Racial/ethnic differences in staff-assessed pain behaviors among newly-admitted nursing home residents

Abstract

Context:

Nonverbal pain behaviors are effective indicators of pain among persons who have difficulty communicating. In nursing homes, racial/ethnic differences in self-reported pain and pain management have been documented. Few studies have examined racial/ethnic differences in nonverbal pain behaviors and pain management among residents with staff-assessed pain.

Methods:

We used the US national Minimum Data Set 3.0, and identified 994,510 newly-admitted nursing home residents for whom staff evaluated pain behaviors and pain treatments between 2010–2016. Adjusted prevalence ratios (aPR) and 95% confidence intervals (CI) estimated using robust Poisson models compared pain behaviors and treatments across racial/ethnic groups.

Results:

Vocal complaints were most commonly recorded (18.3% non-Hispanic Black residents, 19.3% of Hispanic residents, 30.3% of non-Hispanic White residents). Documentation of pain behaviors was less frequent among non-Hispanic Black and Hispanic residents than non-Hispanic White residents (e.g., vocal complaints: aPR _Black: 0.76; 95% CI: 0.73 – 0.78; with similar estimates for other pain behaviors). Non-Hispanic Blacks (47.3%) and Hispanics (48.6%) were less likely to receive any type of pharmacologic pain intervention compared to non-Hispanic White residents (59.3%) (aPR _Black: 0.87, 95% CI: 0.86–0.88; aPR _Hispanics: 0.87, 95% CI: 0.84–0.89).

Conclusions:

Among residents requiring staff assessment of pain because they are unable to self-report, nursing home staff documented pain and its treatment less often in Non-Hispanic Blacks and Hispanics than in non-Hispanic White residents. Studies to understand the role of differences in expression of pain, explicit bias, and implicit bias are needed to inform interventions to reduce disparities in pain documentation and treatment.

INTRODUCTION

Pain is highly prevalent among nursing home residents, with between 35% to 80% of residents experiencing pain at some point during their stay.(1,2,3) Undiagnosed pain leads to ineffective pain management which is associated with several adverse outcomes, including depression,(4) decreased quality of life,(5) increased functional limitations,(6) and behavioral disturbances such as aggression and agitation.(7,8) The identification of pain among nursing home residents remains a challenge(9,10,11) and is even more difficult among patients who cannot self-report pain due to communication difficulties. For residents who cannot self-report, the use of behavioral pain scales is a valid and reliable alternative to self-reported pain.(12,13,14) Nonverbal behaviors such as facial expressions, vocalizations, body movements, and changes in mental status are effective indicators of pain for those unable to self-report.(1,12–14)

Available evidence suggests that the experience of pain varies between racial and ethnic groups, with racial/ethnic minorities reporting higher levels of pain compared to non-Hispanic Whites across various clinical conditions and health care settings.(15,16,17,18) Few studies have examined differences in the expression of nonverbal pain behaviors among different racial/ethnic groups.(18,19,20,21) Ford et al. (2015) noted an increased use of pain words among Caucasians, when compared to Hispanics, and African Americans. Contemporary national studies in nursing home studies exploring health disparities in pain and its management in those unable to self-report pain do not exist.

Differences in the diagnosis of pain among race/ethnic groups contributes to the racial disparities observed in treatment of pain.(9–11, 22, 23) False beliefs about biological differences between Blacks and Whites has been shown to predict racial bias in pain perception and treatment recommendation accuracy.(24) Examining whether these differences persist among nursing home residents with staff-assessed pain behaviors, and whether the treatment of pain differs by race/ethnicity among those with staff-reported pain will contribute to the existing literature. The primary objective of this study was to describe race/ethnic differences in staff-assessed nonverbal pain behaviors and pain management among a national sample of newly admitted nursing home residents. Previous studies have shown racial/ethnic differences in pain prevalence, and treatment in nursing home residents.(11) Thus, we hypothesized that relative to non-Hispanic White nursing home residents, nursing home staff would document pain behaviors less often in non-Hispanic Black and Hispanic residents. We also hypothesized that pain management would be less prevalent in non-Hispanic Black and Hispanic residents relative to White residents.

