Abstract:

The aim of this study was to identify resilience profiles of patients with chronic pain. Using latent class analysis in a sample of 414 patients with chronic musculoskeletal pain, three profiles were identified: primary resilience (40%), consisting of individuals 40 years or younger with high education, who seek medical care, are not working, and without symptoms of psychological stress; secondary resilience (30%), consisting of women over 54 years of age with low schooling, who seek medical care, are not working, and with low likelihood of symptoms of psychological stress; tertiary resilience (29%), women with medium schooling, 40 to 54 years old, working, who do not seek medical care, and with a high likelihood of symptoms of psychological stress. The three profiles display different paths of resilience in chronic pain that are relevant to clinical practice, highlighting the importance of multidisciplinary care for patients with chronic pain.

Keywords:
Chronic Pain; Psychological Resilience; Life Quality

Introduction

Chronic pain is pain that persists in continuous or recurrent form due to chronic disease processes. It has a heavy social impact and is frequently associated with other chronic diseases, placing a burden on health systems and the economy. It is a specific health problem, with clear characteristics of symptoms, incapacity, and mental health problems that are largely independent of the underlying disease or injury ¹1. Kleijnen Systematic Reviews Ltd. Reflection process on chronic diseases in the EU - the role of chronic pain. York: Societal Impact of Pain/European Federation of IASP; 2012. (Systematic Literature Report)..

The prevalence of chronic pain in the general population varies according to the study, from 10.1% to 80% ²2. Harstall C, Ospina M. How prevalent is chronic pain? Pain: Clinical Updates 2003; 11:1-4.^{), (3}3. Abu-Saad HH. Chronic pain: a review. J Med Liban 2010; 58:21-7.. The differences are due mainly to the criterion used to classify chronic pain (< 3 months versus > 6 months), as well as the study population (adults, elderly, war veterans).

Despite advances in the understanding of pain, its prevalence is still high, and it sometimes goes unrecognized or is poorly assessed, underestimated, undertreated, or inadequately treated ⁴4. Hadjistavropoulos T, Craig KD, Duck S, Cano A, Goubert L, Jackson PL, et al. A biopsychosocial formulation of pain communication. Psychol Bull 2011; 137:910-39.. This may be related to the biomedical model's incapacity to deal with all the psychosocial, socio-demographic, behavioral, and sometimes even pathophysiological processes that trigger pain. This emphasizes the biopsychosocial focus, in which the subject's biological, psychological, and social characteristics become important for understanding chronic pain. Specifically, in this article we aim to investigate the interrelations between these variables from the perspective of resilience.

Resilience can be defined as one's capacity to adapt to stressful circumstances, and is associated with decreased perception of the stress. Higher levels of resilience are associated with higher levels of acceptance of pain and adaptation to it, regardless of its duration. Resilience prevents emotional stress and is associated with lower levels of depression and anxiety ⁵5. Ramírez-Maestre C, Esteve R, López AE. The path to capacity: resilience and spinal chronic pain. Spine (Phila Pa 1976) 2012; 37:E251-8..

One major difficulty is the operationalization of scientific findings in clinical practice. Trivedi et al. ⁶6. Trivedi RB, Bosworth HB, Jackson GL. Resilience in chronic illness. In: Resnick B, Gwyther LP, Roberto KA, editors. Resilience in aging: concepts, research, and outcomes. New York: Springer; 2011. p. 181-97. developed the concept of resilience as a construct with three levels, which they called primary, secondary, and tertiary resilience. The objective of this conceptualization is its application to clinical practice in patients with chronic diseases, acknowledging that resilience can change over time, which would allow monitoring the levels with specific responses.

