety/sleep
Directions (pills per day)	Continuous variable	2, 3, 4, 6, 8, 12b
Prescriber in vignette	Categorical variable with 4 values Reference: APRN	MD, APRN, DO, PA
Payment	Categorical variable with 5 values Reference: Cash	Cash, Employer Insurance, Medicare, Medicaid, VA

Note. ER = Extended Release; MD = Doctor of Medicine; APRN = Advanced Practice Registered Nurse; DO = Doctor of Osteopathic Medicine; PA = Physician Assistant; VA = Department of Veterans Affairs.

^aWhite-sounding names: Greg, Brad, Jay, Matthew, Neil, Brett, Allison, Emily, Jill, Kirsten, Meredith, Sarah; African-American sounding names: Ebony, Keisha, Kenya, Layota, Tamika, Latonya, Darnell, Jermaine, Kareem, Jamal, Leroy, Tyron.

^bThe values are presented in text form in the vignette descriptions. Ex: Take 1 tablet every 4 hours, which translates to 6 tablets per day.

Statistical Analysis

Descriptive statistics were used to summarize the participant characteristics. The primary unit of analysis was the vignette. The dependent variable was the rating score which reflected the perceived importance of checking the PMP for a vignette. Because each respondent contributed to 8 vignette responses, the data had a multilevel structure where responses were clustered within the respondent. Thus, a mixed-effects regression model was fitted with vignettes as level 1 and respondents as level 2. First, a random intercept only model with no predictor variables (null model) was fitted to calculate the intraclass correlation that provided an estimate for the variance in scores attributed to the grouping variable, which in our case was the individual respondent. Then, vignette variables and respondent variables were added sequentially, and the corresponding changes in variance estimates were noted. Likelihood ratio tests were used to measure the goodness of fit between the models. All statistical analyses were conducted using R statistical software (version 3.5.0). A p-value < .05 was considered significant.

Results

A total of 178 responses were received which resulted in a response rate of 9.2%. Twenty responses were excluded because they did not answer key demographic or any of the vignette questions. There were only 4 responses from physician assistants, so these were excluded. This left 154 valid responses (7.9%). Table 2 presents the descriptive characteristics of respondents. More than half of the survey respondents (53.2%) were pharmacists followed by advanced practice registered nurse (34.2%) and physicians (12.5%). A majority of respondents were women (73.6%) and were full-time workers (80.7%). Most nurses and physicians (69%) reported working in clinic settings, whereas a majority of pharmacists reported working in a community setting (71%). Most respondents (91.7%) were registered with the PMP.

Table 2.

Characteristics of the Survey Respondents

Characteristics	APRN	Pharmacist	Physician	Totala
	(n = 53)	(n = 82)	(n = 19)	(n = 154)
Age
25-34	13 (28.3%)	15 (19.2%)	2 (11.1%)	30 (21.1%)
35-44	13 (28.3%)	24 (30.8%)	4 (22.2%)	41 (28.9%)
45-54	7 (15.2%)	20 (25.6%)	3 (16.7%)	30 (21.1%)
55-64	11 (23.9%)	15 (19.2%)	4 (22.2%)	30 (21.1%)
65+	2 (4.3%)	4 (5.1%)	5 (27.8%)	11 (7.7%)
Gender
Female	43 (93.5%)	54 (69.2%)	8 (44.4%)	105 (73.9%)
Male	3 (6.5%)	24 (30.8%)	9 (50.0%)	36 (25.4%)
Work Setting
Clinic	41 (77.4%)	4 (4.9%)	9 (47.4%)	54 (35.1%)
Community Pharmacy	0 (0.0%)	72 (87.8%)	0 (0.0%)	72 (46.8%)
Hospital	1 (1.9%)	4 (4.9%)	5 (26.3%)	10 (6.5%)
Hospital (ED)	3 (5.7%)	0 (0.0%)	4 (21.1%)	7 (4.5%)
Other	8 (15.1%)	2 (2.4%)	1 (5.3%)	11 (7.1%)
Work status
Full Time	43 (93.5%)	59 (75.6%)	13 (72.2%)	115 (81.0%)
Part Time	3 (6.5%)	16 (20.5%)	5 (27.8%)	24 (16.9%)
Other	0 (0.0%)	3 (3.9%)	0 (0.0%)	3 (2.1%)
PMP Registered
No	7 (13.2%)	4 (4.9%)	1 (5.3%)	12 (7.8%)
Yes	46 (86.8%)	78 (95.1%)	18 (94.7%)	142 (92.2%)