METHODS

The Institutional Review Board of the University of Massachusetts Medical School Approved this study.

Data Source

The Minimum Data Set (MDS), version 3.0, is a comprehensive data set of mandatory screening, clinical, and functional status assessments of residents in all Medicare- or Medicaid-certified nursing homes.(25) The MDS is typically completed by specially-trained registered nurses with appropriate input from other professionals on the interdisciplinary healthcare team.(25) It is a widely used data source for research with excellent interrater reliability and clinical validity.(26,27)

Sample

Residents were identified as new admissions if the staff indicated that they were completing an admission assessment within 14 days after admission. We focused on residents with staff assessed pain measures. We restricted the sample to racial/ethnic groups with sample sizes able to permit informative analyses (Hispanic, non-Hispanic Black, non-Hispanic White). We included residents without missing information for pain behaviors, pain management, and selected covariates (see covariate section below).

Pain Indicator Variables

Staff-reported pain measures are documented for nursing home residents who are unable or unwilling to self-report pain.(27) The MDS manual outlines the process for determining whether a staff assessment for pain should be conducted, and the sources of information to use in completing the staff-assessed pain section.(25,27). First, staff are instructed to review the medical record for documentation of pain indicators within the 5 days preceding the assessment. Second, the staff conducting the pain assessment are instructed to confirm the information included in the medical record with the direct care staff working most closely with the resident during activities of daily living (across all shifts). This is required because all indicators of pain may not have been documented in the medical record. Direct care staff are also asked about the presence of each indicator. Third, the staff member completing the pain assessment is instructed to observe the resident “during care activities”, noting that these indicators are “more sensitive if the resident is observed during activities of daily living or wound care”. If present, the staff record the pain indicators observed.

Four behavioral pain indicators are included on the MDS 3.0. The presence of nonverbal sounds includes “crying, whining, gasping, moaning, or groaning”. Vocal complaints of pain include language such as “that hurts,” “ouch,” or “stop”. Facial expressions are documented if the staff observe grimaces, winces, wrinkled foreheads, furrowed brows, clenched teeth or jaw (as examples). Staff document protective body movements or postures if they observe (or it is documented in the medical record or reported by direct care staff) movements such as bracing, guarding, rubbing or massaging a body part/area, and clutching or holding a body part during movement. Staff also document if none of these signs or behaviors are observed or reported during the 5-day look-back period. The MDS 3.0 validation study noted that the average agreement (kappa) between specially-trained research nurses and nursing facility nurses for staff-observed pain behaviors was between 0.94 and 0.96.(28) The frequency with which the resident showed possible evidence of pain within the past five days (one to two days, three to four days, and daily) was also documented.(27)

Pain Management

We created five binary variables to characterize pain management over the 5-day look-back period using the items documented in Section J of the MDS 3.0 (25): 1) received scheduled pain medications; 2) received or was offered and declined pro re nata (PRN) pain medications; and 3) received non-pharmacologic intervention. Based on the first two items, we created a binary variable to characterize any pharmacologic pain management (Yes, if scheduled and/or PRN medications were received, No otherwise). We also created binary variables for receipt of:1) Scheduled and PRN pain medications; 2) Scheduled medications (without PRN medications); 3) PRN medications (without scheduled medications); and 4) Non pharmacologic intervention (received any non-medication intervention).

Race/ethnicity

Section A of the MDS 3.0 documents the race of residents. Staff are instructed to record race/ethnicity of nursing home residents through self-report or by family member or significant other for residents unable to respond. If none of these options are available, staff observation or medical record document is used. Race/ethnicity was categorized as American Indian/Alaska Native, Asian, Black/African-American, Hispanic/Latino, Native Hawaiian/Other Pacific Islander, and/or White.(27) We then categorized residents as non-Hispanic Black, Hispanics of any race(s), or non-Hispanic White after excluding residents of other race/ethnicities. Non-Hispanic Whites served as the reference group to ensure comparability with available studies.(11,18, 21, 29)

