Abstract - Bethel College
Running head: INFLUENCES OF CLIMATE
Influences of climate variables on the disease symptoms of a male lupus patient
Sarah Buller
Department of Psychology
Bethel College, North Newton, Kansas
Dwight Krehbiel, Faculty Advisor
Abstract
Lupus erythematosus is an autoimmune disease, affecting over one million Americans. Lupus patients commonly report experiencing increased pain with various weather patterns and barometric pressure fluctuations. The present study is a case-study of an eighteen-year lupus sufferer who charted lupus-associated pain, rash presence, and barometric pressure. One goal of the present study is to describe and explain the patterns of pain and rash ratings charted by this patient. Hand-drawn charts were digitized using Engauge digitizing software. Hierarchical linear modeling (HLM) was used to examine relationships among pain, barometric pressure, temperature, dew point, month, season, and rash with years treated as individuals. Temperature was a significant predictor of pain intensity (p = 0.038). When outliers were removed, barometric pressure was a significant predictor of pain intensity (p = 0.042). Analysis of deviance revealed that a model including barometric pressure, temperature, and dew point was the best model for predicting pain intensity (p < 0.000). Month and season were both significant predictors of rash intensity (p < 0.04). Responses to the McGill Pain Questionnaire indicate that pain experiences differ in intensity, not type of pain. This data provided a more comprehensive description of C.W.’s pain experience. Health-related quality of life was measured using the disease-specific questionnaire, the LupusQoL. These data indicated that C.W. has good health-related quality of life. More importantly, free responses from C.W. regarding quality of life indicate that he is effectively managing his disease and learning from his experience. These results indicate the importance of adopting a biopsychosocial perspective on chronic illness, for multiple factors are involved in the disease experience.
Introduction
Systemic lupus erythematosus (SLE) is an autoimmune disease that affects nearly one million Americans. The body becomes allergic to itself, producing an abundance of antibodies that are directed against body tissue. SLE onset usually occurs between the ages of fifteen and forty-five, mainly affecting women in their childbearing years. Almost ninety percent of lupus patients are female, resulting in lupus being called a “women’s disease” (Wallace, 2000, p. 12). The disproportionately large population of female lupus patients has resulted in a wealth of knowledge on the disease experience of women. Studies regarding the biological symptoms of lupus are comprised of predominately female subjects. In addition, research concerning the psychological and social ramifications of lupus is dominated by female perspectives. The health-related quality of life issues facing female lupus patients are well documented, and lupus quality of life questionnaires are developed primarily for these patients. The availability of females with lupus is far greater than males, so it is inevitable that research endeavors are focused primarily on females.
Male lupus patients are less common, and their disease experiences are not well documented. A basic knowledge of gender differences would suggest that the male disease experience is inevitably different from that of females, at least from a psychological and social standpoint. Because of the small male lupus population, studies regarding male disease experiences and quality of life issues are almost nonexistent. The present study aims to address this empirical void by providing a multifaceted analysis of a male lupus experience.
Despite the abundance of knowledge regarding lupus, there still remains much ambiguity regarding the disease experience for both genders. A long-held belief among rheumatologists and patients is the idea of a correlation between symptom flares and weather variables, specifically barometric pressure. The evidence for this phenomenon is largely anecdotal, and the available empirical support offers little clarification for a clear-cut correlation. This phenomenon is addressed in the present study in the hopes of providing further insight into this belief.
Overview of systemic lupus erythematosus
In 1971, the American College of Rheumatology (ACR) developed criteria for defining lupus. After several revisions, the ACR determined that the presence of four of the eleven criteria confirms a diagnosis of systemic lupus erythematosus. Four criteria are skin-related: sun sensitivity, mouth sores, butterfly rash (rash over the cheeks and nose), and discoid rash (thick, disc-like, scarring rash). Four criteria involve specific organ areas, including the lining of the heart or lungs, the kidneys, the central nervous system, and joints. Three criteria are specifically related to relevant laboratory abnormalities, such as abnormal blood counts (low red blood cell, white blood cell, or platelet counts), positive ANA (antinuclear antibodies) testing, and other blood antibody irregularities. The ANA test is the main diagnostic tool for determining if a patient has lupus (Wallace, 2000). In many clinical studies of SLE patients, these ACR criteria are used as a diagnostic tool for inclusion (Hasan et al., 2004; Dobkin et al., 1999; Khanna et al., 2004).