^aMay not sum to total due to the option to not answer, and in addition, some demographic questions were presented only to those who answered yes to the PMP registration status question.

After excluding 12 responses that were not registered with the PMP and 4 responses with missing data, mixed-effects regression analysis was conducted on 138 responses (7.1%) comprising 1,085 vignettes. The intraclass correlation of the null model (random intercept only model) was 36.8%, and the likelihood ratio test between fixed intercept model and random intercept model was found to be significant (χ2 [1] = 255.04, p<0.001), confirming better fit with mixed-effects modeling approach. Adding vignette variables to the null model reduced the within group variance by 36.7%, and the likelihood ratio test indicated a significant improvement in the model fit (χ2 [25] = 453.75, p<0.001). The final model included respondent level predictors which slightly improved the fit (χ2 [8] = 55.42, p<0.001) and reduced the between-group variance by an additional 32.6%. The individual parameters of the vignette and respondent variables from the full model are presented in table 3.

Table 3.

Parameter Estimates for reported importance to consult the PMP for a given vignette

Variables	Estimate	Std. Error	p
Respondent factors
Respondent Age
25-34 (Ref)	-	-	-
35-44	−0.108	0.413	0.794
45-54	0.397	0.447	0.377
55-64	0.177	0.463	0.703
65+	0.413	0.689	0.55
Respondent Profession
Pharmacist (Ref)	-	-	-
APRN	2.23	0.339	<0.001
Physician	1.43	0.473	0.003
Respondent Gender
Female (Ref)	-	-	-
Male	−0.801	0.374	0.032
Respondent Work status
Full Time (Ref)	-	-	-
Part Time	0.052	0.427	0.909
Vignette variables
Patient Race
African-American (Ref)	-	-	-
White	0.0032	0.123	0.981
Patient Age
Age	−0.014	0.00348	<0.001
Prescribed Drug
Acetaminophen/Codeine 300/30 (Ref)	-	-	-
Alprazolam 0.5mg	0.41	0.445	0.367
Alprazolam 2mg	0.539	0.434	0.224
Hydrocodone/APAP 5/325	1.001	0.242	<0.001
Hydromorphone 2mg	1.67	0.248	<0.001
Oxycodone 5 mg	1.03	0.254	<0.001
Oxycodone (ER) 10 mg	1.23	0.394	0.002
Oxycodone (ER) 40mg	0.93	0.379	0.0226
Oxycodone/APAP 5/325	1.13	0.245	<0.001
Tramadol 50mg	−0.294	0.246	0.201
Prescribed Quantity
Quantity	0.0305	0.00352	<0.001
Prescription Indication
Pain (Ref)	-	-	-
Anxiety or sleep	−0.836	0.391	0.033
Back pain	−0.145	0.192	0.45
Headache	0.579	0.221	0.009
Post-operative pain	−1.57	0.213	<0.001
Tooth pain	0.156	0.218	0.47
Prescription Directions
Directions	0.00539	0.0216	0.803
Prescriber in vignette
APRN (Ref)	-	-	-
DO	−0.0856	0.18	0.635
MD	−0.156	0.178	0.381
PA	−0.331	0.173	0.0565
Patient Insurance
Cash (Ref)	-	-	-
Employer	−1.45	0.198	<0.001
Medicaid	−1.3	0.177	<0.001
Medicare	−1.85	0.295	<0.001
VA	−1.61	0.174	<0.001

Note. Ref = Reference category; APRN = Advanced Practice Registered Nurse; APAP = N-acetyl-para-aminophenol.