Covariates

Various sociodemographic and clinical characteristics that have been shown to be associated with the report and management of pain were included as confounders.(30,31,32) These included age in years (50–64, 65–74, 75–84, and ≥ 85 years), and sex. Clinical characteristics included rejects care (≥1 day versus none), hospice care in past 14 days, diagnosis of dementia/Alzheimer’s disease, anxiety disorder, and depression. Limitations in activities of daily living (ADL) were determined using four items on the MDS ADL short form – personal hygiene, toilet use, locomotion on unit, and eating – to create a seven category ADL performance scale (range 0–6).(33) ADL function was then categorized as minimal limitation (0–2), moderate limitation (3–4), or severe limitation (5–6). Cognitive function was determined by combining scores from the Brief Interview of Mental Status (BIMS; range 0–15) for residents who could communicate(34) or Cognitive Performance Scale (CPS; range 0–6) for residents who could not.(35) Level of cognitive impairment was then categorized as none/mild (BIMS 13–15 or CPS 0–2), moderate (BIMS 8–12 or CPS 3–4), or severe/very severe (BIMS 0–7 or CPS 5–6). Potentially painful health conditions such as cancer, heart failure, coronary artery disease, venous thromboembolism, peripheral vascular/arterial disease, and inflammatory bowel disease/ ulcerative colitis were also included as covariates because pain is commonly reported among individuals with these conditions.(36,37,38) Marital status (married vs. never married, separated, divorced, or widowed), source of admission [community (private home/apt., board/care, assisted living, group home), another nursing home or swing bed, acute hospital, or other (psychiatric hospital, inpatient rehabilitation facility, intellectual disability / developmental disability facility, hospice, long term care hospital, other)], ability to be understood by others (understood or usually understood versus sometimes/rarely never understood), understands others (understands or usually understands versus sometimes/rarely never understands) were excluded because there was a large number of missing observations, or they were correlated with other covariates.

Statistical Analysis

Statistical testing for this large dataset was likely to show significant results for small differences in prevalence in the various groups. Instead, we considered frequency distributions with differences ≥ 5% to be noteworthy. For each outcome of interest, a modified Poisson modeling approach, implemented using generalized estimating equations to account for clustering within nursing homes, was used to estimate adjusted prevalence ratios (aPR) and corresponding 95% confidence limits (CI).(39,40) We also stratified the analysis by gender.

RESULTS

Characteristics of Study Sample

The sample selection procedure is outlined in Figure 1 with 994,510 residents included in the final sample. Overall, the majority of newly-admitted residents were non-Hispanic White (78.9%), women (60.3%), aged ≥75 years (71.5%) and were admitted from an acute care hospital (79.8%). Most had moderate (46.1%) or severe (45.5%) limitations in ADL function.

Figure 1. Sample selection procedure for newly-admitted nursing home residents assessed for pain behaviors by staff.

While 42.5% of non-Hispanic White residents were ≥ 85 years, 25.4% of non-Hispanic Black and 30.6% of Hispanic/Latino residents were ≥ 85 years (Table 1). Across all racial/ethnic groups, the majority entered the nursing home from an acute care hospital. Limitation in ADL function was common with 59.7% of non-Hispanic Blacks, 57.6% of Hispanics, and 41.8% of non-Hispanic Whites experiencing severe compromise in ADL function. Cognitive impairment was common with 30.9% of all residents with severe/very severe cognitive impairment (non-Hispanic Blacks: 38.4%, Hispanics: 38.1%, non-Hispanic Whites: 29.0%). The prevalence of potentially painful conditions appeared to be similar across racial/ethnic groups. Some notable exceptions were arthritis (16.0% of non-Hispanic Blacks, 14.3% of Hispanics, and 21.4% of non-Hispanic Whites) and fractures (5.5% of non-Hispanic Blacks, 9.8% of Hispanics, and 15.9% of non-Hispanic Whites).

Table 1.