The aim of the current article is to draw a resilience profile of patients with chronic pain, combining the strategy of using a specific instrument ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7. with the concept of resilience as a three-level construct ⁶6. Trivedi RB, Bosworth HB, Jackson GL. Resilience in chronic illness. In: Resnick B, Gwyther LP, Roberto KA, editors. Resilience in aging: concepts, research, and outcomes. New York: Springer; 2011. p. 181-97.. The idea is to reflect the individual's life history and social context so that at least part of the enormous range of interactions can be contemplated. We use latent class analysis (LCA) for this purpose, as a person-oriented approach ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.. The focus of LCA is to study individuals based on the patterns of relevant individual characteristics ⁹9. Bergman LR, Magnusson D. A person-oriented approach in research on developmental psychopathology. Dev Psychopathol 1997; 9:291-319., observing the profiles of those displaying similar patterns. This allows identifying more general processes underlying the profiles, which are assembled based on the probability of belonging (conditional probability) to the latent classes furnished by the modeling.

Methodology

Sample

The current study complied with Resolution n. 196/96 of the Brazilian National Research Council, which deals with the guidelines and regulatory standards for research in human subjects, and was approved by the Institutional Review Board of the institution where the research took place (case review n. 1005-55, Research and Graduate Studies Group, Porto Alegre University Hospital). After obtaining authorization for use of the principal investigator's database, the study was approved by the Institutional Review Board of the Sergio Arouca National School of Public Health, Oswaldo Cruz Foundation (CEP/ENSP/Fiocruz, CAAE: 24131413,0,0000,5240).

The data collection methodology has been described by Caumo et al. ¹⁰10. Caumo W, Ruehlman LS, Karoly P, Sehn F, Vidor LP, Dall-Ágnol L, et al. Crosscultural adaptation and validation of the profile of chronic pain: screen for a Brazilian population. Pain Med 2013; 14:52-61.. The sample was recruited in a tertiary clinic for chronic pain (patients referred from primary care units), in addition to volunteers recruited through newspaper ads. The inclusion criterion was diagnosis of musculoskeletal pain by physicians with clinical experience in pain management. The diagnosis was established using the standard protocol for each type of pain. The clinical criteria for defining patients with myofascial pain syndrome were regional pain, normal neurological examination, presence of trigger points, or painful points that cause pain irradiating to the affected muscle when pressed or stimulated, taut bands in the muscle tissue, tender points, and pain described as "dull", "tender", or "deep". The diagnosis of chronic tension headache was based on the criteria of the International Headache Society, and the diagnosis of fibromyalgia was based on the American College of Rheumatology criteria.

The exclusion criteria were inability to understand the native language, illiteracy, or impossibility of appearing at the hospital for the assessments.

Study instrument and variables

The initial classification of individuals according to level of resilience used the version of the Profile of Chronic Pain: Screen (PCP:S) translated to Portuguese and adapted to Brazil. The original version of the PCP:S consists of 15 items distributed across three dimensions ¹¹11. Ruehlman LS, Karoly P, Newton C, Aiken LS. The development and preliminary validation of a brief measure of chronic pain impact for use in the general population. Pain 2005; 113:82-90.: severity - intensity or aversion to pain (four items; score from 0 to 30); interference - impact of pain on pleasant activities, relationships, responsibilities, personal goals, self-care, and cognition (six items; score from 0 to 36); emotional stress - feelings related to pain, such as sadness, strain, anger, pleasure, withdrawal (five items; score from 0 to 25). The translated version adapted to the Brazilian population ¹⁰10. Caumo W, Ruehlman LS, Karoly P, Sehn F, Vidor LP, Dall-Ágnol L, et al. Crosscultural adaptation and validation of the profile of chronic pain: screen for a Brazilian population. Pain Med 2013; 14:52-61. showed the same dimensional structure as the original study.

For classification of individuals as resilient versus non-resilient, the study used a similar procedure to that described by Karoly & Ruehlman ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7.. Thus, individuals were classified as resilient when they obtained high scores on the severity scale (score above the sample's mean on the scale) in combination with a low score on the interference and emotional stress scales (score below the sample's mean on the scale). Non-resilient individuals were classified according to high scores on the interference, severity, and emotional stress scales (score above the sample's mean on the scale). The others were classified as "others". The necessary adjustments were made since the current study's sample was small when compared to that used by Karoly & Ruehlman ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7..