Significant predictors at vignette level included patient’s age, prescribed drug, quantity, indication, and payment. Holding other variables constant, providers reported it was more important to check the PMP for younger patients. Acetaminophen/codeine was used as reference for analysis because it was the least potent drug and viewed as least important for checking the PMP. Except for alprazolam and tramadol, all other drugs have a statistically significant positive association, with hydrocodone having the most influence on the providers’ rating of importance to check the PMP. Drug quantity showed a positive effect with higher quantities associated with higher importance. Compared with an indication for “pain,” respondents were more lenient for an indication of “post-operative pain” and “anxiety or sleep,” but viewed it as more important to consult the PMP for an indication of “headache.” Regarding the payment mechanism, paying with cash was found to strongly predict providers’ perception that checking the PMP was important, compared to other modes of payment. There was no statistically significant influence of the vignette variables for patient race, prescription directions, and prescriber type on how respondents rated the importance of consulting the PMP.

Two factors about the respondents were found to be statistically significant, those being profession and gender. Compared to pharmacists, nurses and physicians found it more important to check the PMP overall. Likewise, women had significantly higher importance scores compared with men. No significant effects were found for respondent age and work status (full time/part-time).

In response to rank order question, most providers rated patient misuse as the most important reason (p < .001) for consulting the PMP. Drug diversion and safety concerns were ranked lower (Figure 2).

Figure 2.

Providers ranking of reasons for consulting Prescription Monitoring Program. APRN = Advanced Practice Registered Nurse.

Discussion

Significant results were found at the vignette level and at the healthcare provider level for influencing the perceived importance of checking the PMP. Key findings from this study were 1) variables such as patient age, type of drug prescribed, patient diagnosis, and type of payment were influential in how respondents viewed the importance of consulting the PMP; 2) there were provider-level differences when evaluating when it was important to consult the PMP, such as provider type and gender, and 3) implicit bias towards race was not identified in our sample.

Influence of Vignette Variables

For this factorial survey experiment, the variables of patient age, type of drug prescribed, diagnosis, and type of payment significantly influenced how important respondents deemed it was to consult the PMP. Our study found that health care providers judged checking the PMP as more important for younger patients than older patients. This was consistent with the literature reporting a higher risk for opioid misuse between ages 18 and 25, versus other age groups (National Institute on Drug Abuse, 2016). However, prescription opioid use has been increasing in older adults. More than 70% of people over the age of 45 have had intense pain, greater mental health needs than the population norm, and experienced disability and serious medical conditions with greater access to legitimate prescription opioids (Arndt, Clayton, & Schultz, 2011; Cicero, Surratt, Kurtz, Ellis, & Inciardi, 2012). There has also been an increased rate for first-time admission to treatment for the older age group (Arndt et al., 2011). For these reasons, it may be equally important to consult the PMP for older patients as for younger patients.

The prescribed drug in the vignette played an important role in how providers evaluated the importance of consulting the PMP. Compared to acetaminophen/codeine, all other drugs except for tramadol and alprazolam were associated with a stronger rating for importance in consulting the PMP, with hydrocodone having the strongest influence. The lack of significant effect for alprazolam and tramadol and the lower scores obtained for patient safety concerns in the rank order question in the survey, together suggests that the primary focus of providers accessing PMP is to investigate opioid misuse rather than safety concerns for an inappropriate combination of benzodiazepines and opioid use. The literature indicates trends and impact of outpatient prescribing of opioids and benzodiazepines increasing the risk of adverse effects such as respiratory depression, overdose, and death, making this a top priority for monitoring (Hirschtritt, Delucchi, & Olfson, 2018; McClure, Niles, Kaufman, & Gudin, 2017). To our knowledge, this is the first study to isolate the influence of different controlled substances on provider judgments related to PMP use.

