|Home | About | Journals | Submit | Contact Us | Français|
Pain assessment and treatment is influenced by patient demographic characteristics and nonverbal expressions. Methodological challenges have limited the empirical investigation of these issues. The current analogue study employed an innovative research design and novel virtual human (VH) technology to investigate disparities in pain-related clinical decision making. Fifty-four nurses viewed vignettes consisting of a video clip of the VH patient and clinical summary information describing a post-surgical context. Participants made assessment (pain intensity and unpleasantness) and treatment (non-opioid and opioid medications) decisions on computerized visual analogue scales. VH demographic cues of sex, race, and age, as well as facial expression of pain, were systematically manipulated and hypothesized to influence decision ratings. Idiographic and nomothetic statistical analyses were conducted to test these hypotheses. Idiographic results indicated that sex, race, age, and pain expression cues accounted for significant, unique variance in decision policies among many nurses. Pain expression was the most salient cue in this context. Nomothetic results indicated differences within VH cues of interest; the size and consistency of these differences varied across policy domains. This study demonstrates the application of VH technology and lens model methodology to the study of disparities in pain-related decision making. Assessment and treatment of acute post-surgical pain often varies based on VH demographic and facial expression cues. These data contribute to the existing literature on disparities in pain practice and highlight the potential of a novel approach that may serve as a model for future investigation of these critical issues.
Despite a burgeoning pain literature, pain-specific curricula , and clinical practice guidelines [68,5,1], pain continues to be misunderstood and mistreated. By at least one estimate in cancer patients, over 80% of pain sufferers receive insufficient relief, largely due to excessively conservative pharmacologic treatment . Such mismanagement likely results from several interacting factors .
Patient demographic characteristics – sex, race, and age in particular – may influence pain-related decision-making. In a laboratory-based study, males undergoing the cold pressor test had their pain underestimated by viewers to a greater degree than females . The clinical literature is mixed, with some reports of females at greater risk of having their pain underestimated by providers [16,6] and others finding no sex differences . The treatment literature is replete with studies indicating females are at increased risk of having their pain under-treated relative to males [10,13,16,19,24,44,45], although such differences are not always found [8,14,64,65].
Studies examining race/ethnic differences in pain assessment are mixed. A multicenter study found minority cancer patients were more likely to have pain underestimated than White patients . Anderson and colleagues  reported 74% and 64% of African-American and Hispanic patients, respectively, had their pain underestimated. Minority patients may also be less likely to have their pain documented altogether . In contrast, several studies found no differences in pain assessment among different racial/ethnic groups [62,61]. There is considerable evidence of race/ethnic disparities in pain management across a range of conditions and settings [6,9,16,15,47,53,54,59,62]. The direction of these differences is consistent, with members of minority groups receiving less aggressive treatment than Caucasians. It should be noted, however, that such disparities are not always found [8,14,40,58].
Pain assessment in the elderly is poor relative to younger populations [25,29,35]. This is likely due to multiple factors including patient and provider beliefs about pain , under-report of pain by elderly patients , lack of standardized assessment instruments , and higher rates of medical comorbidities in elderly patients [18,17,20,26,38,56]. Elderly patients are also at risk of being under-treated for pain ; perhaps especially in post-surgical settings [39,48]. Results of a vignette study also indicated providers’ pain medication decisions are influenced by age, with some nurses predisposed to administer less medication to older patients . The use of opioids for chronic, non-malignant pain may also be underutilized in the elderly [4,7,26].
Although the aforementioned literature suggests patient demographic features play a role in pain assessment and treatment, methodologic limitations of common research designs place constraints on conclusions that may be drawn from these data. Most notably, retrospective studies lack empirical control, while vignette studies lack ecological validity. Furthermore, the literature has focused on the decision-making product to the exclusion of the preceding process. In a previous study , we presented virtual human (VH) technology and lens model methodology as an alternative approach by which to extend the investigation of pain-related decision-making. In the current study, we aim to contribute further to this literature by examining pain assessment and treatment disparities in nurses, who are at the forefront of pain management.