In this study, the variables used to construct the profile via LCA are: age (≤ 30 years, 31-40, 41-59, ≥ 60), sex (male, female), schooling (low: 0-5 years, average: 6-10, high: ≥ 11), working despite the pain (yes, no), specific medical treatment for chronic pain (yes, no), resilience (resilient, non-resilient, other). In addition, to explore the model's validity we used the variable related to diagnosis of chronic pain (myofascial pain syndrome, chronic tensional headache, fibromyalgia).

Statistical modeling

LCA is a finite mixture model ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.^{), (12}12. Muthen B, Shedden K. Finite mixture modeling with mixture outcomes using the EM algorithm. Biometrics 1999; 55:463-9.^{), (13}13. Agresti A. Categorical data analysis. Hoboken: John Wiley Sons; 2002.^{), (14}14. Magidson J, Vermunt JK. Latent class models. In: Kaplan D, editor. The Sage handbook of quantitative methodology for the social sciences. Thousand Oaks: Sage Publications; 2004. p. 175-98.^{), (15}15. Vermunt JK, Magidson J. Latent class analysis. In: Lewis-Beck MS, Bryman A, Liao TF, editors. Encyclopedia of social science research methods. Thousand Oaks: Sage Publications; 2004. p. 549-53., the application of which is adequate when postulating that a population consists of two or more underlying latent subgroups, defined by the intersection of numerous individual characteristics, i.e., when postulating the existence of an unobserved underlying categorical variable (construct) that divides the population into mutually exclusive and exhaustive latent classes ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.^{), (16}16. Hagenaars JA, McCutcheon A. Applied latent class analysis. New York: Cambridge University Press; 2002.. The categories' pertinence to the classes is unknown, but can be deduced through a set of measurements (items, variables), generally categorical.

LCA allows classifying cases with categorical indicators, similar to factor analysis with continuous variables. However, the first type of analysis includes the probability that a group of individuals with particular characteristics belong to a latent class that is exclusively dependent on the data ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.^{), (16}16. Hagenaars JA, McCutcheon A. Applied latent class analysis. New York: Cambridge University Press; 2002..

Two parameters are essential for understanding LCA: the probability of belonging to a given latent class (prevalence of latent classes) and the probability of the category's response within each class (conditional probabilities), which can be used to obtain the expected proportion in each cell in the contingency table ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.. If the model fits the data well, the expected proportion of cells coincides with the observed proportion ⁸8. Collins LM, Lanza ST. Latent class and latent transition analysis: with applications in the social, behavioral, and health sciences. New York: Wiley; 2010.. The latent classes can be defined (named, interpreted) by the combination of conditional probabilities, which in turn characterize the individuals belonging to the latent class in question.

The modeling for choosing the number of latent classes considered the following criteria:

Pearson's χ² goodness of fit and log-likelihood ratio chi-square (G²): the model is considered adequate when it minimizes the values for χ² and G² without an excessive number of parameters ¹⁷17. Linzer DA, Lewis J. poLCA: an R package for polytomous variable latent class analysis. J Stat Softw 2011; 42:1-29.^{), (18}18. Linzer DA, Lewis J. poLCA: polytomous variable latent class analysis: R package version 1.4. https://github.com/dlinzer/poLCA (acessado em 10/Dez/2014).
https://github.com/dlinzer/poLCA... ; in this case, we obtain p > 0.05, i.e., it is not necessary to include more latent classes in the model.

Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC): the best models show the lowest BIC and/or AIC values ¹⁶16. Hagenaars JA, McCutcheon A. Applied latent class analysis. New York: Cambridge University Press; 2002.. BIC is more appropriate for latent classes, since it is simple and considers the sample size in its calculation ¹⁶16. Hagenaars JA, McCutcheon A. Applied latent class analysis. New York: Cambridge University Press; 2002.^{), (19}19. Lin TH, Dayton CM. Model selection information criteria for non-nested latent class models. J Educ Behav Stat 1997; 22:249-64.^{), (20}20. Forster MR. Key concepts in model selection: performance and generalizability. J Math Psychol 2000; 44:205-31.^{), (21}21. Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Modeling 2007; 14:535-69.. We also have consistent AIC (CAIC) and sample-size adjusted BIC (SSBIC or adjBIC), which follow the same logic as the other criteria, with sample-size adjusted BIC preferred for small samples ²¹21. Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Modeling 2007; 14:535-69..