Demographic and clinical characteristics of newly admitted nursing home residents with staff-assessed pain, by racial/ethnic groups (n=994,510)

Characteristic	Non-Hispanic Black (n=146,058)	Hispanic (n=63,673)	Non-Hispanic White (n=784,779)
	Percentage*
Age, years                50–64	21.8	15.5	9.5
65–74	23.2	20.4	15.4
75–84	29.6	33.5	32.6
85+	25.4	30.6	42.5
Women	58.4	55.0	61.2
Married**	25.0	35.3	36.7
Admitted from: Community	5.8	7.8	10.4
Acute hospital	84.4	81.9	78.8
Rejects care	11.7	10.6	16.1
Activities of daily living limitations: Moderate	32.7	34.9	49.6
Severe	59.7	57.6	41.8
Cognitive Impairment: Moderate	33.7	36.7	38.3
Severe	38.4	38.1	29.0
Makes self understood**	45.0	42.6	55.8
Understands others**	46.6	43.1	55.2
Diagnoses: Dementia/Alzheimer’s	43.7	47.5	51.7
Anxiety disorder	11.3	19.9	23.0
Depression	19.2	27.1	32.8
Hospice	5.5	7.8	7.7
Skilled nursing facility stay	53.9	48.3	58.7
Potentially painful conditions
Cancer	9.2	6.9	9.3
Cardiovascular
Heart Failure	18.0	14.9	17.7
Coronary Artery Disease	18.5	20.3	22.7
Venous Thromboembolism	5.1	3.6	3.0
Peripheral Vascular/Arterial Disease	8.0	6.8	5.9
Inflammatory Bowel Disease/Ulcerative Colitis	0.4	0.4	0.9
Skin problems**	17.2	20.4	20.7
Musculoskeletal:
Arthritis	16.0	14.3	21.4
Osteoporosis	3.9	7.7	11.2
Fracture (hip and other)	5.5	9.8	15.9
Urinary Tract Infection (last 30 days)	15.4	15.9	18.0

^*May not total 100% due to rounding.

^**Missing data: marital status (non-Hispanic Black=6,902, Hispanic=2,853, non-Hispanic White=19,218), makes self understood (non-Hispanic Black=1,159, Hispanic=386, non-Hispanic White=3,981), understand others (non-Hispanic Black=1,693, Hispanic=487, non-Hispanic White=5,567)

^**Includes foot problems (infection of the foot, diabetic foot ulcers, other open lesions of the foot); burns (second or third degree); surgical wounds; and open lesions other than ulcers, rashes, and cuts.

Pain behaviors

The most common pain behavior documented was vocal complaints followed by facial expressions of pain, nonverbal sounds and protective body movements or postures (Table 2). Figure 2 shows the frequency of pain behaviors (none, 1–2 days, 3–4 days, daily) by racial/ethnic group. Among non-Hispanic Whites, 12.2% had pain behaviors documented daily, compared to 7.0% of non-Hispanic Blacks, and 7.2% of Hispanics. Supplemental Figure 1 demonstrates that the results are similar in men and women.

Table 2.

Association between race/ethnicity and pain behaviors among newly admitted nursing home residents (n=994,510)

	Non-Hispanic Black (n=146,058)	Hispanic (n=63,673)	Non-Hispanic White (n=784,779)
	Vocal complaints
Percent	18.3	19.3	30.3
Prevalence ratio (95% Confidence Interval):
Crude	0.60 (0.58 – 0.63)	0.63 (0.60 – 0.67)	1.00 (reference)
Adjusted*	0.76 (0.73 – 0.78)	0.76 (0.73 –0.80)	1.00 (reference)
	Facial expressions
Percent	15.7	179	21.3
Prevalence ratio (95% Confidence Interval):
Crude	0.74 (0.72 – 0.76)	0.84 (0.80 – 0.88)	1.00 (reference)
Adjusted*	0.77 (0.76 – 0.79)	0.83 (0.79 – 0.86)	1.00 (reference)
	Non-verbal sounds
Percent	9.8	10.0	13.4
Prevalence ratio (95% Confidence Interval):
Crude	0.74 (0.71 – 0.76)	0.75 (0.72 – 0.79)	1.00 (reference)
Adjusted*	0.75 (0.73 – 0.77)	0.72 (0.69 – 0.75)	1.00 (reference)
	Protective body movements or postures
Percent	4.6	5.5	8.4
Prevalence ratio (95% Confidence Interval):
Crude	0.55 (0.52 – 0.57)	0.65 (0.61 – 0.69)	1.00 (reference)
Adjusted*	0.66 (0.63 – 0.68)	0.71 (0.67 – 0.75)	1.00 (reference)
	No signs observed or documented
Percent	68.2	66.7	55.8
Prevalence ratio (95% Confidence Interval):
Crude	1.22 (1.21 – 1.24)	1.19 (1.17 – 1.22)	1.00 (reference)
Adjusted*	1.13 (1.12 – 1.14)	1.14 (1.12 – 1.16)	1.00 (reference)