Fifty-four currently practicing nurses participated in this study. Participants needed to be at least 18 years of age and a licensed practicing Registered Nurse (RN). Students and those with advanced nursing degrees were included if they met the aforementioned criteria. Participant recruitment occurred at the local and national level. Local recruitment strategies included presentations at class lectures, advertisements displayed in local hospitals and clinics, and presentations at association meetings. National recruitment occurred via nursing email listservs and national meeting presentations. Consistent with national data, the majority of participating nurses were female (83%) and self-reported Caucasian (93%), with an average age of approximately 42 years (SD = 11.90). A wide range of U.S. geographical locations was represented; Florida (n = 23) was the modal state of residence. Participants with Associate (n = 22), Bachelor (n = 17), and graduate (n = 15) degrees in nursing were included. Approximately 72% were not currently enrolled in an academic program at the time of study participation. Of the 15 nurses who were students, the majority (n = 11) was pursuing graduate degrees. Examination of self-reported professional background data indicated that the average years of nursing experience was approximately 14 (SD = 10.52). The three most frequently endorsed current practice areas were critical care (n = 22), primary care (n = 16), and oncology (n = 14). With one exception, all nurses reported experience working in a hospital setting.
The current study employed a lens model design and was powered for the idiographic analyses of this approach. The lens model is an analogue method for capturing how individuals use information in their environment to form judgments. Inherent in the lens model approach is the assumption that judgment processes are contextually determined. That is, an individual’s judgment is determined based on his/her attention to and weighting of the information (cues) available in the immediate environment. In lens model applications, individuals are presented a series of profiles containing cues that may be used to form a judgment. The profiles depict cases or situational contexts for the individual to process, and each contains a unique combination of cues. The outcome of the judgment process for each profile is obtained using a quantifiable response mode, such as a numerical rating scale (NRS) or visual analogue scale (VAS). Policy capturing occurs at the idiographic level first (see Statistical Analyses section below). Idiographic analyses refer to statistical tests that are conducted at the individual participant level. Following this level of analyses, data can then be aggregated for nomothetic analyses. Nomothetic analyses refer to the more common approach of statistical testing at the group level.
A principal concern of lens model designs is the ratio of profiles to cues. A balance must be achieved that considers the ratio needed for estimation of stable regression coefficients while also taking into account the demand placed on study participants. The smallest recommended profile-to-cue ratio is 5:1, but a 10:1 ratio may be preferred given logistical feasibility . Idiographic power of this study was maximized by employing a ratio that exceeded the acceptable 5:1 ratio. The 10:1 ratio would likely have imposed undue burden on study participants through the creation of a large number of profiles and, thus, was not adopted. This study investigated 4 contextual cues (age, race, sex, and pain expression) and used a total of 32 profiles, which is a profile-to-cue ratio of 8:1. This ratio permitted each possible cue combination to be presented twice, which further enhanced statistical power. It was expected that this ratio would ensure adequately powered nomothetic analyses when the idiographic data were aggregated. Lens model designs that employ a sufficient profile-to-cue ratio have enhanced power at the nomothetic level due to greater reliability of each individual’s data as a result of multiple observations. Thus, policy-capturing investigations like the current study can achieve adequate power with a smaller sample size than traditional research designs .
Each VH profile consisted of a vignette and 20-second looped video clip (ie, the video continuously restarts without pause) of a VH representing the patient. Presentation of these profiles was randomized for each participant to prevent order effects. The same vignette (see Appendix) was used for each profile and described a patient who underwent an open appendectomy; it also contained patient clinical information indicating the status of the patient, pain complaint (duration and location), and prescription medication orders. Physiological information (temperature, blood pressure, pulse rate, respiration rate, and mental status) was also provided, and these values varied among profiles but were always within normal limits. The VH videos contained dynamic facial expressions of pain (low and high) and were generated with People Putty software, an innovative technology for the creation of virtual characters. This software eliminates from the development of the actual stimuli the very biases the present study is intending to investigate. A diverse collection of characters was created via the systematic manipulation of VH sex, race, and age. The facial expressions of pain for the VH were also systematically manipulated based on the Facial Action Coding System (FACS). The FACS is an objective, anatomically-based system that permits a full description of the basic units of facial movement associated with private experience, including pain. Forty-four different action units (AUs) have been identified. Core action units representing the facial expression of pain in adults are: brow lowering (AU4), tightening of the orbital muscles surrounding the eye (AU6&7), nose wrinkling/upper lip raising (AU9&10), and eye closure (AU43) [23,50]. The pain expression can be differentiated from other negative subjective states, such as disgust, fear, anger and sadness [41,42,31].