Entropy: this is a measure of dispersion (or concentration) of a probability mass function ¹⁷17. Linzer DA, Lewis J. poLCA: an R package for polytomous variable latent class analysis. J Stat Softw 2011; 42:1-29.^{), (18}18. Linzer DA, Lewis J. poLCA: polytomous variable latent class analysis: R package version 1.4. https://github.com/dlinzer/poLCA (acessado em 10/Dez/2014).
https://github.com/dlinzer/poLCA... . In practice, entropy seeks to assess whether the separation of classes proposed by the modeling displays acceptable quality, i.e., whether the individuals' classification in classes is satisfactory. The current study adopts the measure of relative entropy ²²22. Ramaswamy V, Desarbo W, Reibstein D, Robinson W. An empirical pooling approach for estimating marketing mix elasticities with PIMS data. Marketing Science 1993; 12:103-24., which varies from 0 to 1, where 0 is the worst situation and 1 is the best: the closer to 1, the clearer the definition of classes ²³23. Celeux G, Soromenho G. An entropy criterion for assessing the number of clusters in a mixture model. J Classif 1996; 13:195-212..

Residuals analysis: this analysis is essential for assessing the assumption of local independence. The most common method consists of comparing the observed frequency of variables with the expected frequency according to LCA. The bivariate residuals are observed for each comparison pair; thus, a 2X2 table presents four residuals to be evaluated. This study uses the normalized residual, and results greater than │1.96│ are considered significant, indicating violation of the local independence assumption ²⁴24. Vermunt JK, Magidson J. Latent GOLD 4.0 user's guide. Belmont: Statistical Innovations Inc.; 2005.^{), (25}25. Vermunt JK, Magidson J. Technical guide for latent GOLD 4.0: basic and advanced. Belmont: Statistical Innovations Inc.; 2005.. It should be noted that there is no golden rule concerning the acceptable number of residuals; however, to add more than four residuals can make the model unwieldy ²⁶26. Muthén LK, Muthén BO. Technical appendices: standardized residuals in Mplus. 2007. https://www.statmodel.com/download/StandardizedResiduals.pdf (acessado em 20/Dez/2014).
https://www.statmodel.com/download/Stand... .

Having determined the best latent class model, we next endeavored to verify the model's quality. Thus, we compared the latent classes as a function of the score obtained on the PCP:S dimensions, using one-way ANOVA with post-hoc Bonferroni test. To assess external validity, we used the maximum log-likelihood test (G², similar to χ² ), comparing the distribution of individuals according to diagnosis of chronic pain with the classification predicted by the latent class modeling.

Modeling used the poLCA package ¹⁷17. Linzer DA, Lewis J. poLCA: an R package for polytomous variable latent class analysis. J Stat Softw 2011; 42:1-29.^{), (18}18. Linzer DA, Lewis J. poLCA: polytomous variable latent class analysis: R package version 1.4. https://github.com/dlinzer/poLCA (acessado em 10/Dez/2014).
https://github.com/dlinzer/poLCA... for R software. Significance was set at 5%.

Results

The study sample included 414 patients recruited at the Pain and Palliative Care Clinic of the University Hospital in Porto Alegre, Rio Grande do Sul, Brazil. Of these, 347 were women with chronic musculoskeletal pain, with a mean of 11.03 years of schooling (standard deviaton - SD = 5.43) and mean age 50.23 years (SD = 17.10).

Of all the patients, 65.9% were seeking medical care and 69.8% were not working because of the pain (Table 1). The method for classification of resilience resulted in a total of 11.6% of resilient individuals, 33.1% non-resilient, and 55.3% not classified either as resilient or non-resilient, thus classified as "others". This "others" category was removed from the original study by Karoly & Ruehlman ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7., and consisted of individuals with the following characteristics on the PCP:S scale: low severity (mean 18.20) and intermediate emotional stress (mean 11.97) and interference (mean 16.87). Still, in all the dimensions it is the category with the widest variability of data, with SD of 5.47, 6.42, and 8.63, respectively (Table 1).

