Nursing Staff, Patient, and Environmental Factors Associated with Accurate Pain Assessment

Article Outline

• Abstract
• Introduction
• Methods
  • Measures
    • Patient Characteristics
    • Nursing Staff Characteristics
    • Environmental Characteristics
  • Analysis
• Results
• Discussion
• Conclusions
• Disclosures and Acknowledgments
• References
• Copyright

Abstract 

Context

Although pain ranks highly among reasons for seeking care, routine pain assessment is often inaccurate.

Objectives

This study evaluated factors associated with nurses (e.g., registered) and other nursing support staff (e.g., licensed vocational nurses and health technicians) discordance with patients in estimates of pain in a health system where routine pain screening using a 0–10 numeric rating scale (NRS) is mandated.

Methods

This was a cross-sectional, visit-based, cohort study that included surveys of clinic outpatients (n=465) and nursing staff (n=94) who screened for pain as part of routine vital sign measurement during intake. These data were supplemented by chart review. We compared patient pain levels documented by the nursing staff (N-NRS) with those reported by the patient during the study survey (S-NRS).

Results

Pain underestimation (N-NRS<S-NRS) occurred in 25% and overestimation (N-NRS>S-NRS) in 7% of the cases. Nursing staff used informal pain-screening techniques that did not follow established NRS protocols in half of the encounters. Pain underestimation was positively associated with more years of nursing staff work experience and patient anxiety or post-traumatic stress disorder and negatively associated with better patient-reported health status. Pain overestimation was positively associated with nursing staff’s use of the full NRS protocol and with a distracting environment in which patient vitals were taken.

Conclusion

Despite a long-standing mandate, pain-screening implementation falls short, and informal screening is common.

Key Words: Pain measurement, veterans, outpatients

Back to Article Outline

Introduction 

Shortcomings in symptom management begin with a failure of symptom recognition, and previous literature indicates that providers consistently underestimate patients’ pain, although the reasons for this inaccuracy are poorly understood.¹, ², ³ Differences in pain measurement and settings make it difficult to compare findings across studies, although some studies present common metrics allowing for comparison. In one study reporting pain on a 0–10 numeric rating scale (NRS), nurses underestimated patients’ pain by 2.4 points on average in a triage setting and by 3.3 points in the treatment area of an emergency room.⁴ In a postoperative hospital setting, investigators found a 3-point average nurse underestimation of pain.⁵ An evaluation of a hospital burn unit found that when patients were at rest, nurse and research pain reports agreed 49% of the time, but agreement dropped to 30% during procedures.⁶ A different study of burn unit patients found a 38% agreement between nurse and research pain ratings.⁷

Despite the fact that pain is a frequent reason for seeking medical attention,⁸ it is often not assessed in ways consistent with current practice recommendations,⁹, ¹⁰, ¹¹, ¹² thus hindering successful pain management.¹³, ¹⁴, ¹⁵, ¹⁶ Recent studies suggest that underestimation of pain occurs despite the increasingly widespread practice of routine screening,⁸, ¹⁷ and persistent underrecognition impairs progress in improving the health-related quality of life for those with chronic conditions. To improve pain screening, the American Pain Society first coined the term “Pain as the 5th Vital Sign” to promote routine pain assessment, along with pulse, temperature, blood pressure, and respirations.¹⁸ In the United States, the Veterans Administration (VA) has required routine pain screening nationally since 2003, although many VA facilities adopted the 5th Vital Sign earlier.¹⁹ In VA outpatient practice, vital signs are routinely obtained by registered nurses (RNs) or other nursing support staff who have some relevant training (e.g., health technicians or licensed vocational nurses [LVNs]), whom, for the sake of brevity and to be consistent with other literature, we refer to collectively as “nursing staff.”

Some of the specific barriers to successful pain assessment include knowledge deficits and not using direct evaluation approaches.⁹, ¹¹, ²⁰, ²¹, ²² Patient barriers that might influence their pain reports include fears of addiction and side effects from pain medication, lack of pain education, and fears of disease progression.¹⁰, ²³, ²⁴ Although limited research suggests progress in addressing several of these factors,¹³ health professionals express major concerns with the large gap between knowledge of effective pain assessment and management approaches and the realities of clinical practice.¹¹, ¹⁷, ²⁵

Good pain measurement requires consistent use of valid, reliable instruments. The most commonly used pain screen is the 0–10 NRS, where 0 reflects no pain and 10 reflects the worst pain imaginable. Efforts to promulgate the “5th Vital Sign” have endorsed this approach. The psychometric properties of the NRS are among the best characterized of any pain-screening tool; the NRS is highly valid and reliable for detecting acute and chronic pain in many settings, populations, and time frames.¹⁵, ²⁶