Each VH contained four cues: sex (male or female), race (Caucasian or African American), age (young or old), and pain (low pain expression or high pain expression). A total of 32 unique scenarios were created, permitting each possible cue combination to be presented twice. Sample still frame images captured from the VH videos are contained in our previously published manuscript . Participants used computerized visual analogue scales (VASs) to provide pain assessment (intensity and unpleasantness) and treatment (non-opioid and opioid medication) ratings for each VH profile. Pain assessment ratings were recorded on separate VASs with endpoints at no pain sensation and most intense pain sensation imaginable for pain intensity, and not at all unpleasant and most unpleasant imaginable for pain unpleasantness. Pain treatment ratings were based on participants’ likelihood of administering a non-opioid and opioid analgesic within the prescribed dosage. Separate VASs were used for each rating, with endpoints at not at all likely and complete certainty.
Participants also completed the Gender Role Expectations of Pain (GREP). The GREP  is comprised of 12 VASs that assess an individual’s view of the typical man and woman with respect to pain sensitivity, pain endurance, and willingness to report pain. It also assesses the individual’s personal attribution of his or her pain sensitivity, pain endurance, and willingness to report pain relative to the typical man and woman. Psychometric properties of the GREP are sound. The factor structure is consistent with the theoretical formulation of the scales and accounts for 76% of the variance in scores. The GREP has good test-retest reliability with individual item correlations ranging from .53 to .93. High correlations (-.71 to -.81) between individual items reflecting the opposite gender role (i.e., typical male endurance of pain correlated with typical female endurance) demonstrates internal consistency. Finally, sex differences in the endorsement of items on the GREP are large, with the largest differences (46% of variance) shown for “willingness to report pain” items. These differences provide evidence for the construct validity of the measure . The GREP has also been demonstrated to be a significant predictor of experimental pain ratings in undergraduate men and women, accounting for a significant proportion of the sex differences in pain report . Consistent with previous research , two theoretically important items from the GREP were included in the subsequent analyses to determine if gender stereotypes about “endurance of pain” and “willingness to report pain” influence clinical decisions regarding pain assessment and management practices.
This study employed a WEB-based delivery model. The majority of participants completed the study on a personal computer; the remainder participated on a laboratory computer. Each participant provided electronic consent. A demographic questionnaire was completed next. The order of the GREP and VH profile administration was counterbalanced and followed the demographic questionnaire. Presentation of the 32 VH profiles was also counterbalanced to prevent order effects. The following procedure was used for all administrations of the VH profiles: 1) participants read the clinical information and view the video simultaneously; 2) participants complete questions that ask them to provide pain assessment and treatment ratings. Prior to the patient profile portion of the study, participants read an “instructions” document that informs them about how to approach the task and how to use the electronic VASs to give pain assessment and treatment ratings. Participants are instructed to fully complete the questions for each profile and are not permitted to return to previously completed profiles. To maximize compliance with instructions and provide answers to frequently posed questions, a help menu was provided and accessible at all times.
At study conclusion, participants completed a short task-validity questionnaire that asked them to guess at the purpose(s) and/or hypothesis(es) of the study. Participants were then debriefed regarding the variables of interest and the study hypotheses. A brief educational tutorial regarding pain practice with sex, age, and ethnically diverse patients was then provided, after which participants completed a short test of their knowledge in this area. Continuing education credits or financial compensation were provided for participation. All data were collected and stored in an electronic database. The time necessary to complete the study varied between approximately 60 and 90 minutes, and was primarily a function of participant computer specifications.