^*Model adjusted for age, sex, rejects care, hospice, anxiety, dementia/Alzheimer’s disease, depression, limitations in activities of daily living, level of cognitive impairment, and all painful conditions

Figure 2. Frequency of any pain behaviors by racial/ethnic groups.

After adjustment for sociodemographics, conditions that may affect the experience of pain, and the presence of potentially painful conditions, the prevalence of each pain behavior was less frequently recorded among non-Hispanic Black and Hispanic residents than in non-Hispanic Whites [e.g., vocal complaints aPR _Black: 0.76 (95% CI: 0.73–0.78); aPR _Hispanic: 0.76 (95% CI: 0.73–0.80)]. Non-Hispanic Blacks and Hispanics were more likely than non-Hispanic Whites to have no signs of pain documented [aPR non-Hispanic _Blacks: 1.13 (95% CI: 1.12–1.14); aPR _Hispanics: 1.14 (95% CI:1.12–1.16)] (Table 2). Supplemental Tables 1 and 2 show that the results are similar in men and women.

Pain Management

Use of pharmacologic pain management was common (Table 3). Non-Hispanic Blacks and Hispanics were less likely than non-Hispanic Whites to receive any pharmacological pain treatment (aPR _{non-Hispanic Blacks}: 0.87, 95% CI: 0.86–0.88), aPR _Hispanics: 0.87, 95% CI: 0.84–0.89), and any non-pharmacologic intervention (aPR _{non-Hispanic Blacks}: 0.71, 95% CI: 0.69–0.74), aPR _Hispanics: 0.81, 95% CI: 0.77–0.85). Similar race/ethnic differences were observed for the receipt of scheduled and PRN medications, and PRN medications.

Table 3.

Type of pain management intervention by race/ethnicity among newly admitted nursing home residents (n=994,510)

	Non-Hispanic Black (n=146,058)	Hispanic (n=63,673)	Non-Hispanic White (n=784,779)
	Any pharmacologic pain management
Percent	47.3	48.6	59.3
Prevalence ratio (95% Confidence Interval):
Crude	0.80 (0.79 – 0.81)	0.82 (0.80 – 0.84)	1.00 (reference)
Adjusted*	0.87 (0.86 – 088)	0.87 (0.84 – 0.89)	1.00 (reference)
	Scheduled and PRN pain medications
Percent	11.0	11.3	17.1
Prevalence ratio (95% Confidence Interval):
Crude	0.64 (0.62 – 0.66)	0.66 (0.63 – 0.70)	1.00 (reference)
Adjusted*	0.76 (0.74 – 0.78)	0.75 (0.71 – 0.78)	1.00 (reference)
	Scheduled medications
Percent	11.3	10.9	10.7
Prevalence ratio (95% Confidence Interval):
Crude	1.05 (1.01 – 1.10)	1.02 (0.97 – 1.06)	1.00 (reference)
Adjusted*	1.10 (1.06 – 1.14)	1.04 (1.00 – 1.08)	1.00 (reference)
	PRN medications
Percent	25.1	26.4	31.6
Prevalence ratio (95% Confidence Interval):
Crude	0.79 (0.78 – 0.81)	0.84 (0.81 – 0.87)	1.00 (reference)
Adjusted*	0.85 (0.83 – 0.87)	0.87 (0.84 – 0.90)	1.00 (reference)
	Non pharmacologic intervention
Percent	16.2	19.1	25.1
Prevalence ratio (95% Confidence Interval):
Crude	0.65 (0.62 – 0.67)	0.76 (0.72 – 0.80)	1.00 (reference)
Adjusted*	0.71 (0.69 – 0.74)	0.81 (0.77 – 0.85)	1.00 (reference)