Because the NRS is the most commonly used pain-screening tool, evaluating the factors associated with NRS variation in daily clinical practice settings may suggest targets to improve pain management. Various provider, patient, and environmental factors influence pain assessment. For example, several nurse characteristics have been associated with discordance between nurse and patient pain reports under research conditions. Better pain concordance is associated with fewer years of work experience,⁶, ¹² more pain management education,¹, ³ younger age,²⁷ and greater contact with patients.²⁷ Poorer concordance is associated with lower confidence in the ability to accurately assess pain and the time constraints faced in completing nursing tasks.²⁸ Qualitative evidence from inpatient settings suggests that nurses may use informal assessment approaches (e.g., “How are you?”) that directly contribute to underrecognition and suboptimal pain management.¹⁰, ¹¹ Nurses are more likely to underestimate pain when patients are suspected of or known to have a substance use disorder.²³, ²⁹ Additionally, when patients exhibit behaviors consistent with addiction, nurses are more likely to underestimate pain.²⁷

Patient characteristics also have been shown to contribute to discordance with nurse pain reports under research conditions. Improved concordance between the patients’ reports and the nurses’ assessments of patient pain is associated with female sex, older age, presenting with positive physical and behavioral signs of pain, being immobile, and having cancer.¹², ³⁰ In one study, patients with musculoskeletal conditions, abdominal pain, or cellulitis had lower concordance with nurse pain ratings.⁴ In addition, research on blood pressure screening suggests that the environment in which screening takes place may affect the accuracy of the screen results.³¹, ³² This may include patient reports of their comfort with the health care setting and nursing staff.

Although the previously cited studies document discordance in nurse and patient pain reports, these studies have been mostly limited to inpatient settings; most of the studies only considered nurse or patient factors associated with discordance as separate domains and did not consider them jointly, and none explored the role of the environment in contributing to pain assessment discordance. Additionally, the literature generally focuses only on the underestimation of pain and not the overestimation of pain, which may also affect the quality of subsequent care.

We aimed to characterize variability in pain-screening practices and pain ratings and understand a range of factors that contribute to variability in the crucible of daily clinical practice. Specifically, we evaluated the independent association of patient, nursing staff, and environmental factors with discordance between patient pain ratings routinely obtained by nursing staff with patient self-reports of pain provided during the study survey. We evaluated these relationships in the VA outpatient setting, where routine NRS estimation is mandated, to identify targets for quality improvement. We build on previous research by examining this variation in the ambulatory care setting and exploring factors associated with both the under- and overestimation of pain by nursing staff.

Back to Article Outline

Methods 

The VA funded the Helping Veterans Experience Less Pain Study to evaluate outpatient use of routine pain screening. We randomly sampled patients who attended one of 19 clinics from two hospitals and six affiliated sites in three large urban counties in Southern California (Los Angeles, Ventura, and Orange) between March 2006 and March 2007. Institutional review boards approved the study at each clinical site. To assure that we broadly addressed the care of chronic illness, we included primary care, urgent care, women’s health, cardiology, and oncology patients. Before surveying veterans, we asked all nurses and nursing support staff responsible for vital sign assessment to complete a self-administered survey that included questions about pain assessment attitudes and experience (described in more detail as follows).

Patients were randomly approached after provider visits and those who had their vital signs measured during the preceding encounter passed a brief cognitive screening test,³³ had intact hearing, spoke and understood English, had not participated previously, and agreed to medical record review, were invited to participate in the study. To ensure both healthy and frail participants, we interviewed every other patient who reported his or her health as excellent, very good, or good, and every patient who reported his or her health as fair or poor.

Data sources included an interviewer-administered patient survey, a self-reported survey of nursing staff, and patient chart review. To develop both patient and provider surveys, we reviewed existing pain surveys, conducted informal qualitative work, and completed cognitive interviews and field testing. The resulting 53-item patient survey included a number of standardized pain measures as well as demographic and health history questions. Trained research assistants administered the survey after informed consent.

We developed the 63-item self-administered nursing staff survey from both existing pain instruments³⁴ and team-derived items. Nursing staff survey items reflect established pain management content consistent with the U.S. Agency for Healthcare Research and Quality and the World Health Organization guidelines.³⁵, ³⁶ We obtained data about diagnoses and nursing staff documentation of pain assessments from the patient’s electronic medical record. A single abstractor with extensive professional abstraction experience reviewed charts under the supervision of two physicians.

Measures 

As the primary outcome, we evaluated agreement between the nursing staff’s documentation of the patient’s pain during routine vital sign intake (e.g., the “5th Vital Sign” or N-NRS) and the patient’s survey response using the same question (S-NRS) immediately after the clinic visit. Time between these measurements was approximately 30–45 minutes. Based on expert judgment of clinical significance of NRS difference,³⁷ we defined agreement between the pain reports as cases where the N-NRS was one point greater or less than the S-NRS. We defined a positive difference (e.g., N-NRS−S-NRS≥+2) as the case where the nursing staff overreported patient pain. Likewise, a negative difference (e.g., N-NRS−S-NRS≤−2) reflected nursing staff underestimation. The outcome was constructed as a three-level categorical variable reflecting underestimation, agreement, or overestimation.