Descriptive statistics were conducted for the demographic and background characteristics of the sample. Both idiographic (individual-based) and nomothetic (group-based) analyses were conducted. At the idiographic level, simultaneous multiple regression equations were generated for each participant to capture their decision making policies. The term ‘decision policy’ refers to the consistent approach an individual takes in weighting contextual cues to make a given decision. VH cues of sex, race, age, and pain expression were the independent variables in each model. Pain assessment and treatment ratings were the dependent variables in their respective models. The standardized regression coefficients (β) in each equation represent the weight of each cue in the formation of the assessment and treatment judgments. This weight represents the unique contribution and relative importance of each cue in the participant’s clinical decision. The coefficient of multiple determination (R2) represents the amount of variance in assessment and treatment policies accounted for by the independent variables, or the overall function of the cues in each individual’s policy. Overall policies (R2) and individual cues (β) were considered significant at p < .05. Nomothetic analyses were conducted next. For each participant, average assessment and treatment ratings were calculated across VH at each level of cue. Paired samples t-tests (for normally distributed data) and Wilcoxon signed-rank tests (for non-normally distributed data) were then used to compare ratings within cue for the entire sample. Effect sizes were calculated for significant results. For paired samples t-tests, Cohen’s d was calculated based on the standard deviation of the difference in ratings. Coefficient r was used as the measure of effect for Wilcoxon signed-rank tests.
Individual regression equations were conducted for each participant to model their decision policy. Results of these idiographic analyses are presented below and in Table 1.
Results indicated that 33 out of 54 nurses had significant (p < .05) decision policies for pain intensity assessment. Six of these 33 participants used sex as a significant cue in their policy. Five nurses gave higher pain intensity ratings for female VH; the reverse was true for 1 nurse. Race was a significant cue in the policies of 5 of these 33 nurses, with 4 more likely to judge higher pain intensity in African-American VH and 1 more likely to judge higher pain intensity in Caucasian VH. All seven nurses who used age as a significant cue were more likely to judge older VH as experiencing greater pain intensity than younger VH. Finally, pain expression was a significant cue for 32 participants, and each rated VH displaying high levels of pain expression as experiencing more pain.
Similar results were obtained for pain unpleasantness ratings, such that 34 out of 54 participants had significant overall policies for this decision domain. Examination of the contribution of the specific contextual cues indicated that VH sex, race, age, and pain expression were significant cues in the policies of 10, 3, 9, and 34 of these nurses, respectively. Eight participants provided higher ratings for female VH; the converse was true for 2 participants. African-American VH were judged to be experiencing more pain unpleasantness by 2 nurses, whereas 1 nurse rated Caucasian VH as experiencing more pain. All 9 participants who had a significant coefficient for the age cue provided higher pain ratings to older VH compared to younger VH. Finally, all 34 nurses with a significant pain expression cue judged those VH with high expressivity to be experiencing greater pain unpleasantness than those VH with low expressivity.
Results indicated that 23 out of 54 nurses had significant decision policies for treatment with an opioid medication. All 4 participants who had a significant coefficient for the VH sex cue provided higher ratings for female VH compared to male VH. Similarly, all 4 nurses who significantly weighted VH race in their opioid medication decision policies were more likely to administer opioid treatment to African American VH compared to Caucasian VH. Of the 7 nurses who used age as a significant cue, 6 were more likely to engage in this treatment with older versus younger VH. The converse was true for 1 nurse. Finally, all of the 23 nurses with a significant pain expressivity cue were more likely to administer opioid medication to high expression VH patients.
Thirteen participants had significant policies for non-opioid treatment. VH sex was significantly weighted by 2 of these nurses. Both gave higher ratings to female VH patients and, thus, were more likely to engage in this treatment practice with them compared to male VH. Only 1 nurse used race as a significant cue, and this participant was more likely to administer a non-opioid medication to African-American VH. Age was a significant cue in the policies of 4 of these nurses, with all 4 providing higher ratings for younger VH compared to older VH. Lastly, pain expression served as a significant cue for 12 nurses, with low expression VH patients being more likely to receive non-opioid medication by 7 nurses and high expression VH patients more likely to receive this treatment by 5 nurses.