The diagnosis for prescribing opioids was influential for healthcare providers’ ratings of PMP consultation importance. Headache was viewed as a potential red flag by providers and resulted in more stringent checking of the PMP when patients presented with or were prescribed opioids for headache. This may be a manifestation of their awareness of guidelines that opioids should not be used in the routine treatment of headache, which may result in inappropriate prescribing (Dowell, Haegerich, & Chou, 2016; Loder, Weizenbaum, Frishberg, & Silberstein, 2013). Our findings indicated post-operative pain was seen as a less important context to check the PMP as there was more leniency of checking the PMP when patients were prescribed opioids for postoperative pain. There is some evidence, however, that PMP consultation could be valuable for the post-operative group as they could obtain opioids from multiple prescribers (Morris, Zumsteg, Archer, Cash, & Mir, 2014), for example, if they already had an existing opioid prescription for another condition.

Healthcare providers strongly considered checking the PMP for patients paying for their prescriptions in cash. This finding confirmed that health care providers viewed cash transactions involving controlled substances as an indicator of high-risk opioid use. Cash payments generally signal a warning sign or red flag because some people abusing prescription drugs used cash as a method to evade monitoring (Cepeda, Fife, Chow, Mastrogiovanni, & Henderson, 2013; Hartung et al., 2017). In a survey of Texas community pharmacists, it was found that patients paying for controlled substances with cash or obtaining early refills were the most influential triggers for PMP use (Fleming et al., 2014). Several studies related similar findings where providers reported accessing the PMP when they were suspicious of the patient’s drug-seeking behavior (Hildebran et al., 2014; Irvine et al., 2014; Leichtling et al., 2017). Our study showed that health care providers rated patients’ misuse as the most important reason to consult with the PMP.

Influence of Respondent Variables

The present study found two influential provider level variables related to the overall level of how important a given provider judged checking the PMP was for any prescription. Checking the PMP was rated of higher importance by APRNs and physicians than pharmacists. Although pharmacists showed positive intention to use PMP in previous studies (Fass & Hardigan, 2011; Fleming et al., 2014; Gavaza, Fleming, & Barner, 2014), the findings suggested the intention may be more conditional compared to prescribers. Possible reasons for the relatively lower values for PMP utilization may include lower self-efficacy among pharmacists to address prescription misuse, limited access to patient information, and technical barriers like PMP complexity and usability (Cochran, Field, Lawson, & Erickson, 2013; Hartung et al., 2018; Norwood & Wright, 2016). In addition, pharmacists may assume the prescriber has taken responsibility for checking the PMP, or pharmacists may rely on other data sources to evaluate controlled substance use, such as pharmacy records and automated alerts from insurers when processing prescription claims.

The gender of the respondent also was found to be influential. Males overall rated checking the PMP as less important than females independent of vignette characteristics. This matches a finding from a study of New York physicians who were surveyed about their patterns of PMP use (Blum et al., 2016). In this study, male physicians reported significantly lower rates of PMP use even though New York State regulations mandated PMP checks before prescribing controlled substances. They also found men were more likely to have negative views of the mandatory PMP use policy compared to women.

Implicit Bias

This study did not identify an implicit bias toward race. Similar findings were reported in studies that investigated opioid-related clinical decision making. One study found no association between patient race as represented by skin color and name in the clinical vignettes and opioid analgesic prescribing by emergency physicians (Tamayo-Sarver et al., 2003). Another study used virtual patient stimuli and found no effect of provider racial bias on pain management decisions (Hirsh, Hollingshead, Ashburn-Nardo, & Kroenke, 2015). Implicit bias has been found to have important effects in other analyses such as regarding the treatment of pain in African-American patients and for patients with stigmatized conditions such as mental illness, disability, obesity, and HIV status (Meghani, Byun & Gallagher, 2012; Sabin & Greenwald, 2012; Chapman, Kaatz & Carnes, 2013; Maina et al., 2018; McHugh et al., 2013; Pokela et al., 2010; Sadowski et al., 2009). More work is needed to investigate the potential role of implicit bais on PMP use using other tests of implicit association.