Consistent with prior literature, we included independent variables in our model, reflecting patient, nursing staff, and environmental characteristics that may influence pain assessment and reporting. Although not explored in previous literature, our preliminary qualitative interviews identified factors related to the environment in which pain assessment takes place that might affect the concordance of pain reports. We also identified other measures included in our surveys that we hypothesized might influence the accuracy of pain measurement. These variables are described briefly in the following sections.

We asked patients if the nursing staff person asked about their pain and whether or not he or she used the NRS; we included an indicator variable reflecting their response in our models (yes/no). Clinical variables derived from the survey or chart review included dichotomous indicators for the presence of musculoskeletal conditions, cardiovascular disease, and cancer. Consistent with previous findings that nurses may underestimate pain among patients with known or suspected substance abuse problems, active or prior alcohol or drug abuse/dependence was constructed as a three-level variable (active, past history, or no history of abuse/dependence) from the chart review. In addition to these measures, we also considered the role that mental health may play in pain assessment. To that end, we included an indicator variable for depression based on the Patient Health Questionnaire-2³⁸, ³⁹, ⁴⁰ and an indicator for the presence of post-traumatic stress disorder (PTSD) or other anxiety disorder.⁴¹ We also controlled for patient age (in years); race (white, black, Hispanic, and others); hearing status; mild cognitive impairment (scoring zero to one wrong answers vs. two to three wrong answers on cognitive screener);³³ and self-reported health status (excellent/very good/good vs. fair/poor) from the patient survey and recruit logs. Because the sample comprised mostly men, we were unable to explore the role of patient sex in this analysis.

We derived the following variables directly from the nursing staff survey responses. Consistent with the literature, we included measures of work experience (in years), confidence with pain management, and other attitudinal measures. We constructed the Confidence in Pain Management Scale to assess self-efficacy and included items addressing degree of confidence in “detecting pain,” “determining pain severity,” “detecting depression among patients in pain,” and “evaluating pain in patients with substance abuse” (five-category Likert scale, very confident to very unconfident; range=0–16, Cronbach’s α=0.82). A higher score on the Confidence Scale reflected greater confidence in abilities. A Negative Pain Belief (“Attitude”) Scale consisted of four items assessing attitudes/beliefs about patients in pain: “Alcohol and drug abusers often exaggerate their pain,” “Patients don’t report their pain accurately,” “Patients cannot distinguish between acute and chronic pain,” and “The cause of pain is usually not clear” (five-category Likert scale, strongly agree to strongly disagree; range=0–16, Cronbach’s α=0.79). A higher score on the Attitude Scale reflected more negative views about patient reports of pain. In addition to these two scales, we included the following individual attitudinal measures: “My peers regard pain assessment as an important skill,” “The limited time I have available keeps me from thoroughly evaluating pain,” and “Nurses do not have any control over how a patient’s pain is treated.” Each of these statements was evaluated using a five-category Likert scale from “strongly agree” to “strongly disagree.” For analyses, we constructed dichotomous variables reflecting those who strongly agreed vs. all others.

We included a composite measure of the environment in which vitals were taken, derived from six questions to evaluate patient perception, including “How private was the room or space you were in?”; “How free from interruption was the time you spent with the nurse?”; “How easy was it for you to understand what the nurse said?”; “How rushed did your time with the nurse feel?”; “How distracted by other things did the nurse seem to be?”; and “How caring a person did the nurse seem to be?” Response options included “extremely,” “moderately,” “not very,” or “not at all.” We scaled items so that higher scores reflected positive evaluation and calibrated the scale to range from 0 to 100 (Cronbach’s α=0.70).

Analysis 

After data cleaning and descriptive analyses, we evaluated the association between the patient, nursing staff, and environment characteristics with the outcome of nursing staff/patient NRS agreement using multinomial logistic regression. We modeled the outcome where the N-NRS and the S-NRS were in agreement (±1 point) as the reference group. We conducted a sensitivity analysis to determine what effect patient, nursing, and environment characteristics had on the outcome, controlling for the portion of patients who reported that their pain changed between the time nursing staff screened them for pain and when they began the study survey.

The patient and nursing staff survey had few (<5%) missing values, but, because of the clustered nature of the data, where one nursing staff person could have evaluated the vitals for more than one study patient, a single missing value on the nursing staff survey could lead to multiple missing values in the analytical data set. Therefore, we imputed missing values on the nursing staff survey using regression techniques (for years of work experience, because it is highly correlated with age) or median substitution on five other variables missing between one and nine responses. Analyses were weighted for nonresponse to avoid bias in this regard. We used a probability-weighting scheme based on a logistic regression model using covariates available on both responders and nonresponders. We accounted for the clustering of patients within nursing staff through the use of robust standard error estimation.

