Putn, 44 (1991) 157-163 0 1991 Elsevier Science Publishers ADONIS 030439599100070R

157 B.V. 0304-3959/91/%03.50

PAIN 01730

Interethnic differences in pain perception Howard P. Greenwald Health Services Administration

Ptogram,

(Received

School of Public Administration,

9 May 1990, revision received

University of Southern California,

30 July 1990, accepted

10 August

Los Angeeles, CA (U.S.A.)

1990)

Summary

While several investigators have reported relationships between ethnic background and expression of pain, such relationships are in fact highly problematical. Few studies of pain and ethnicity have used quantitative measures of pain combined with multivariate methods of data analysis. Most have focussed on populations which, unlike many in the United States today, are characterized by highly distinct ethnic groups. The study reported here interviewed 536 persons recently treated for forms of cancer known to cause significant pain. Pain was assessed using standard, well validated inst~ments, including Graphic Rating Scales anchored in several alternative time-frames and the McGill Pain Questionnaire. The study took place in an area with a low proportion of recent immigrants and only small concentrations of distinct ethnic minorities. No statistically significant relationships were observed between ethnic identity and measures of pain sensation. However, pain described in affective terms according to the McGill Pain Questionnaire did vary among ethnicities. This observation suggests that cultures associated with specific ethnic identities still condition individual expression of pain despite the high degree of assimilation that has occurred among ethnic groups in the United States. Key words: Pain; Ethnicity;

Disease impacts; Health behavior

Introduction

Many researchers have addressed the connection between ethnicity and beliefs, attitudes, and actions related to health and illness. The belief that ethnic background affects pain perception is widespread among health care professionals and lay people alike. An understanding of variation in the experience of pain is important for ensuring high quality medical care for patients in whom pain is part of the clinical picture. Pain is one of the most frequent reasons patients visit their physicians. The proportion of drug prescriptions comprising analgesics and associated medications in the United States (about 9% in 1981) [7] underscores the importance of pain assessment and management in health care. In discussing the relation between ethnicity and pain, most commentators cite the pioneering work of Zborowski 1161,who studied differences in expression of

Correspondence to: Dr. Howard P. Greenwald, Associate Professor and Director, Health Services Administration Program, School of Public Administration, University of Southern California, University Park, Los Angeles, CA 90089-0041, U.S.A.

pain among 4 ethnic groups: Irish, Italian, Jewish, and Old American (i.e., native born Anglo-Saxon). On the basis of interviews in a New York City hospital in the early 1950s Zborowski reported that Jewish and Italian patients were more likely to express pain freely than Irish or Old American patients. Zola, another early researcher, compared Boston hospital patients of Italian Catholic, Irish Catholic, and Anglo-Saxon background [17]. The Irish and Anglo-Saxon patients tended to deny the presence of pain, while the Italian patients, again, expressed this symptom more freely. A later study by Flannery et al. comparing pain in c~ldbirth among Black, Irish, Italian, Jewish, and Protestant Anglo-Saxon women detected no statistically significant differences [l]. Other investigations, however, have observed differences among Blacks, Caucasians, and Puerto Ricans in pain associated with a variety of conditions [2,5,15]. These and related studies leave much uncertainty about the refation between ethnicity and pain today. The variety of methods used to assess expression of pain make comparison of research findings difficult. Most studies do not control for social background variables other than ethnicity that may affect perceptions or interpretation of pain. Perhaps most important, nearly all previous studies have taken place in locations where

15X

migration and immigration patterns have led to the predominance of readily identifiable ethnic groups in the population. In places where ethnic groups are less distinct or have become better assimilated, differences in perceptions of pain and other health-related forms of perception and behavior may disappear. The study reported here helps clarify the place of ethnicity in determining individual expression of pain by observing persons recently treated for cancers (lung, pancreatic, prostatic, and cervical) frequently associated with pain, and in an age range (mean = 68 years) where chronic diseases involving pain are prevalent. Data available in this study permit analysis including key sociaf background factors and widely used, standardized indices of pain perception and interpretation. The study took place in Western Washington state, a region with relatively few residents born outside the United States or members of distinct minority groups. According to the 1980 U.S. Census, the foreign born constituted 7.4% of the population in the Seattle-Everett SMSA, compared with 21.3% in the New York-New Jersey SMSA and 22.3% in the Los Angeles-Long Beach SMSA. Blacks, Asians, and persons of Spanish-speaking origin constituted 3.6, 4.0, and 2.0% of the Seattle SMSA. Comparable percentages were 21.3, 3.0, and 16.4 for New York, and 12.6, 5.8. and 27.6 for Los Angeles.