Systemic lupus erythematosus is diagnosed if a patient exhibits both internal and external disease features and ACR criteria are met. SLE can be either organ-threatening or non-organ-threatening. Common symptoms of non-organ-threatening SLE include achiness, fatigue, fevers, difficulty with deep breathing, swollen glands and joints, or rashes. Because their internal organs are not afflicted, these patients can expect to have a normal life expectancy. Organ-threatening SLE involves the heart, lungs, or kidneys, and liver or blood abnormalities may be present. This form of lupus can be fatal if not properly treated. Five to ten percent of patients who fulfill the ACR criteria also have another autoimmune disease, such as scleroderma (tight skin and arthritis), dermato-polymyositis (muscular inflammation), or rheumatoid arthritis (painful joint inflammation) (Wallace, 2000). Disease flares are characteristic of lupus, and such flares are often the subject of clinical studies. Flares are characterized by constitutional symptoms, musculoskeletal involvement, cutaneous (skin) involvement, and low complement levels. Flares can be quantified by SLEDAI (Systemic Lupus Erythematosus Disease Activity Index) scores, a scale used to measure disease activity (Grossman, 2002).
Both the humoral and cellular aspects of the immune system are affected by lupus, and most of the endogenously produced antibodies are directed against cell nuclei. Lupus patients produce many different types of antibodies, but the most common are the antinuclear antibodies (ANA), present in 95% of patients (Trethewey, 2004). The typical lupus patient has a positive ANA test and at least one or two other autoantibodies. Two antibodies that are very specific to SLE are the anti-DNA antibody and the anti-Sm antibody. Anti-DNA antibodies are implicated in the direct damage of bodily tissue, attacking the nuclear DNA that is responsible for protein production. Anti-Sm antibodies interfere with the nuclear transcription process, resulting in the cell’s inability to synthesize RNA from DNA (Wallace, 2000).
Several antibodies to cytoplasmic components are also present in many lupus patients. Common autoantibodies found in SLE patients are directed against the phospholipid cell membrane, function against RNA processing, or affect patient mental functioning. Lupus patients can also have antibodies to specific cell types, including red blood cells, white blood cells, platelets, and nerve cells. Antibodies are also made to antigens expressed by the body’s own cells. The combination of antibody and antigen creates a circulating immune complex that may induce inflammation (Wallace, 2000).
Due to the variable nature of systemic lupus erythematosus, treatment options must be tailored to the individual patient. General treatment options are available, but these depend on the symptoms and severity of the disease. For mild manifestations, non-steroidal anti-inflammatory drugs are usually effective. For severe disease manifestations, high-dose corticosteroid and immunosuppressive drugs are often prescribed. Sun avoidance is also important in preventing lupus flares (Trethewey, 2004).
The Lupus-Weather Connection
Rheumatic diseases and the weather. In searching for answers regarding the causes of lupus pain, many patients and practitioners turn to climatologic influences. In The Lupus Book, Wallace jokes that “patients with rheumatic disease make excellent meteorologists” (2000, p.186). Barometric pressure, temperature, and humidity changes have long been thought to cause lupus patients to experience increased stiffness and aching in the joints. However, empirical data have yet to irrefutably confirm or deny such a pattern.
Several international studies have examined the hypothesized weather and disease activity correlation. In subtropical Israel, Amit et al. (1997) found no seasonal variation for the symptoms of fever, fatigue, weight loss, inflammation, tendon pain, or myalgia. Photosensitivity, a cutaneous disease manifestation, did show seasonal variation, with greater disease symptoms during the summer months. Haga et al. (1999) found the same seasonal pattern in photosensitivity in a population of subarctic SLE patients. Both Amit et al. and Haga et al. found no significant variation in disease activity throughout the year. Amit et al. conclude that while their results revealed limited seasonal patterns for SLE symptoms (specifically photosensitivity), it is possible that there may be seasonal patterns in the disease activity of an individual lupus patient.