Back to Article Outline

Results 

We approached 5,667 clinic patients, of whom 783 were eligible and 528 completed the interview (67.4% response rate). Of the 155 nursing staff approached, 145 completed the survey (93.5% response rate). After matching patient surveys with the survey of the nursing staff member who took their vital signs, we were left with 476 patients and 94 nursing staff. In 11 cases, nursing staff did not document any pain rating, leaving 456 patients with complete data on the dependent variable pain measures (N-NRS and S-NRS), who are the subject of this report.

Table 1 presents descriptive statistics. Patients were predominantly males (93.8%) and averaged 62 years of age (range=23–89 years). About half the patients were whites and one-quarter (24.3%) were blacks. Few respondents were hearing impaired, and approximately 16.6% had probable mild cognitive impairment. Approximately half reported their health as fair or poor, and nearly one-third had chart documentation of active alcohol or drug abuse/dependence. The most common comorbidities were musculoskeletal conditions (45.4%) and depression (40.2%). Although pain was evaluated in all encounters, less than half the patients reported that the nursing staff formally rated their pain using the NRS.

Table 1. Patient Characteristics (n=465)

Mean age (±SD)	61.3 (12.9)
% Male	93.8
Race
White	49.2
Black	24.3
Hispanic	15.7
Other	7.3
% Hearing impaired	9.9
% Cognitively impaired	16.6
Self-reported health status
% Excellent/very good/good	47.3
% Fair/poor	52.3
History of alcoholism/drug abuse/dependence
% With no abuse/dependence documented	57.8
% With abuse/dependence documented/not active	14.0
% With active abuse/dependence	28.0
Clinical conditions
Musculoskeletal	45.4
Cardiovascular	26.0
Cancer	11.2
% Depressed	40.2
% Post-traumatic stress disorder/anxiety disorder	21.5
% Nursing staff asked patient to rate pain on 0–10 scale	48.4

SD=standard deviation.

The 94 nursing staff included RNs (25.5%), LVNs (61.7%), and health technicians (12.8%); 87.2% were females and 42.6% were blacks (Table 2). The nursing staff members averaged 18 years of experience (range=1–41 years). The nursing staff reported high confidence in their pain management skills and a moderate number of negative beliefs about patients’ pain reports. The mean value of the Confidence in Pain Management Scale was 13.0 (±2.3, range=0–16) and that of the Negative Pain Belief Scale was 7.2 (±3.4, range=0–16). Almost 40% of all nursing staff strongly agreed that their peers regard effective pain management as an important skill. Fewer then 10% strongly agreed that time constraints keep them from correctly evaluating pain, and 10.6% strongly agreed that nursing staff have no control over pain management. The scale characterizing the patient evaluation of the environment in which nursing staff took their vital signs had a mean value of 87.7 (±13.8, range=0–100), reflecting a positive environment as viewed by the patient.

Table 2. Nursing and Environment Characteristics (n=94)

Male (%)	12.8
Race (%)
White	12.8
Black	42.6
Hispanic	18.1
Other/missing	26.5
Nursing staff training (%)
Registered nurses	25.5
Licensed vocational nurses/health technicians	74.5
Years of work experience (mean±SD)	17.6 (10.7)
Confidence in Pain Management Scale (mean±SD)a	13.0 (2.3)
Negative Pain Belief Scale (mean±SD)b	7.2 (3.4)
“Peers regard effective pain management as an important skill” (% strongly agree)c	39.4
“The limited time I have available keeps me from thoroughly evaluating pain” (% strongly agree)c	8.5
“Nurses have no control over pain treatment” (% strongly agree)c	10.6
Physical Environment Scale (mean±SD)d	87.7 (13.8)

SD=standard deviation.

Table 3 presents mean nursing staff-documented and patient-reported pain measures (N-NRS and S-NRS, respectively) as well as the distribution of our outcome measure. The mean S-NRS was 3.21 (±3.16), and the mean N-NRS was lower (2.30, ±3.27). Most of the time, the nursing staff’s pain documentation matched the patient’s subsequent report during the survey within one point (67.1%). Where discordance occurred, in most cases, nursing staff underestimated pain (25.4%). A smaller proportion (7.5%) of nursing staff overestimated pain.

Table 3. Study Outcomes (n=465 Patients)

Patient pain rating (mean±SD)	3.21 (3.16)
Nursing staff pain rating (mean±SD)	2.30 (3.27)
Nursing staff-patient pain concordance/discordance
% Patient rates pain higher than nursing staff	25.4 (n=118)
% Patient and nursing staff pain ratings match	67.1 (n=312)
% Nursing staff rates pain higher than patient	7.5 (n=35)

SD=standard deviation.