Implications for Pain Management Nurses

These findings have key direct implications for pain management APRNs and registered nurses as a call for additional education, clinical practice, research, and policy. We know from this study that younger age, diagnosis such as headache, payment in cash, and use of high potency opioids all influence provider perceptions of the importance of consulting the PMP. This calls to attention a much-needed renewal of efforts to provide safe and effective pain management for patients that may be overlooked given the provider tendencies highlighted in our analysis.

Education of healthcare providers

Education of healthcare professionals, and preparing pre-licensure healthcare professionals should include standardized competencies regarding when to access the PMP. In an article by Herr et al. (2015), recommendations encompass an implementation strategy to affect pain management in communities by teaching access to PMPs. Further recommendations for education entail identifying and discussing the biases that may impact patients when pain management care is withheld due to assumptions rather than through objective data gathering. People in pain and receiving opioids for the treatment of pain are vulnerable to stigmatizing and marginalizing behaviors of their healthcare professionals (St Marie, 2014).

Furthermore, in our study healthcare provider emphasis on accessing PMP when prescribing or monitoring patients on higher potency opioids may have further implications for training health care professionals. Prescribers and pharmacists may be underestimating the potential role for lower potency opioids and anxiolytics to be associated with problems that could be identified using the PMP. For example, while hydromorphone and extended-release oxycodone may be associated with diversion, benzodiazepines may warrant extra attention in older adults where coprescribing of benzodiazepines and opioids can be particularly harmful (Jones & McAninch, 2015; Park, Saitz, Ganoczy, Ilgen, & Bohnert, 2015; Sun et al., 2017). Universally accessing the PMP on all patients receiving opioids for pain and not just those who provoke suspicion for misuse may be needed. Educating healthcare professionals on the use of this important monitoring program is essential because it allows the health care team to evaluate the benefits and potential harm of opioid prescribing.

Clinical practice changes

Providing clinical care that treats pain responsibly through acknowledging risks for opioid use disorder may lead us out from under these population health problems of undertreated pain and the opioid crisis. Accessing the PMP based on diagnosis may not be the best approach because substance misuse can occur in patients with all diagnoses and in all healthcare settings. Healthcare providers may decide not to check the PMP based on diagnoses such as acute pain or hospice and palliative care. However, evidence indicates these are areas of risk that could go unidentified. Opioid prescriptions from the emergency department are associated with chronic opioid use (Barnett, Olenski, & Jena, 2017; Hoppe, Kim, & Heard, 2015; Volkow, McLellan, Cotto, Karithanom, & Weiss, 2011). Persistent postoperative pain is a common problem occurring in a range of 5% to 75% of patients following a variety of surgeries (Deumens et al., 2013). Persistent postoperative pain and prolonged opioid use after surgery are highly correlated (Issa, Ahmed Ali, Bouwense, van Santvoort, & van Goor, 2014; Puolakka et al., 2010; Raebel et al., 2013). In a survey of 34 hospice and palliative medicine fellowship programs, 77.2% of respondents reported seeing more than one patient with substance use disorder and nearly 44% treated patients when there was suspicion of misuse (Childers & Arnold, 2012). Although the Centers for Disease Control and Prevention guidelines state that the guidelines are not applicable to palliative care or hospice, access of the PMP for this patient population may help patients suffering from the disease of addiction receive appropriate treatment. The importance of protection of sobriety was considered a very important factor, as reported in a study in primary care (St Marie, 2016).

