The Clinical Journal of Pain 17:365–374 © 2001 Lippincott Williams & Wilkins, Inc., Philadelphia
Psychometric Properties of a Spanish Version of the McGill Pain Questionnaire in Several Spanish-Speaking Countries *Carlos Lázaro, M.D., †Xavier Caseras, Ph.D., ‡Victor M. Whizar-Lugo, M.D., §Roberto Wenk, M.D., 㛳Fernando Baldioceda, M.D., ¶Rodrigo Bernal, M.D., ¶Abdiel Ovalle, M.D., †Rafael Torrubia, Ph.D., and *J. E. Baños, M.D., Ph.D. Departments of *Farmacologia, de Terapéutica i de Toxicologia and †Psiquiatria i Medicina Legal, Universitat Autònoma de Barcelona, Spain; ‡Servicios Profesionales de Anestesiología y Clínica del Dolor, Centro Médico del Noroeste, Tijuana, Baja California, México; §World Health Organization Focal Point, Programa Argentino de Medicina Paliativa, San Nicolás, Argentina; 㛳Trastornos Temporo-mandibulares y Dolor Facial, San José, Costa Rica; and ¶Instituto Oncológico Nacional, El Dorado, Panamá
Abstract: Objective: Versions of the McGill Pain Questionnaire are available in a several languages and are used in clinical studies and sociocultural or ethnic comparisons of pain issues. However, there is a lack of studies that compare the validity and reliability of the instrument in the countries where it is used. The current study investigates the psychometric properties of a Spanish version of the McGill Pain Questionnaire in five Spanish-speaking countries. Design: The authors conducted a multicenter and transnational study with one investigator in each center. Patients were evaluated once with a Spanish version of the McGill Pain Questionnaire, a visual analog scale, and a verbal rating scale. Setting: The study was performed in pain clinics and acute pain units of four Latin American countries (Argentina, Costa Rica, Mexico, and Panama) and Spain. Patients: The study included 205 patients (84 with acute pain, 121 with chronic pain) from Latin America. Their data were compared with those of 282 Spanish patients. Interventions: The McGill Pain Questionnaire, visual analog scale, and verbal rating scale were administered once to all patients. The McGill Pain Questionnaire was administered again to patients from Latin America countries to ascertain descriptor comprehension. Outcome measures: Demographic data, McGill Pain Questionnaire parameters, and visual analog scale and a verbal rating scale scores were obtained from patients with chronic and acute pain. Psychometric properties of the Spanish version of the McGill Pain Questionnaire were established for each country by calculating the ordinal consistency by means of rank-scale correlation (Spearman test), intercategory correlation, and interparameter correlation (Pearson test). Concurrent validity was also calculated by comparing scores from the visual analog scale (Pearson test) and verbal rating scale (Spearman test) with questionnaire parameters (qualitative-toquantitative comparisons). Results: The Spanish version of the McGill Pain Questionnaire maintained a high internal validity when tested in different countries. Ordinal consistency, intercategory, interparameter, and qualitative-to-quantitative parameter correlations were similar in all countries. Few descriptors were considered to be inappropriate or difficult to understand.
Received July 10, 2000; revised April 7, 2001; accepted July 2, 2001. Address correspondence and reprint requests to Dr. Josep-E. Baños, Department de Farmacologia, de Terapéutica i de Toxicologia, Facultat de Medicina, Universitat Autònoma de Barcelona, 08193-Bellaterra, Spain; e-mail:
[email protected]
365
366
LÁZARO ET AL. Conclusions: The psychometric properties of the Spanish version of the McGill Pain Questionnaire assessed in different Latin-American countries suggest that the questionnaire may be used to evaluate Spanish-speaking patients. The validity of this test should be extended with reliability studies to further establish its usefulness in the evaluation of pain. Key Words: Cultural differences in pain—McGill Pain Questionnaire—Pain assessment—Transnational studies.