Table 4 presents the results of the multinomial logistic regression evaluating the concordance of nursing staff and patient pain measures as a function of patient, nursing staff, and environmental characteristics. We observed the strongest association with patient reports of being asked to report pain by the nursing staff person using the NRS as opposed to being informally queried about their pain. Nursing staff were significantly more likely to overestimate pain when they asked patients to rate their pain using the NRS (relative risk=4.23; 95% confidence interval=1.64, 10.0; P=0.003). The nursing staff were also less likely to underestimate pain when they asked patients to rate their pain using the NRS, but this event was less common, and the association was marginally significant (P=0.058). We found no significant difference in the frequency with which the nursing staff asked patients to rate their pain using the NRS by education level (e.g., RN vs. LVN or health technician) (data not shown). Patients with PTSD or other anxiety disorder were almost twice as likely to report higher pain levels as those documented by the nursing staff. Additionally, nursing staff were less likely to underestimate patient pain when the patient self-reported excellent, very good, or good health status (compared with fair or poor health status). Having any musculoskeletal condition was marginally associated with pain underestimation (P=0.06).

Table 4. Multivariate Results^a

	S-NRS>N-NRSb		N-NRS>S-NRSc
Patient, Nursing, and Environmental Characteristics	RRR	95% CI	RRR	95% CI
Self-reported health statusd	0.59	(0.37, 0.95)	1.03	(0.43, 2.45)
Alcohol/drug abuse/dependence (current)e	0.80	(0.48, 1.33)	1.31	(0.47, 3.68)
Alcohol/drug abuse/dependence (not active)e	0.56	(0.25, 1.24)	1.20	(0.37, 3.89)
Musculoskeletal condition	1.59	(0.98, 2.57)	1.30	(0.58, 2.94)
Cardiovascular condition	0.99	(0.56, 1.75)	1.79	(0.77, 4.17)
Any cancer	0.72	(0.36, 1.46)	0.47	(0.09, 2.39)
Depression	0.88	(0.54, 1.42)	0.36	(0.13, 1.02)
PTSD/anxiety disorder	2.00	(1.11, 3.6)	2.46	(0.96, 6.26)
Nurse asked patient to rate pain on 0–10 scale	0.59	(0.34, 1.02)	4.23	(1.64, 10.9)
Years of work experience	1.02	(1.01, 1.04)	1.01	(0.98, 1.05)
Confidence in Pain Management Scalef	1.00	(0.91, 1.09)	0.83	(0.69, 0.99)
Negative Pain Belief Scaleg	0.99	(0.92, 1.06)	0.95	(0.84, 1.08)
“Peers regard effective pain management as an important skill”h	1.21	(0.80, 1.84)	1.41	(0.56, 3.57)
“The limited time I have available keeps me from thoroughly evaluating pain”h	0.82	(0.31, 2.17)	1.21	(0.22, 6.73)
“Nurses have no control over pain treatment”h	1.18	(0.61, 2.30)	1.54	(0.47, 5.03)
Physical Environment Scalei	0.99	(0.98, 1.01)	0.97	(0.95, 0.99)

RRR=relative risk ratio; CI=confidence interval.

Nursing staff with more work experience were more likely to underestimate pain; an additional five years of experience was associated with a 10% increase in the likelihood of underestimation. However, this measure did not have a significant association with pain overestimation. Because of the high correlation between nursing education level and years of work experience, we report only the model that included work experience. Still, we learned from the models that include nursing education level (RN vs. LVN/health technician) that RNs were both more likely to over- and underestimate patient pain (data not shown). Nursing staff who scored higher on the Confidence Scale were less likely to overestimate pain, but there was no similar relationship with underestimation. We found no significant association between negative beliefs and pain under- or overestimation. None of the other attitudinal measures regarding peer perceptions of the importance of pain management, time constraints, and having no control over pain treatment were associated with either pain under- or overestimation. The measure reflecting the environment in which vital signs were documented was significantly and negatively associated with nursing staff pain overestimation but not associated with underestimation.

We conducted sensitivity analyses to evaluate the influence of the time lag between when the nursing staff documented patient vitals and when the patient completed the survey because of the possibility that pain levels might have improved or worsened (data not shown). We asked patients whether their pain changed since they entered the clinic that day, and 17.4% reported that their pain either improved or worsened. However, we found no significant association between whether or not the patient reported a change in his or her pain during the clinic visit and pain discordance. Anxiety disorders or PTSD were significantly more common among patients in whom pain changed during the visit, present in 33% of those who noted an improvement or worsening of their pain but in only 19% of those whose pain was unchanged (P=0.007). The association of PTSD or anxiety disorders with nurse-researcher discordance was diminished when we limited the analysis to cases in which pain did not change during the visit.