^*Model adjusted for age, sex, rejects care, hospice, anxiety, dementia/Alzheimer’s disease, depression, limitations in activities of daily living, level of cognitive impairment, and all painful conditions

DISCUSSION

Among newly admitted nursing home residents for whom staff assessed their pain, we found that each documented pain behavior was less frequently recorded for non-Hispanic Black residents and for Hispanic residents than for non-Hispanic White residents; the differences persisted after adjustment for a variety of factors including sociodemographic characteristics, and potentially painful and other clinical conditions. Further, non-Hispanic White residents were more likely to have had daily pain behaviors recorded as compared to Hispanic/Latinos and non-Hispanic Black residents. Our findings were similar in men and women which is consistent with research on pain subgroups in nursing home residents.(41)

We found that the most common indicators of pain documented in newly admitted nursing home residents with staff-assessed pain were vocal complaints and facial expressions. This is in alignment with previous studies where pain words/noises and pain faces were the most prevalent pain behaviors among nursing home residents.(21,42) Protective body movements or postures were the least frequently noted indicators of pain. This suggests that facial expressions of pain and pain vocalizations are more readily observed than other nonverbal pain behaviors.(21,41)

Our findings build on previous research showing that non-Hispanic Black and Hispanic residents were less likely to have “any” pain documented compared to non-Hispanic White residents.(18,22,43,44,45 ) Ford et al. (2015) evaluated the ability of certified nursing assistants (CNAs) to identify nonverbal pain behaviors while viewing videotapes of nursing home residents with dementia completing activities of daily living.(21) They found differences in the CNAs’ identification of “pain words” among Caucasians (37.7%), Hispanics (25%), and African Americans (4.2%). Furthermore, bracing and restlessness as indicators of pain were only observed by CNAs among Caucasian residents. In studies that have combined self-reported and staff-assessed pain, “any pain” is documented less frequently in racial/ethnic minorities.(11,43,46,47) These findings are consistent with the current study which showed that pain behaviors are documented less frequently in Black and Hispanic residents relative to non-Hispanic White residents.

Our study was unable to further explore reasons for the racial/ethnic differences observed, but others have posited that implicit bias by health care providers may either effect perceptions and clinical decisions or influence patient–provider communication.(48) Implicit bias has been shown to be associated with markers of poor visit communication and poor ratings of care,(49) with other evidence showing less patient engagement in medical dialogue.(50) Explicit bias arising from erroneous beliefs about biological differences in pain experience and expression between Blacks and Whites (24,51) may also be a contributor to the observed disparities. Interestingly, results from a recent randomized field experiment which used standardized patients differing only by race and activation, showed no race effect in pain assessment in the setting of advanced cancer.(52) Similar findings have not been demonstrated in nursing home residents with staff-reported pain. Our study found non-Hispanic Black and Hispanic residents were also less likely to have pharmacological and non-pharmacological pain management strategies documented. Thus, it is imperative to evaluate the extent to which implicit bias may contribute to the racial/ethnic differences observed in nursing home settings.

Culture may influence individual pain-related behavior and emotions and providers’ interpretations of those behaviors. Differences in cultural and social norms impact the expression of emotions among different racial/ethnic groups,(53,54) with some cultures promoting more stoic expressions of pain,(19) making accurate perception of pain (55) more difficult for nursing home staff. Providers whose cultural experience and knowledge differs from that of the residents they care for may interpret resident emotional expressions and behaviors through a different lens, if the provider is not cognizant of this difference, and does not account for it during the assessment.(56,57,58) Differences in attitudes and beliefs about the origin and role of pain influence how individuals perceive and react to their pain, and the pain of others.(54–56) Such differences can impact the provider-patient interaction during the pain assessment process.