Many methods have been used in recent years to assess clinical pain. These tools are commonly taken from the psychosocial field to explore the multiple components involved in the experience of pain. Nonetheless, a common limitation of these instruments has been a strong sociocultural dependence. With such a limitation, there is an increased risk that misunderstandings will occur when each tool is not individually adapted to each sociocultural or ethnic environment. The absence of cross-cultural comparisons and validations may also increase such risk.1 All of these considerations are especially important when dealing with pain, because this experience has strong sociocultural influences. Moreover, pain is highly prevalent in Spanish society and in other Western countries. For instance, a recent epidemiologic study has found that 76% of the adult population (>18 years) had experienced a painful event in the 6 months before the study.2 Although data from Latin American countries are not available, a similar high prevalence of pain can be assumed. Ethnic and cultural differences influence coping style,3 behavioral distress,4 treatment preferences,5 and reporting4 of several pain syndromes. This diversity might be due to sociocultural differences related to patient expectations, availability of health care systems, physician training and practice styles, or to the existence of financial compensation programs linked to disability.6 The McGill Pain Questionnaire (MPQ) is one of the most widely used tools in pain assessment,7 and several translations and versions of this test are available.8–20 Simple direct translations are of limited value because of the difficulty of finding conceptual semantic (meanings) and metric equivalencies (descriptor-associated intensity) of the items.20 In contrast, the adapted version is a modified model of the original questionnaire, which has been adjusted according to the culture in which it will be administered. Such versions should assure the accurate understanding of the questionnaire by the population to be studied and that the validity of the test will remain intact in the new cultural context. Although useful, the development of an adapted version is more time consuming and laborious than mere direct translations. Furthermore, it is usually necessary to perform an analysis of The Clinical Journal of Pain, Vol. 17, No. 4, 2001
psychometric properties such as validity and reliability when creating these adaptations. The original MPQ has been used to assess pain in English-speaking populations, which comprise many different sociocultural backgrounds. In the United States, Canada, and the United Kingdom, multiethnicity is the rule.8 The same situation applies for the MPQ versions in other languages, especially when that language is spoken in several countries. To our knowledge, studies that analyze the psychometric properties of the MPQ in such a variety of countries have not been published. Spanish is one of the most-spoken languages in the world, and the growing use of Spanish in many countries parallels that of English. Demographic forecasts estimate that Spanish-speaking persons will become the secondlargest ethnic group in the United States within this century; in fact, these persons comprise the majority of the population in some Southern states.11 In recent years, at least two Spanish versions of the MPQ with psychometric analysis have been published,11,21 but other versions and translations are also available.13,14,17 The questionnaire designed by Lázaro et al.21 was not a mere translation; the authors followed a process similar to that used to create the original MPQ, but some differences should be highlighted. First, the descriptors used by patients were weighted more than those coming from dictionaries or translations in the final selection. Second, students performed the ascription of descriptors into subclasses and were free to vary the names or to create new ones. As a result, subclass punishment was deleted because of a lack of descriptors accomplishing the criteria to be included in this subclass. Later, a second process of inclusion was performed to classify descriptors given by patients and not classified by students. These descriptors were included in three miscellaneous subclasses.21 However, like other Spanish versions, the Spanish version of the MPQ (MPQ-SV) has not been tested in countries other than Spain. This evaluation is needed before the instrument can be recommended for clinical use by investigators from other Spanish-speaking countries. We hypothesize that the MPQ-SV might maintain its psychometric properties in different countries with a common language background. To prove this
SPANISH VERSION OF THE MCGILL PAIN QUESTIONNAIRE sumption, we investigated the psychometric properties and comprehension of the MPQ-SV in patients from several Spanish-speaking Latin American countries. METHODS Study design and setting Thirty-one members of the International Association for the Study of Pain (IASP) from Latin American countries (four from Argentina, three from Colombia, one from Costa Rica, one from Chile, one from Ecuador, four from Mexico, three from Panama, one from Puerto Rico, two from Uruguay, and two from Venezuela), and seven U.S. members (from Arizona, California, Florida, New Mexico, and Texas) were approached by mail. Members were randomly selected from the IASP directory and no contacts had been previously established. The letter that was sent explained the purpose of the study and invited these members to participate in the validation of MPQSV in their country. Seven