To quantify the amount of variance accounted for by each cue in the various decision policies, individual standardized regression coefficients for each cue within each policy across nurse were squared (Table 2). Results of these calculations indicated that sex, race, age, and expression cues accounted for as much as 13%, 15%, 26% and 77%, respectively, of the variance in policies for pain intensity assessments and 14%, 13%, 28%, and 79%, respectively, of the variance in policies for pain unpleasantness assessments. Examination of the regression coefficients for treatment policies revealed that sex, race, age, and expression cues accounted for a maximum of 17%, 22%, 14%, and 85%, respectively, of the variance in opioid decisions and 21%, 30%, 26%, and 38%, respectively, of the variance in non-opioid decisions. The aforementioned values represent the maximum amount of variance accounted for by each cue. Additional descriptive data, including mean values, are presented in Table 2.
For each nurse, average assessment and treatment ratings were calculated across VH patients at each level of cue (Tables 3 and and4).4). Normality assumptions were satisfied for pain intensity assessment and opioid treatment ratings. Ratings for pain unpleasantness assessment and non-opioid treatment deviated slightly from a normal distribution. Paired samples t-tests (for normally distributed data) and Wilcoxon signed-rank tests (for non-normally distributed data) were then used to compare ratings within cue for the entire sample.
For pain intensity and unpleasantness ratings, statistically significant differences were present within each cue (Table 3). Nurses assessed VH who were female, African-American, older, and displaying high pain expressivity to be experiencing both greater pain intensity and unpleasantness than their demographic counterparts. Follow-up analyses were conducted to test the a priori hypothesis that gender-role expectations about pain would influence participants’ assessment ratings. Since within-cue sex differences emerged for both pain intensity and unpleasantness ratings, follow-up analyses were conducted separately for these decision domains. Correlation analyses indicated the GREP factor of “willingness to report pain” was significantly associated with average pain intensity assessment ratings for both male (r = .31, p < .05) and female (r = .30, p < .05) VH patients. The association between this factor and average pain unpleasantness ratings approached significance for both male (rs = .26, p = .056) and female (rs = .27, p = .053) VH patients. The GREP factor of “stereotypic endurance” for pain was not significantly associated with average pain intensity or unpleasantness ratings for either male or female VH patients; however, the magnitude of these relationships (coefficient range: -.15 to -.23) was sufficient as to warrant follow-up analyses. Analysis of Covariance (ANCOVA) results indicated the significant within-cue sex differences persisted [pain intensity: F(1,51) = 10.73, p < .01; pain unpleasantness: F(1,51) = 13.24, p < .01] even after controlling for the GREP factors of “willingness to report pain” and “stereotypic endurance” for pain.
Nomothetic results for pain treatment decisions are presented in Table 4. There were no significant differences within cue for non-opioid treatment decisions. Relative to their demographic counterparts, opioid treatment decisions were more likely to be endorsed for VH who were female, African-American, older, and displaying high pain expressivity. Follow-up analyses were conducted to test the a priori hypothesis that gender-role expectations about pain would influence participants’ treatment decisions. These analyses were confined to ratings for opioid treatment, since sex differences only emerged for this treatment domain. Correlation analyses indicated no significant association between the GREP factor of “willingness to report pain” and average opioid treatment ratings for male (r = .14, p > .05) and female (r = .15, p > .05) VH patients. Similarly, no significant association emerged between the GREP factor of “stereotypic endurance” for pain and average opioid treatment ratings for male (r = .03, p > .05) and female (r = .08, p > .05) VH. Due to these non-significant findings, no further analyses were conducted to control for the effects of gender-role expectations about pain.
Pain is frequently inadequately assessed and under-treated [60,68]. Patient sex, race/ethnicity, and age are potential sources of these deficiencies [6,16,35,44,47,48,52]. This study addressed several methodologic limitations of the literature through use of VH technology and lens model methodology. Results indicated this approach successfully captured the pain-related decision policies of participants. Although replication is needed, this investigation illustrates an alternative and promising approach by which to continue the study of medical decision-making.