Methods Patient sample The researchers collected data on patients with cancers of the lung, pancreas, prostate, and uterine cervix. Although the frequency and degree of reported pain varied by disease, a significant proportion of individuals with each malignancy evidenced serious pain problems [5]. Samples of patients with the above malignancies in King and Pierce Counties, Washington, were obtained from the Cancer Surveillance System (CSS), a population-based tumor registry maintained by the Fred Hutchinson Cancer Research Center in Seattle. An evaluation of the CSS has recently demonstrated that fewer than 2.5% of the cancer cases in its catchment area are missed by the registry 1131. The study reported here was restricted to recently diagnosed, primary malignancies and to individuals older than 20 years and younger than 80 years, whose names were entered into the CSS registry during an 18 month data collection period between 1980 and 1981. The investigators identified a total of 877 individuals meeting the above criteria who remained alive by the time their names were entered in the CSS. They obtained physician approval (required for human subjects considerations) to request participation in the study

from 599. The research team succeeded in contacting a total of 591, of whom 536 agreed to interviews and provided sufficient data to be included in the analysis. A comparison of CSS records of those interviewed with those not interviewed revealed no statistically significant differences on the dimensions of gender. marital status, race. method of diagnosis, age, and cancer stage. Pain meusurement Of several several instruments incfuded in the patient interview protocols for assessment of pain, mood, and function, findings from two are reported below: (1) a series of Graphic Rating States (GRS) combining visual, numerical, and descriptive indicators, asking the subject to focus successively on the worst pain he or she had experienced in the preceding 24 h. 7 days, and 2 months; (2) the McGill Pain Questionnaire (MPQ). The GRS consisted of printed 10 cm lines, above which appeared the numbers 1 through 5 evenly spaced along the line, and below which appeared the words “no pain,” “slight pain,” “moderate pain,” “very bad pain.” and “pain as bad as can be,” positioned to correspond closely with the numbers. Subjects received instructions to circle the number which most closely corresponded to their worst pain over the specified period. Through this indicator, the investigators attempted to combine several measures of pain intensity whose validity has received strong support in clinical and laboratory studies [6,12.14]. The MPQ consists of blocks of words capable of describing the quality and intensity of pain, individual words within each block having been assigned weights corresponding to pain severity on the basis of evaluation by panels of physicians, patients, and students [lo]. Scores are generated on the basis of words appearing on the MPQ form selected by the subject to describe his or her pain. The MPQ has an especially valuable feature in its ability to measure specific aspects of the pain experience according to distinct classes of words. Examples include: (I) sensory qualities of pain stated in terms describing temporal, spatial, pressure, thermal, and other properties; (2) affectioe qualities, in terms of tension, fear, and similar qualities related to emotions. MPQ findings presented here are based on a weighted version of the instrument, subscale scores reflecting distinct qualities of pain, each felt at a specific Ievel of intensity. The reliability and validity of the MPQ have been demonstrated on a variety of populations [3,9]. The ability of the sensory and affective subscales to discriminate among specific aspects of pain is supported by tests of external validity [4]. Ethnic identificution the principal independent variable in “ Ethnicity,” this study, was assessed in terms of the subject’s identification of ethnic origins. As part of the interview

159

protocol, each subject was handed a card listing 25 countries or groups of countries and asked to identify the “national or ethnic group” that best described his or her ethnic origin. Subjects were instructed to designate as many indicators of ethnic origin as they felt were applicable. Subjects were considered members of a specific ethnic group if they selected the corresponding country or set of countries, whether or not they selected additional ethnic designations. The self-identification feature of this procedure for assessing ethnicity corresponds closely to that used in the Bureau of the Census Current Population Survey (CPS) and the National Center for Health Statistics Health Interview Survey (HIS) [ll]. In HIS questionnaire pretests, more objective approaches such as assessment of ethnicity by parents’ national origin, language spoken at home, etc., proved unusable because more than 95% of respondents were born in the United States. A variable designating individuals as belonging to a specific ethnic group only if they identified a single

ethnic origin was not used here, because fewer than half the subjects made designations of this kind. In view of foregoing research on ethnicity and pain, it appeared important to identify “Jewish” as an ethnic category. Because no specific geographical lineage indicated this ethnicity, the interview protocol asked subjects to identify their father’s and mother’s religion. A variable was computed to indicate Jewish ethnicity if either one or both parents were Jewish. Data

analysis

This study used 2 methods of data analysis: (1) comparison of percentages of subjects in each ethnic category reporting specific levels of pain according to the GRS, with associated &i-square tests; (2) multiple regression analysis of scores on both the GRS and MPQ subscales. In the first procedure, the 5 pain levets of the GRS were collapsed into 2 categories, and comparisons made between those indicating “no pain” or “slight pain,” and those indicating “ moderate pain,” “ very bad