Means in the sample were 21.60 on the severity dimension (standadrd error - SE = 0.28), 14.76 on emotional stress (SE = 0.35), 21.05 on interference (SE = 0.48), and 57.41 on the total scale (SE = 0.97).

Table 2 shows the goodness of fit for the latent class models, from one to five classes. The fit generally decreased as the number of classes increased, but BIC and CAIC found their optimal point with the three latent classes model (BIC = 3,619.69; CAIC = 3,648.69). Although the other fits point to the model's adequacy with four or five latent classes, we found that the addition of a fourth or fifth latent class made it redundant, with the additional classes failing to add explanatory power to the model from the theoretical point of view, also reflecting the observed p-value (0.143). There is thus evidence that after the third latent class, it is no longer necessary to add more classes. Given the above, the study adopted the model with three latent classes, with prevalence rates and conditional probabilities shown in Table 3.

The class nomenclature was that proposed by Trivedi et al. ⁶6. Trivedi RB, Bosworth HB, Jackson GL. Resilience in chronic illness. In: Resnick B, Gwyther LP, Roberto KA, editors. Resilience in aging: concepts, research, and outcomes. New York: Springer; 2011. p. 181-97. and relates to the predominant category of resilience in the latent class. Thus, class 1 was called primary resilience, and with a prevalence of 40% it has the highest probability of consisting of individuals up to 40 years of age, with 13 years of schooling or more, who seek medical care, not working, and classified as "others" on the PCP:S scale. As for gender, males showed a higher probability of belonging to this class than to others.

Class 2 was called secondary resilience, and with a prevalence of 30% it has a higher probability of consisting of women over 54 years of age, with up to eight years of schooling, who seek medical care, are not working, and classified as "resilient" on the PCP:S scale.

Class 3 was called tertiary resilience, and with a prevalence of 29% it has a higher probability of consisting of women 41 to 54 years of age, with nine to 12 years of schooling, who do not seek medical care, are working, and were classified as "non-resilient" on the PCP:S scale.

As shown in Table 4, the results show that the resilience profiles based on latent classes is able to discriminate between individuals. Thus, individuals classified with primary resilience displayed lower scores on all the dimensions of PCP:S, followed by those with secondary resilience and tertiary resilience, with significant differences between all the dimensions in the post-hoc Bonferroni test (p-value < 0.05).

Importantly, LCA is a finite mixture model, meaning that the classes consist of a combination of variables, and that we can find men and women in the same class, for example. What characterizes a class is that the category shows higher probability in this class than in the others.

Residuals analysis to assess local independence did not indicate any significant residual (greater than 1.96). The data thus meet the assumption of local independence.

Table 5 shows the results of the external validity test. The null hypothesis is that the distribution of individuals in the categories is independent. Comparison of latent classes and the classification of the diagnosis of chronic pain proved significant (G² = 16.060; p-value = 0.003): individuals with fibromyalgia belong predominantly to the tertiary resilience profile (63.2%). Meanwhile, individuals with chronic tension headache appear predominantly in the secondary resilience profile (46%) and those with myofascial pain syndrome belong predominantly to the primary resilience profile (42.6%), as evidence of the latent classes' external validity.

Discussion

This article identifies and describes a set of variables and its interrelationship with resilience in individuals with chronic pain, using LCA. It also assesses the validity of latent classes, observing the pattern found with an external validator, and in this case the results pointed to the validity of the existence of three latent classes of resilience: primary, secondary, and tertiary.