Idiographic analyses indicated over 60% of nurses had significant decision policies for pain assessment. Stated differently, the contextual information (cues) provided in the clinical scenarios (clinical situation including vignette) was sufficiently weighted by the majority of nurses to result in a reliable decision equation. This suggests highly relevant information was provided for decisions about pain assessment. Analyses of treatment policies indicated almost twice as many nurses had significant opioid than non-opioid policies. For these providers, information contained in the scenarios was apparently more serviceable for decisions about use of opioid than non-opioid medications. These results may also be due to nurses’ greater familiarity with opioids for acute pain and/or the relative paucity of guidelines for non-opioids.
An advantage of this methodology is its ability to capture the decision-making process. Analyses at this level indicated VH demographic cues played a significant role in many nurses’ decisions. Most who used sex as a significant cue assigned higher assessment ratings to female VH. Sex played a relatively minor role in non-opioid decisions and a much larger role in opioid decisions, with females generally more likely to be administered medication.
These results are interesting considering evidence that females are at greater risk of having their clinical pain under-assessed [6,16], whereas in the experimental setting, females are judged to be experiencing greater pain than males . There is also substantial evidence females receive sub-optimal pharmacologic pain management relative to males [10,13,16,19,24,44,45]. The current study differs in many respects from previous investigations. A hybrid design was employed wherein participants made clinical ratings in an experimental context. Differences in participants are also noteworthy. We primarily enrolled female nurses, whereas previous clinical investigations consisted mainly of male physicians. We tested the hypothesis proposed by Bond  that provider stereotypic beliefs about gender and pain drive these effects. Results did not support this hypothesis; thus, it would appear that, as measured by the GREP, provider beliefs do not explain the sex difference in pain assessment and treatment. Additional work is needed to further elucidate the impact of patient and provider sex on pain-related decisions.
Race was also a contributor in many assessment policies. Most nurses who significantly weighted race assigned higher ratings to African-American VH. At the treatment level, race was significantly weighted by 9% and 6% of nurses, respectively, in their opioid and non-opioid decisions; most were more likely to administer medication to African-Americans. The relevant assessment literature is small and mixed, with some evidence African-Americans are at greater risk of pain underestimation  and some reports of no racial/ethnic differences . To our knowledge, this is the first study to find African-Americans received higher pain ratings than Caucasians. Perhaps participants were sensitive to medical disparities and took caution not to underestimate the pain of African-American VH. Perhaps female nurses exhibit less racial/ethnic biases. Multicultural competence is prominent in the nursing literature and emphasized in nursing education [27,36,43]. It is difficult to ascertain the meaning of these assessment differences. Did nurses discount Caucasian VH pain? Did they view African-Americans as less pain-tolerant? Regardless, it is interesting to note these ratings are consistent with self-report differences in experimental and clinical pain ratings .
The treatment disparities literature is more robust with data indicating minority individuals’ pain is under-treated relative to the dominant racial/ethnic group [6,9,16,15,47,53,54,59,63,61]. In addition to racism, a widely-held explanation is patient race serves as a proxy for the true operating variables (e.g., differential communication, SES, access) that drive these disparities. Aside from the cues of interest, all factors in the current scenarios were held constant. That African-American VH did not receive sub-optimal treatment ratings compared to Caucasians – and often received more aggressive care ratings – is more consistent with a “race-as-proxy” explanation than provider racism. Future investigations that include additional cues could address these issues.
The clinical context is also of consideration. Our post-surgical scenario may be less likely to elicit non-medical influences on treatment-related decisions than more ambiguous scenarios (e.g., migraine, sickle cell crises) . Study methodology is also important. Despite innovations, this study is properly classified as a vignette-based approach. Two recent vignette studies found patient race did not influence analgesic practice [14,58]. These designs may provide shelter from time and financial pressures of real practice that may make biased decisions more likely.
Age also emerged as a significant factor. Almost one-quarter of nurses weighted age in their assessment decisions; most provided higher ratings for elderly patients. Treatment ratings were less affected, and the results less consistent, particularly for non-opioids. The majority with a significant opioid policy were more likely to administer this medication to older VH. The literature documenting under-assessment of pain in older individuals is robust [18,17,20,26,25,29,35,38,56], whereas the treatment literature is small. Two studies found post-surgical elderly at risk of receiving less pain medication [48,14]. That the current results indicate elderly VH often received higher treatment ratings, particularly for opioids, is therefore surprising. Since older individuals are more likely to present with medical comorbidities that complicate pain management [34,67], these findings may be a consequence of the standardization of clinical presentation across VH. These results tentatively suggest when older and younger patients present with similar conditions, older patients are typically seen as experiencing greater pain, receive equivalent pharmacologic treatment, and may even receive more aggressive treatment.