PAIN ~~ODE~TE~ TO "AS BAD AS CAN BE*

N

ETHNIC! ORIGIN

35.0

(254)

ENGLAND, SCOTLAND, WALES

41.7 63.0

GERMANY

36.8

(155)

43.2 59.4 41.2

IRELAND

46.1 66.7 46.3

SCANDINAVIA

53.7 64.2 FRANCE

37.5

(56)

44.6 67.9 EASTERN EUROPE

34.2 44.7 68.4

ITALY

28.6 35.7 64.3

JEWISH

44.4 44.4 77.8 -

WORST IN PAST DAY

r//y./, WORST IN PAST WEEK Fig. 1. Worst pain by ethnic origin.

WORST IN PAST TWO MONTHS

%

I60

pain,” or “pain as bad as can be.” Chi-square tests were based on 4-fold tables cross-tabulating percentages of subjects reporting each pain level by dichotomous variables representing membership versus non-membership in each ethnic group. In the second procedure. GRS and MPQ scores served as dependent variables in regression equations which included the following variables on the right-hand side: disease site and stage; patient sex, age, education, and income; ethnicity represented by a series of dichotomous variables. In the regression equations. the dichotomous variables representing ethnicity indicated differences between individuals inside and outside each category. Subjects, of course. could be included in multiple ethnic categories; several ethnic categories were no! included in the equations due to low numbers of observations.

lapping time periods: the pas! 24 h. the past week, and the past 2 months. The percentages indicated in Fig. 1 represent the proportion of subjects who selected the words ” moderate pain.” “ very bad pain.” or “pain II> bad as can be” to indicate the worst pain they experienced in the specified time periods. Percentages are presented for the 6 most frequently selected designations of ethnicity identified in the sample, plus Italian and Jewish, of special relevance because of their importance in earlier studies. The number of cases with each designation is presented in the figure. These numbers total more than the sample size because muitiple designations were evident in many cases. Chi-square tests comparing subjects in any given ethnic category with those not included in that category indicated no statistically significant differences for worst pain over any of the 3 time periods. Table I presents a multiple regression analysis of responses to the GRS items for the same time periods. Coefficients in this table are from ordinary least squares (QLS) equations prediciting GRS responses coded 1 through 5, c(~rresp~)nding to the verbal designations of pain level on the scale. These models include variables representing cancer site and stage and social background variables (sex, age. education. and income) in

Findings Fig. 1 and Tables I and II study’s findings. Fig. 1 presents findings based to GRS items requesting them pain they experienced over three

TABLE

below

summarize

this

on subjects’ responses to indicate the worst distinct though over-

I

REGRESSION

OF WORST

PAIN

ON ETHNIC

ORIGIN

AND

SELECTED

BACKGROUND

VARIABLES ______

Worst pain in past day

Worst pain in past week

B

B

SE.

SE.

Worst pain in past 2 months B

SE.

--_I___

Diseuse chu~acteri~~i~ Site Pancreas

-0.315

Prostate

-0.319

Cervix

- 0.646 * *

Stage

*

(0.269)

0.104

(0.281)

(0.142)

- 0.238

(0.149)

-0.119

(0.242)

-0.619

(0.253)

-0.771

(0.072)

I--0.060

0.009

(0.069)

Sex(M=l;F=2)

- 0.084

(0.159)

Age

- 0.057

(0.031)

-0.066

Education

-0.015

(0.018)

Income

- 0.016

England, etc. Germany

**

0.034

0.544

(0.322) (0.170) **

(0.290) (0.0X2)

Social background 0.084

(0.167) *

0.095

(0.190) **

(0.033)

-0.129

- 0.014

(0.019)

- 0.001

(0.0221

(0.019)

- 0.014

(0.020)

- 0.024

(0.023)

(0.037)

-0.133

(0.117)

-0.151

(0.123)

- 0.068

(0.140)

-0.155

(0.121)

- 0.076

(0.126)

0.071

10.144)

Ethnrc origin

Ireland

0.064

(0.128)

-0.051

(0.134)

0.202

(0.153)

Scandinavia

0.044

(0.154)

- 0.036

(0.161)

0.024

(0.184)

France

- 0.042

(0.166)

- 0.012

(0.174)

0.159

(0.198)

Eastern Europe

- 0.248

(0.233)

- 0.062

(0.244)

0.296

(0.279)

Italy

- 0.249

(0.335)

-0.21R

(0.350)

0.072

(0.401)

Jewish

0.202

(0.450)

(0.471)

0.275

(0.539)

R’ =

0.05

* P -C0.05. ** P