Hasan et al. (2004) found that ECLAM (European Consensus Lupus Activity Measurement) scores were significantly higher in spring (p=0.006) and were somewhat higher in the summer (p = 0.051) compared to in the winter. The main conclusion by Hasan et al. was that aggravation of disease activity in SLE patients occurs during the sunny season, and this can be attributed to UV exposure. In addition, SLE disease activation was mostly the result of non-cutaneous reasons and was measurable via laboratory and clinical variables.
These studies indicate similar results regarding photosensitivity: disease activity is influenced by the sun, and thus symptoms are intensified during the summer months. However, photosensitivity is only one of the many possible symptoms of a lupus patient. Seasonal variations in pain and fatigue were not confirmed in these studies, indicating that there is a definite lack in empirical support for hypothesized climate correlations. The literature specifically devoted to SLE and seasonal and climatic correlations is significantly limited. A greater amount of literature is available regarding weather and other rheumatic diseases, specifically rheumatoid arthritis (RA).
Guedj and Weinberger (1990) addressed the influence of barometric pressure, relative humidity, temperature, and rain on pain in sixty-two rheumatic patients. Patients were asked to complete daily questionnaires in which they rated the severity of their joint pain and swelling and their everyday activity ability level. For RA subjects, increases in barometric pressure and temperature were correlated with increased arthritic pain (p < 0.05).
In a study by Aikman (1997), rheumatoid arthritis and osteoarthritis patients of Bendigo, Australia, kept daily records of the severity of their symptoms. Lower temperature and higher relative humidity were significantly correlated with mean pain and mean rigidity, and this finding was supported by previous literature cited by Aikman. Barometric pressure was significantly correlated with mean pain (p < 0.01), and this was a positive relationship (r = 0.1297). The weak correlation between pain and barometric pressure is consistent with the findings of several others, including Guedj and Weinberger (1990).
Like Guedj and Weinberger (1990) and Aikman (1997), Gorin et al. (1997) used a daily diary experimental design to investigate RA patients’ sensitivity to temperature, sunlight, barometric pressure, and relative humidity. At the end of each day, patients completed a book of standard measures, including a rating of the present day’s pain. Weather data were compiled from a local weather service by the experimenters. An interesting addition to the study by Gorin et al. was the multifaceted examination of weather effects on pain: (1) effects of weather on same day pain; (2) effects of previous day’s weather on current pain; and (3) effects of weather changes on next day pain. The authors found weak evidence for a correlation between weather and RA pain, but it was found that pain severity was highest on cold and less sunny days. In addition, pain was more severe after days with higher barometric pressure, and relative humidity change was associated with higher pain ratings. The authors concluded that these findings provide some support for patients’ claims of weather sensitivity, but individual differences in patients may confound a definitive correlation. In addition, the authors emphasized the importance of a future examination of psychological factors that could influence beliefs about the presence of such a weather-pain relationship.
In a study on osteoarthritis pain, McAlindon et al. (2007) found that changes in barometric pressure and ambient temperature influenced knee pain severity. Increased barometric pressure was found to be associated with higher knee pain.
Explanations for a correlation between weather and symptoms. It is apparent that the association between weather variables and rheumatoid disease symptoms is unclear. In examining the association between arthritis and the weather, Aikman (1997) found that 92% of participants believed the weather affected their arthritis symptoms. Participants reported that arthritis symptoms were worse with cold temperatures, humidity, frost, or when the weather changed. Aikman compared this finding to the work of Sibley (1985, as cited in Aikman, 1997). Sibley found that only 62% of a sample included RA and OA patients believed that the weather influenced symptoms. Aikman’s justification for these difference percentages is random error. A belief in the influence of weather on symptoms may be moderated by memory of recent weather changes or disease flares.