Pain expression was also an important contributor to clinical ratings. All nurses with a significant assessment policy judged VH expressing high pain to be experiencing greater pain. Similar consistency was found for opioid treatment ratings, whereas less consistency was evident for non-opioids. This cue had the largest absolute effect on nurses’ assessment and treatment policies. The methodological implications of these results are encouraging. Although facial manipulations were guided by the FACS and empirically-validated pain expression [22,50], these findings further validate the manipulation. Clinically, it is reassuring that pain expression was the most salient cue in decision policies. This finding is consistent with work highlighting the importance of nonverbal pain expressions [3,22,31].
Use of patient demographic cues in pain decision-making, even if found in a small cluster of nurses, is significant given current practice. If the California-mandated 1:5 patient-to-nurse ratio for medical/surgical units  is adopted as a conservative estimate of the country-wide average, then a given nurse may assess and treat the pain of hundreds of patients each year. Thus, even a small propensity to use demographic characteristics in pain-related decision-making is highly important. Additionally, since precepting is important to nursing practice [32,57], if a veteran nurse conveys that patient demographic characteristics should be considered in pain-related decisions, the consequences could be dramatic. Thus, any one nurse with a propensity to weight demographic cues could have far-reaching influence.
This study highlights advantages of lens model approaches. For example, if analyses of VH age and opioid decisions were limited to a nomothetic approach, the conclusion would be: older VH received higher opioid ratings than younger VH, and this difference was moderate-to-large. Through idiographic analyses, we see that far from being ubiquitous, age-based differences in opioid treatment were present in 13% of nurses, and one nurse used this cue in the opposite direction. The resulting implications are very different.
Study limitations should be noted. First, analogue studies must consider their representativeness. We attempted to address this through VH technology. In our pilot work, over 70% of participants indicated the VH facial expressions were realistic depictions of pain. Over 90% considered the clinical scenario to be reflective of a real post-operative scenario. Although this suggests good representativeness, participants still made clinical decisions about a hypothetical VH in an experimental setting via computer. Moreover, some participants may have had difficulty understanding the nature of the task, which could have affected their decisions. Second, lens designs must restrict the number of available cues because of human cognitive limitations  and the geometric expansion of vignettes that must be viewed as cues increase . Thus, potentially important cues are excluded. Third, although participants’ demographic characteristics were consistent with national norms, it is possible they were unique in some way that influenced how they approached the study. Finally, this study may have been face-valid and elicited socially-desirable responses. Although many nurses expressed awareness of study hypotheses, they did not differentially weight VH cues relative to those who did not (data not presented).
In summary, this study demonstrated the use of VH technology and lens model methodology in the investigation of pain assessment and treatment disparities. Demographic and pain expression cues were significant contributors to many nurses’ pain-related decision policies. Continued research is needed to address the many questions raised herein, with the goal of improving the care of all patients in pain.
Support for this research was provided from Grant F31 (NS049675) to A. T. Hirsh from the National Institutes of Health, National Institute of Neurological Disorders and Stroke. Support was also provided, in part, from the National Institutes of Health, National Institute of Child Health and Human Development, National Center for Medical Rehabilitation Research (T32HD007424). The authors would also like to acknowledge Roger Fillingim, Ph.D. and William Perlstein, Ph.D. for their contributions to this research.
Patient presents with abdominal pain 22 hours post open appendectomy surgery. Patient reports that the pain began immediately following surgery. The pain is localized to the right lower abdomen in the area around the surgical incision. Patient also reports occasional generalized pain throughout the entire abdominal area. The pain limits patient’s ability to move around freely. Patient reports no prior surgical treatments and has current prescriptions for anti-inflammatory and analgesic medications.
The authors have no conflicts of interest to declare.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.