Back to Article Outline

Discussion 

Nursing staff reports of patient pain and patient self-reported ratings were often discordant, and nursing staff underestimated one-quarter of all patients’ pain; overestimation of pain was unusual. Still, we observed smaller differences than those previously reported, and nursing staff were somewhat more likely to agree with patient pain reports compared with those in prior studies. About half of all nursing staff used informal screening approaches rather than the NRS. Nursing staff and patient factors associated with greater underestimation included informal screening, more nursing work experience, and patient anxiety or PTSD disorders, whereas lower risk of underestimation was associated with better health status. Pain overestimation was associated with formal screening, and we identified a small but novel association of a better clinic environment with lower pain overestimation.

We considered the possibility that the use of informal compared with formal (e.g., NRS) approaches to pain screening reflected an implicit strategy of more clinically skilled nursing staff. However, there was no simple association between the use of this approach and years of experience, and nursing staff-patient pain concordance increased with higher-intensity pain (data not shown), suggesting that formal screening may have been driven by recognition of more severe pain, as noted in qualitative research characterizing inpatient nurse pain assessment.¹⁰ This finding also is supported by studies in which nurse agreement with patient pain assessments was higher when patients were at rest than when receiving potentially painful treatments.⁶ Additionally, nurses also may be more likely to agree or overestimate patient pain when the patient reports little or no pain.¹

Nursing staff were more likely to overestimate pain and, although marginally significant, were also less likely to underestimate pain when patients reported that their nursing staff used the NRS to formally rate pain. This reinforces the call for the implementation of standardized tools for pain assessment to maximize detection, because maximizing sensitivity is an appropriate goal for screening. Without these tools, nursing staff would rely on their own clinical judgment,⁴² which may be influenced by many of the preconceptions and attitudes about patients in pain noted earlier.

Having PTSD or another anxiety disorder was significantly associated with an increased risk of underestimating patient pain. This is cause for great concern, because anxiety conditions frequently co-occur with pain, and psychiatric disorders are significantly associated with increased disability.⁴³ Underestimation of pain for such persons may result in poor pain and anxiety treatment outcomes. Having a cancer diagnosis or cardiovascular condition did not influence the likelihood of nursing staff-patient pain discordance, although previous inpatient studies suggest that nursing staff may have difficulty assessing pain in these conditions.⁴, ³⁰ Possibly, interventions that inflict pain may be more likely to occur in such patients when hospitalized.⁶ In addition to increasing awareness among nursing staff of both physical and psychiatric conditions associated with significant pain, our findings suggest that clinical investigation of pain should be prompted by lower thresholds in the presence of psychiatric disorders.

The only nursing staff characteristic associated with pain rating discordance was years of work experience. Nursing staff with more years of work experience were more likely to underestimate pain. Previous research has reported a negative correlation between years of work experience and accuracy of pain assessment among nursing staff.⁶, ¹² In one study, nursing staff with more work experience were more likely to underestimate pain, whereas nursing staff with less work experience were more likely to overestimate pain.⁶ Although experience was not associated with the use of informal vs. formal pain screening, less experienced nursing staff may be more familiar with standardized pain assessments than those who started working before these assessments were widely used in 2003. There is a high correlation between years of work experience and nursing education level, and we also found in separate models that nursing education level was associated with increased pain report discordance. Recent nursing education also has emphasized pain management skills.¹¹, ⁴⁴

Our study suggests that the environment may have a small effect on the likelihood of overestimating but not underestimating patient pain. Nursing staff were less likely to document patient pain higher than his or her actual pain in cases where the patient favorably rated his or her environment. There is little research that explores the role of the environment in the accuracy of assessment and none dealing specifically with pain assessment. We hypothesized that a more favorably rated environment would have a negative association with pain discordance; this hypothesis was partially supported by our study results. Our finding is novel and warrants further investigation.

Limitations of our study include our focus on VA care settings only, although the VA’s experience may represent a “best case scenario,” given its relatively long history of routine pain screening and known excellence for chronic illness care.¹⁶, ¹⁹ Because our study was conducted in a predominantly male cohort, it may conservatively estimate the challenge of routine pain screening in women in whom population-based surveys suggest that pain is more severe.⁴⁵ It also focused on an outpatient population, although factors that cause pain differ among settings. Thus, these findings underscore the need for additional research on routine pain screening in other health care settings and patient populations.

Back to Article Outline

Conclusions 

The care of patients with pain requires a team approach, and nursing staff spend more time with patients than any other health care provider.¹¹ When comparing our findings with previous literature, this study suggests that health care has made modest progress in improving concordance between patient reports and provider documentation of pain. Additionally, these findings suggest that when standardized pain measures are not consistently used, pain discordance is more prevalent. Education programs should be developed to raise awareness that specific patient populations are at risk for poor pain assessments. These programs could also encourage the use of standardized tools for all pain assessments. Our findings are also relevant to the application of various screening tools in clinical practice and likely differences in sensitivity or specificity between research and clinical settings. Efforts to improve pain assessments will require multiple interventions targeted at patient, nursing staff, and environmental factors, to mitigate the suffering of those who live with pain.