Primary resilience reflects individuals with low-grade pain, while also apparently reflecting different patterns of resilience. This difference could indicate both a hereditary and environmental effect, and the findings are thus consistent with the work of Boardman et al. ²⁷27. Boardman JD, Blalock CL, Button TMM. Sex differences in the heritability of resilience. Twin Res Hum Genet 2008; 11:12-27.. The authors investigated the inheritance of psychological resilience in adults (monozygotic twins) older than 25 years in the United States and found that men and women display the same degree of inheritance in relation to positive affect; still, after controlling for other psychological effects, the inheritance of resilience appears to be greater in men. Meanwhile, when compared to women, men appear to derive additional benefits in the environmental domain, which can allow other manifestations of resilience, independently of gender ²⁷27. Boardman JD, Blalock CL, Button TMM. Sex differences in the heritability of resilience. Twin Res Hum Genet 2008; 11:12-27.. The explanations for this environmental domain, according to the three authors, are based on the social role developed by women (housework and family), making them less resistant to stressful factors involving friends and family (divorce and domestic and relationship problems), since stressful events in life in general are similar for women and men ²⁷27. Boardman JD, Blalock CL, Button TMM. Sex differences in the heritability of resilience. Twin Res Hum Genet 2008; 11:12-27.. This would also support the age issue, since the primary resilience class includes male individuals that are younger than the women in the secondary resilience class, which could explain the development of distinct resilience paths in the two groups.

The secondary resilience class consisted predominantly of women more than 54 years old, with low schooling, not working, but who seek medical care. With 30% prevalence, the main difference between this class and tertiary resilience were schooling, age, the search for medical care, and work. The difference in relation to primary resilience was gender, age, and schooling.

The result of latent classes as a function of age bracket is similar to the study by Gooding et al. ²⁸28. Gooding PA, Hurst A, Johnson J, Tarrier N. Psychological resilience in young and older adults. Int J Geriatr Psychiatry 2012; 27:262-70., in which the level of overall resilience and sub-dimensions of emotional regulation and problem-solving were greater in the elderly than in the young. Meanwhile, resilience related to social support was greater in the young than in the elderly.

In the current study, not working was a factor related to primary and secondary resilience, and individuals with this profile also belonged to a higher age bracket. In this case, not working could be a way of anticipating retirement, which is often a worker's goal in the sense of a definitive vacation, since in Brazil, despite all the dissent, pensioners' basic rights are still safeguarded. This is important, because studies that use the "work" variable, like Giuntoli et al. ²⁹29. Giuntoli G, South J, Kinsella K, Karban K. Mental health, resilience and the recession in Bradford. York: Joseph Rowntree Foundation; 2011., evaluated individuals that lost their jobs, in whom they identified high prevalence of depression when assessing mental health and resilience during the recession in Bradford, England. The authors further highlight that men were more reluctant to seek emotional support than women. Meanwhile, young males reported alcohol and drug abuse to deal with the stress. Even acknowledging their stress and depression, many participants refused to seek medical help due to the stigma of mental illness. In the study by Wright et al. ³⁰30. Wright LJ, Zautra AJ, Going S. Adaptation to early knee osteoarthritis: the role of risk, resilience, and disease severity on pain and physical functioning. Ann Behav Med 2008; 36:70-80., no differences were found according to gender in the variables related to resilience.

In Karoly & Ruehlman ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7., which used PCP:S to classify individuals with chronic pain in two groups (resilient and non-resilient), the resilient group had more schooling, while the non-resilient group was significantly more prone to receiving treatment for their pain when compared to resilient individuals (78% of the non-resilient group sought medical treatment, compared to 60% of the resilient group). The authors' findings differ from those of the current article, but this difference should be observed with caution. First, in the study by Karoly & Ruehlman ⁷7. Karoly P, Ruehlman LS. Psychological "resilience" and its correlates in chronic pain: findings from a national community sample. Pain 2006; 123:90-7., although non-resilient individuals sought more medical care, 60% of resilient individuals did likewise. Second, in relation to work, we are apparently seeing cultural differences in the way work and retirement are viewed in the two countries; thus, as found in our study, persons with low schooling are submitted to more physically demanding work than those with more schooling, so it is understandable for them to view retirement as a "prize", "vacation", or "second life" ³¹31. Graeff L. Representações sociais da aposentadoria. Textos Envelhecimento 2002; 4:19-34..