While a recent meta-analysis described the patient, observer, and assessment level factors that affected the accuracy of pain assessment, the role of race/ethnicity was not discussed.(59) We believe differences in pain behaviors in racial/ethnic groups observed in our study could be due to a variety of factors. Providers are often unaware of their own implicit biases.(60) This could negatively affect the recognition of pain. Implicit and explicit bias can also influence how care providers interact with their patients.(57,61,62) For instance, one study found that health care providers assumed that White residents were more sensitive to pain than African American and Hispanic residents.(63) The role of racial/ethnic concordance between residents and direct care nursing home staff must be considered because shared language and culture may promote better pain assessment, as has been suggested with patient-physician interactions.(64,65) In our study, we lack information on the MDS to further our understanding of which of these possible mechanisms may explain the observed racial/ethnic differences in pain behavior recognition.

The differences in documented pain behaviors and pain management in racial/ethnic groups may be related to the resources available in the nursing homes in which racial/ethnic minorities typically reside. Nursing homes are highly racially segregated, and this may strongly influence care quality,(66) and adversely impact racial/ethnic disparities on quality outcomes in nursing homes.(67) Care quality is also influenced by the racial/ethnic composition of nursing homes.(68) Racial/ethnic minorities may live in nursing homes where staff have less time available to conduct thorough MDS assessments. The MDS manual provides a process for staff-assessment of pain that includes talking to direct care staff, family members, and review of notes. How thorough this process is may be associated with the time available to the nursing home staff.(69) The amount of time spent completing the MDS assessments are not available and as such we were unable to explore this hypothesis further.

Strengths/limitations

The MDS is a large standardized data set that includes nearly all U.S. nursing home residents. The use of a 5-day look-back period to assess the occurrence and frequency of pain behaviors ensures that any pain within that period is captured, since pain may not be present at the time of the assessment. However, because pain behaviors are staff assessed, pain behaviors during the look-back period that were not documented in the medical record or recalled by proxies during the assessment may be missed. The MDS 3.0 does not include details regarding pain severity and frequency of individual pain behaviors for those with staff-assessed pain. Thus, there are no guidelines on how to translate specific pain behaviors into indicators of pain severity to guide pain management. We had no information on the race/ethnicity of the nursing home staff and as such were unable to explore the extent to which racial/ethnic concordance between staff and residents reduced disparities in recognition of pain. Additionally, facility-level variables such as percent of residents on Medicaid, staffing level and other measures of resource availability were not included in the analysis. Nevertheless, the results of this study have important implications for the assessment of pain behaviors among nursing home residents of different racial/ethnic groups. This study was carried out with a sample of newly admitted nursing home residents. As such, lack of pain behaviors observed may be due to effective pain treatments received from the point of referral, instead of care provided in the nursing home. Future work can focus on how the race/ethnicity of providers and residents interact in the interpretation of nonverbal pain behaviors. Whether the observed behaviors reflect pain or discomforts (e.g., overstimulation, uncomfortable environment, constipation) cannot be discerned from these data. Providers’ perceptions of the pain behaviors of an ethnically diverse group of nursing home residents, and how that impacts treatment during a stay in the nursing home is also another area that needs further elucidation.

CONCLUSIONS

In nursing homes, the prevalence of pain is high.(70) We found that among residents unable to self-report, the documentation of pain behaviors is less in racial/ethnic minorities relative to non-Hispanic White residents. Studies to understand the extent to which the disparity in pain documentation and management by race/ethnicity is due to implicit bias, differences in pain expression, poorer documentation of the expressed pain by racial/ethnic minority nursing home residents, and/or explicit bias are warranted. Given that documentation of pain in nursing home residents is the first step in putting together appropriate pain management plans, such studies are needed to inform best practices in recognizing and treating pain in racially/ethnically diverse nursing home residents unable to self-report pain. Improvements in the recognition of pain and effective pain management will help improve quality of life (71) and ensure dignity in care.(72)

KEY MESSAGE.

Nursing home residents who cannot communicate express pain using facial expressions, sounds, and body language. Black and Hispanic residents had pain documented and received medications to manage pain less frequently than White residents. Identifying factors causing disparities in pain recognition and management among racial/ethnic groups in nursing homes is needed.