Back to Article Outline

Disclosures and Acknowledgments 

Dr. Karl Lorenz was supported by a VA HSR&D Career Development Award. This study was funded by The Veterans Administration (IIR-030150). The funder was not involved in the conduct of the study or development of the submission. The authors declare no conflicts of interest.

Back to Article Outline

References 

1. Bergh I, Sjostrom B. A comparative study of nurses’ and elderly patients’ ratings of pain and pain tolerance. J Gerontol Nurs. 1999;25:30–36
   • View In Article
   • MEDLINE
2. Field L. Are nurses still underestimating patients’ pain postoperatively?. Br J Nurs. 1996;5:778–784
   • View In Article
   • MEDLINE
3. Hovi SL, Lauri S. Patients’ and nurses’ assessment of cancer pain. Eur J Cancer Care (Engl). 1999;8:213–219
   • View In Article
   • MEDLINE
   • CrossRef
4. Puntillo K, Neighbor M, O’Neil N, et al. Accuracy of emergency nurses in assessment of patients’ pain. Pain Manag Nurs. 2003;4:171–175
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (137 KB)
   • CrossRef
5. Bowman JM. Perception of surgical pain by nurses and patients. Clin Nurs Res. 1994;3:69–76
   • View In Article
   • MEDLINE
   • CrossRef
6. Choiniere M, Melzack R, Girard N, et al. Comparisons between patients’ and nurses’ assessment of pain and medication efficacy in severe burn injuries. Pain. 1990;40:143–152
   • View In Article
   • Abstract
   • Full-Text PDF (1130 KB)
   • CrossRef
7. Van der Does AJ. Patients’ and nurses’ ratings of pain and anxiety during burn wound care. Pain. 1989;39:95–101
   • View In Article
   • Abstract
   • Full-Text PDF (830 KB)
   • CrossRef
8. Mularski RA, White-Chu F, Overbay D, et al. Measuring pain as the 5th vital sign does not improve quality of pain management. J Gen Intern Med. 2006;21:607–612
   • View In Article
   • CrossRef
9. Herr K, Titler MG, Schilling ML, et al. Evidence-based assessment of acute pain in older adults: current nursing practices and perceived barriers. Clin J Pain. 2004;20:331–340
   • View In Article
   • MEDLINE
   • CrossRef
10. Dihle A, Bjolseth G, Helseth S. The gap between saying and doing in postoperative pain management. J Clin Nurs. 2006;15:469–479
    • View In Article
    • MEDLINE
    • CrossRef
11. McCaffery M, Pasero C. Pain: Clinical manual. St. Louis, MO: Mosby; 2002;
    • View In Article
12. McKinley S, Botti M. Nurses’ assessment of pain in hospitalised patients. Aust J Adv Nurs. 1991;9:8–14
    • View In Article
    • MEDLINE
13. Brockopp DY, Downey E, Powers P, et al. Nurses’ clinical decision-making regarding the management of pain. Int J Nurs Stud. 2004;41:631–636
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (197 KB)
    • CrossRef
14. Dworkin S, Sherman J. Relying on objective and subjective measures of chronic pain: guidelines for use and interpretation. In:  Turk D editors. Handbook of pain assessment. 2nd ed.. New York: The Guilford Press; 2001;p. 619–639
    • View In Article
15. Jensen M, Karoly P. Self-report scales and procedures for assessing pain in adults. In:  Turk D editors. Handbook of pain assessment. 2nd ed.. New York: The Guilford Press; 2001;p. 15–34
    • View In Article
16. U.S. Department of Veterans Affairs . Healthcare inspection: Veterans Health Administration Pain Management Initiative. Washington, DC: VA OIG; 2001;
    • View In Article
17. Krebs EE, Carey TS, Weinberger M. Accuracy of the pain numeric rating scale as a screening test in primary care. J Gen Intern Med. 2007;22:1453–1458
    • View In Article
    • CrossRef
18. Lynn J, Schuster JL, Kabcenell A. Improving care for the end of life: A sourcebook for health care managers and clinicians. New York: Oxford University Press; 2000;
    • View In Article
19. U.S. Department of Veterans Affairs . VHA Directive 2003-021: Pain management. Available from http://www1.va.gov/geriatricsshg/docs/PainManagement.pdf2003;Accessed August 8, 2008
    • View In Article
20. McCaffery M, Ferrell BR. Nurses’ knowledge of pain assessment and management: how much progress have we made?. J Pain Symptom Manage. 1997;14:175–188
    • View In Article
    • Abstract
    • Full-Text PDF (1208 KB)
    • CrossRef
21. Cohen MZ, Easley MK, Ellis C, et al. Cancer pain management and the JCAHO’s pain standards: an institutional challenge. J Pain Symptom Manage. 2003;25:519–527
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (70 KB)
    • CrossRef
22. Manias E, Bucknall T, Botti M. Nurses’ strategies for managing pain in the postoperative setting. Pain Manag Nurs. 2005;6:18–29