Further, the search for medical care at a specialized center for chronic pain allows individuals to receive adequate therapy for their condition, which can result in the patient's enhanced perception of the pain. Thus, individuals with low schooling may develop a feeling of trust in the physician, an essential condition for successful treatment. The perception of well-being and emotional welfare are factors that help maintain mental health ³²32. Matos A, André R, Cherpe R, Rodrigues D, Figueira C, Pinto A. Estudo psicométrico preliminar da Mental Health Continuum - Short Form - for youth numa amostra de adolescentes portugueses. Psychologica 2010; 53:131-56.^{), (33}33. Vosgerau MZS. Indicadores de bem-estar emocional e doenças crônicas: associação da autopercepção da felicidade, amor e bom humor à condição de saúde de adultos e idosos de Matinhos, Paraná [Tese de Doutorado]. Londrina: Universidade Estadual de Londrina; 2012.. Thus, the perception that one is receiving adequate treatment for a problem has a protective side for an individual's health.

In the analysis of external validity, individuals with fibromyalgia were predominantly classified with tertiary resilience. This appears to relate to the severity of the disease, characterized by diffuse pain in various parts of the body, concurrent presence of myofascial pain, and (although controversial) greater comorbidity and other physical symptoms, such as fatigue and sleep and mood disorders ³⁴34. Bennett RM, Goldenberg DL. Fibromyalgia, myofascial pain, tender points and trigger points: splitting or lumping? Arthritis Res Ther 2011; 13:117.. One can infer from this severity the difficulty in seeking psychological rebalance and state of resilience (reinforced by the fact that individuals with fibromyalgia obtained higher scores in the dimensions of severity, interference, and emotional stress).

Resilience is a complex construct, and it is not possible to establish a universal resilient trait ⁽³⁵35. Rutter M. Implications of resilience concepts for scientific understanding. Ann N Y Acad Sci 2006; 1094:1-12.^{), (36}36. Luthar SS, Brown PJ. Maximizing resilience through diverse levels of inquiry: prevailing paradigms, possibilities and priorities for the future. Dev Psychopathol 2007; 19:931-55.; it is subject to cultural, environmental, and genetic variations, generating multiple and sometimes unexpected paths ³⁷37. Bonanno GA. Loss, trauma, and human resilience: have we underestimated the human capacity to thrive after extremely aversive events? Am Psychol 2004; 59:20-8.. However, the current study's findings corroborate the position of Trivedi et al. ⁶6. Trivedi RB, Bosworth HB, Jackson GL. Resilience in chronic illness. In: Resnick B, Gwyther LP, Roberto KA, editors. Resilience in aging: concepts, research, and outcomes. New York: Springer; 2011. p. 181-97.⁾ and the implications for clinical practice, allowing better patient follow-up. Thus, individuals with the primary resilience profile would only require the usual care and follow-up to guarantee equilibrium and well-being. Individuals with secondary resilience profile may need support and reinforcement in their coping resources, but can largely recover their equilibrium on their own, without ruling out specialized follow-up in their recovery path. The individuals that require the most attention are those with the tertiary resilience profile, since they may need multidisciplinary professional intervention to avoid greater deterioration of their health and well-being.

Conclusions

LCA largely proved useful for developing resilience profiles; in addition, the use of a specific instrument for resilience in chronic pain (PCP:S) contributed to this article's findings. The complexity of the phenomenon in question, resilience, requires a detailed study of biopsychosocial aspects; in this case, socio-demographic variables act as a proxy for these aspects and allow constructing a solid base for understanding resilience. The specific instrument (PCP:S) allows orienting the process of identifying resilience groups.

The current findings' clinical relevance merit special attention, since the resilience profiles reflect different patient needs in terms of clinical follow-up. Again, harm to the individual's mental health increases when moving from primary resilience to the secondary level and finally to the tertiary level.

However, some limitations should be highlighted. Generalization of the results is mainly limited by the fact that the sample consists of individuals from a tertiary care clinic (a specialized service). Another limitation results from lack of identification of the type of pain (neuropathic versus nociceptive), which could shed more light on this phenomenon. One more limitation is lack of inclusion of other pain conditions besides fibromyalgia, myofascial pain syndrome, and chronic tension headache; adding them might have led to the formation of different latent class profiles than those found in the current study.

Finally, since the resilience construct is heavily related to the individual's life history, the article's conclusions should be interpreted with caution since they are not based on longitudinal data. Subsequent studies should thus attempt to reveal individuals' life histories, seeking latent class models for longitudinal data.

References

Publication Dates

History