members requested further information (one from Argentina, one from Costa Rica, two from Mexico, two from Panama, and one from Puerto Rico) and were sent a new letter with a draft of the protocol and a request for suggestions. Only five members (one from Argentina, one from Costa Rica, one from Mexico, and two from Panama) agreed to participate. A steering committee in Spain was established to coordinate the study and comprised two psychologists, one pharmacologist, and one anesthesiologist. The committee mailed and collected the questionnaires, processed the data, performed the statistical analyses, and wrote the final report. The researchers interviewed the patients, completed the case report forms and returned the forms to the steering committee, and approved both the final protocol and the final report. The researchers were asked to use the MPQ-SV on patients in their usual clinical practice. Inclusion criteria were normal cognitive and communicative abilities as evaluated by the researcher, actual pain at the time of the interview, and verbally informed consent to participate. Case report forms collected information regarding sociodemographic variables (sex, age, country) and pain characteristics (acute or chronic, type, etiology if known). Afterwards, patients were asked to score their pain intensity using a visual analog scale (VAS) of 100 mm and a verbal rating scale (VRS) of five points (leve ⳱ mild, moderado ⳱ moderate, fuerte ⳱ severe, muy fuerte ⳱ agonizing, and insoportable ⳱ unbearable). Then, the investigator administered the MPQ-SV. Administration of the Spanish version of the McGill Pain Questionnaire Researchers read each group of MPQ descriptors (subclass) to the patients in the same order of appearance as
367
in the MPQ-SV. After reading the descriptors of each subclass, patients were asked to choose one or none of the descriptors according to the quality of pain he or she felt at that moment. The researcher marked the descriptors chosen by the patient in the case report form, thus completing the clinical portion of the administration. Patients were then asked to read the MPQ-SV descriptors on a separate page of the case report form. This time, they were invited to point out what descriptors other than those chosen might describe any kind of pain. Patients were also asked to report if they did not understand a descriptor. Descriptors chosen by at least 50% of the patients from each country, and not previously chosen by more than 20% of the patients during the clinical administration, were deemed inappropriate. Statistical analysis Results are presented to compare values from Latin American countries with those from Spain, where the instrument was initially developed and validated. The mean, standard deviation, and range of the variables are reported. For the VRS scores, the median and 25th to 75th percentile values are given. The descriptors considered as inappropriate or unknown are expressed as percentages. To establish psychometric properties of the MPQ-SV, we studied ordinal consistency of descriptors using Spearman correlation among rank (descriptor position within the subclass) and scale values (intensity of pain associated with a descriptor obtained in the original development of MPQ-SV). To determine intercategory correlation, correlation coefficients among the different parameters of the MPQ-SV were calculated, and relations among the total, sensorial, affective, and evaluative pain rating indices (PRIs) were examined. Interparameter correlation of the MPQ-SV (e.g., total number of words chosen) was correlated to pain rating index values. Additionally, quality or intensity parameters of the MPQSV were correlated to the VRS (Spearman test) and to the VAS (Pearson test). RESULTS General characteristics of the sample The study recruited 205 Latin American patients from Argentina (n ⳱ 40), Costa Rica (n ⳱ 24), Mexico (n ⳱ 96), and Panama (n ⳱ 45). The Spanish group included 282 patients. All patients who were approached agreed to participate in the study. Detailed information regarding gender, age, and type of pain is given in Table 1. In each country, women accounted for the majority of patients. Patients in countries with more cases of acute pain The Clinical Journal of Pain, Vol. 17, No. 4, 2001
368
LÁZARO ET AL. TABLE 1. General characteristics of the sample
Argentina Costa Rica Mexico Panama Spain
N
Men (%)
Women (%)
40 24 96 45 282
17 (42.5) 2 (8.7) 46 (47.9) 8 (17.8) 87 (30.7)
23 (57.5) 22 (91.3) 50 (52.1) 37 (82.2) 195 (69.3)
Age mean (SD) [range] 61.7 39.2 45.7 40.5 57.3
had a younger mean age compared with patients in countries with more cases of chronic pain. The majority of patients were affected by chronic pain, but in some countries (e.g., Mexico, Panama) more acute-pain patients were seen. The type of pain was different in each country. In Argentina, patients were affected mostly by cancer pain (56.1%). In Costa Rica, pain of a vascular etiology was the most frequent (84.6%), whereas in Mexico inflammatory pain of rheumatic disorders was most prominent (62.6%), followed by cancer pain (20.2%). A similar pattern was seen in Panama and Spain, where 71.1% and 50.5% of patients experienced inflammatory pain, respectively. Pain intensity Table 2 summarizes the pain measurement results obtained using VAS and VRS. The mean VAS scores were diverse and values ranged between both extremes (1– 100). An ANOVA was performed and, when significant, pairwise comparisons based on multiple-range test were carried out. Significant differences were observed among the different countries (F [df ⳱ 4, 475] ⳱ 6.7, p