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (152 KB)
    • CrossRef
23. Randall-David E, Wright J, Porterfield DS, et al. Barriers to cancer pain management: home-health and hospice nurses and patients. Support Care Cancer. 2003;11:660–665
    • View In Article
    • MEDLINE
    • CrossRef
24. Wells N, Johnson RL, Wujcik D. Development of a short version of the Barriers Questionnaire. J Pain Symptom Manage. 1998;15:294–298
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (108 KB)
    • CrossRef
25. Hall-Lord ML, Larsson BW. Registered nurses’ and student nurses’ assessment of pain and distress related to specific patient and nurse characteristics. Nurse Educ Today. 2006;26:377–387
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (145 KB)
    • CrossRef
26. von Korff M. Epidemiological and survey methods: assessment of chronic pain. In:  Turk D editors. Handbook of pain assessment. 2nd ed.. New York: The Guilford Press; 2001;p. 603–619
    • View In Article
27. Elander J, Marczewska M, Amos R, et al. Factors affecting hospital staff judgments about sickle cell disease pain. J Behav Med. 2006;29:203–214
    • View In Article
    • MEDLINE
    • CrossRef
28. Kaasalainen S, Coker E, Dolovich L, et al. Pain management decision making among long-term care physicians and nurses. West J Nurs Res. 2007;29:561–580discussion 581–588
    • View In Article
29. Morgan BD. Knowing how to play the game: hospitalized substance abusers’ strategies for obtaining pain relief. Pain Manag Nurs. 2006;7:31–41
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (180 KB)
    • CrossRef
30. Loveman E, Gale A. Factors influencing nurses’ inferences about patient pain. Br J Nurs. 2000;9:334–337
    • View In Article
    • MEDLINE
31. Littler WA, Komsuoglu B. Which is the most accurate method of measuring blood pressure?. Am Heart J. 1989;117:723–728
    • View In Article
    • MEDLINE
    • CrossRef
32. Punzi HA. Why ambulatory blood pressure monitoring?. Am J Health Syst Pharm. 1998;55(Suppl 3):S12–S16
    • View In Article
    • MEDLINE
33. Callahan CM, Unverzagt FW, Hui SL, et al. Six-item screener to identify cognitive impairment among potential subjects for clinical research. Med Care. 2002;40:771–781
    • View In Article
    • MEDLINE
    • CrossRef
34. Pain treatment topics: clinical concepts—pain assessment tools. Available from http://pain-topics.org/clinical_concepts/assess.php2008;Accessed July 7, 2008
    • View In Article
35. Foley K. Appraising the WHO analgesic ladder on its 20th anniversary. Cancer Pain Release. Available from http://whocancerpain.bcg.wisc.edu/?q=node/862006;Accessed July 7, 2008
    • View In Article
36. Institute for Clinical Systems Improvement (ICSI) . Assessment and management of chronic pain. Bloomington, IN: Institute for Clinical Systems Improvement; 2007;87
    • View In Article
37. Dy SM, Asch SM, Naeim A, et al. Evidence-based standards for cancer pain management. J Clin Oncol. 2008;26:3879–3885
    • View In Article
    • CrossRef
38. Kroenke K, Spitzer RL, Williams JB. The Patient Health Questionnaire-2: validity of a two-item depression screener. Med Care. 2003;41:1284–1292
    • View In Article
    • MEDLINE
    • CrossRef
39. Kroenke K, Spitzer RL, Williams JB, et al. Anxiety disorders in primary care: prevalence, impairment, comorbidity, and detection. Ann Intern Med. 2007;146:317–325
    • View In Article
40. Lowe B, Kroenke K, Grafe K. Detecting and monitoring depression with a two-item questionnaire (PHQ-2). J Psychosom Res. 2005;58:163–171
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (169 KB)
    • CrossRef
41. Lang AJ, Stein MB. An abbreviated PTSD checklist for use as a screening instrument in primary care. Behav Res Ther. 2005;43:585–594
    • View In Article
    • MEDLINE
    • CrossRef
42. Florin J, Ehrenberg A, Ehnfors M. Patients’ and nurses’ perceptions of nursing problems in an acute care setting. J Adv Nurs. 2005;51:140–149
    • View In Article
    • MEDLINE
    • CrossRef
43. McWilliams LA, Cox BJ, Enns MW. Mood and anxiety disorders associated with chronic pain: an examination in a nationally representative sample. Pain. 2003;106:127–133
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (102 KB)
    • CrossRef
44. In:  Ferrell BR,  Coyle N editor. Textbook of palliative nursing. New York: Oxford University Press; 2001;
    • View In Article
45. Hardt J, Jacobsen C, Goldberg J, et al. Prevalence of chronic pain in a representative sample in the United States. Pain Med. 2008;9:803–812
    • View In Article
    • CrossRef