Further support for clinical practice change to enhance pain management while minimizing risk can occur through comprehensive assessment, patient education, using a wide range of treatment options, and monitoring to mitigate risk (Arnstein, St. Marie, & Zimmer, 2017). The comprehensive assessment includes gathering data on the history of opioids and from where are they being dispensed. The PMP is the best resource across the country that contains that data in a timely fashion. A patient print out from the PMP can be used to educate and counsel the patient on their prescribed history, as well as the potential for dangerous drug combinations when benzodiazepines are concurrently prescribed. Accessing the PMP with each opioid or benzodiazepine prescription written or dispensed can facilitate the early identification of risk and subsequent mitigation with appropriate prescription adjustments or referral to treatment.

An interdisciplinary approach to pain is necessary to facilitate positive change in practice. The example in our study of cash payment triggering the pharmacist to access the PMP illustrates this concept. Pharmacists have access to how the patient pays for the prescription opioid, and prescribers do not. Important communication between pharmacists and prescriber on this finding cannot be overemphasized.

Future Research

The findings from this study suggest several avenues for future research. The question of implicit bias could be assessed in epidemiological studies of PMP use. Although one would have to carefully adjust for confounding that was avoided in the present study, looking at real PMP use data may provide for triangulation of this issue. Also, more work remains in the appropriateness of PMP use once the provider is examining data. For example, if a provider is only focused on multiple provider prescribing, he or she may miss a safety issue like a coprescribing of a benzodiazepine and an opioid.

Health policy

Access to medically necessary data through the PMP has proven worth. Mandatory use of PMP for every opioid prescription has been shown to reduce overdose deaths (Patrick et al., 2016), however, this comes with an added time burden to providers (Bachhuber et al., 2018; Poon et al., 2016). Local and system policy should be encouraged to creating ease of access for all clinicians to reduce barriers to access such as integrating PMP data into electronic medical records (Centers for Disease Control and Prevention, 2017; Rutkow et al., 2017). If the length of time is increased for the provider to access the PMP and provide education and counseling, providers often can authorize delegates to access PMP reports, although little is known about how delegates interpret PMP data. Federal policy should be encouraged to fund a National Prescription Drug Monitoring Program that connects each state monitoring system through easy computer-human interface and in real time. This will help monitor and improve communication for patients who travel between state, so they can be safely cared for and receive early identification of risk behaviors or hazardous drug combinations.

Limitations

This study has several limitations. The sample was only drawn from one state in the U.S. Variation in regulations, state initiatives, and rates and trajectories of opioid-related deaths may lead providers in other states to having different results. Registered nurses without prescriptive authority may have a unique role in receiving delegation to access PMP; however, this was beyond the scope of this study. The study might be subject to response bias if more motivated or conscientious providers were more likely to respond. Furthermore, the dependent variable related to respondent judgments about the importance to check the PMP, not an actual check of the PMP. Also, the African-American sounding names may not have sufficiently tested for potential implicit bias related to race. However, several studies of racial disparities in other fields have used such racially identifiable names to experimentally manipulate race and found significant effects (Bertrand & Mullainathan, 2004; Ewens, Tomlin, & Wang, 2014; Giulietti, Tonin, & Vlassopoulos, 2017; Holbrook, Fessler, & Navarrete, 2016; Okonofua & Eberhardt, 2015). Other methods to reveal implicit bias such as priming tasks or using images and videos may be more salient (Burgess et al., 2008; Green et al., 2007; Mathur, Richeson, Paice, Muzyka, & Chiao, 2014).

Conclusions

The variables patient age, prescribed drug and quantity, indication for prescribing, and payment mode each were independently found to increase how important a provider deemed it to check the PMP for a given hypothetical controlled substance prescription. At the provider level, women compared with men, and nurses and physicians compared with pharmacists were found to rank checking the PMP database as more important for a hypothetical controlled substance prescription. This study isolated the influence of different controlled substances on the importance of consulting the PMP. Additionally, most health care providers rated patients’ misuse as the most important reason to consult the PMP rather than patient safety. Findings in this study have implications for pain management nurses for education, clinical practice, research, and health policy.