Redelmeier and Tversky (1996) question this lack of empirical consensus: why is this belief so prevalent, despite empirical support, and what is maintaining this belief? The authors hypothesized that this belief is partially due to the tendency to perceive patterns where none are present. When presented two uncorrelated sequences and primed with a statement about a negative relationship between barometric pressure and RA pain, 3% of participants identified the sequences as positively related, 79% identified them as negatively related, and 18% identified the sequences as unrelated. The authors attribute this finding to the phenomenon of selective matching; participants focus on prominent coincidences and ignore contrary evidence. Redelmeier and Tversky believe that selective matching can explain both the widespread belief that weather influences RA pain and the lack of empirical support for this phenomenon.
Selective matching is analogous to the concept of illusory correlations. According to Elliot Aronson, illusory correlations occur when we “perceive a relationship between two entities that we think should be related – but…they are not” (2004, p. 115). Illusory correlations confirm one’s original beliefs or stereotypes by allowing a perceived relationship to be sufficient evidence for the belief’s validity. In attempting to explain the systematic errors occurring in psycho-diagnostic tests, Chapman and Chapman (1967) hypothesize that such errors are caused by variables inherent to the stimuli. A study on illusory correlations and stereotypic beliefs by Hamilton and Rose (1980) found that subjects “perceived a pattern of evidence” that tended to validate a prior stereotypic belief. Believing in a particular relationship may result in the seeking out of evidence to support that relationship, even if no such evidence exists (Nickerson, 1998). By taking the concept of illusory correlations into account, a clear relationship between rheumatoid disease symptoms and climate variables may be confounded. In an attempt to explain variability in disease symptoms, patients and physicians may search for logical explanations, settling on a theory that may reflect only a limited period of time. Barometric pressure, temperature, and other climate variables may provide a simple solution for medical phenomena that are truly inexplicable and random.
The phenomenon of illusory causation may also be intertwined in the weather-symptom debate. Illusory causation occurs when causality is inappropriately attributed to a stimulus due to more noticeable or prominent qualities of the stimulus compared to others. A series of social-attribution studies conducted by Lassiter et al. (2002) concluded that illusory causation is “a species of perception” (p. 304). An individual’s point of view dramatically affects the extraction of information from an observed interaction. This subsequently affects judgments regarding causal influences of the involved variables. In regards to rheumatoid disease activity and climate influences, it may be possible that researchers, physicians, and patients are mistakenly attributing causal power to weather. In the absence of clear-cut causative agents in times of disease flares, dramatic climate changes or weather events may mentally stand out, resulting in an attribution of causality to such events.
The weather-symptom debate is further complicated by the absence of a biological explanation for such correlations. The biological mechanism for the explanation of a correlation between weather and rheumatoid disease activity is not well understood. In a study on osteoarthritis pain, McAlindon et al. (2007) state that evidence indicates that barometric pressure affects joint integrity. A study of cadaveric hips by Wingstrand et al. (1990) found the pressure within joints to be below atmospheric pressure in normal situations. When joint pressure was equilibrated to atmospheric pressure, the hip bone was partially dislocated within the joint by eight millimeters. Thus, it was concluded that atmospheric pressure plays an important role in stabilizing the hip joint. McAlindon et al. (2007) looked to the joint pains experienced by divers for some sort of explanation regarding the effects of changing pressure on joint pain. These compression pains may “result from the sudden increase in tissue gas tension surround the joints causing fluid shifts and interfering with joint lubrication” (p. 433). Golde states that temperature could directly affect the surrounding joint structures as well as the viscosity of synovial fluid (fluid surrounding the joint). In addition, temperature may have secondary effects on inflammation by affecting capillary permeability (as cited in McAlindon et al., 2007).
In conclusion, these studies indicate that climatic influences on rheumatoid disease pain are not consistent. Some studies have found that pain and symptoms are influenced by barometric pressure and humidity, while other studies indicate that only cutaneous symptoms show any relationship to seasonal and weather variations. At this time, the biological mechanisms for weather effects on pain are not well understood, but it is clear that biological factors are at work. In an attempt to explain why patients and practitioners still insist on a weather and pain correlation, some authors turn to the phenomena of illusory correlations and causations, arguing that these relationships are maintained by prior beliefs and anecdotal evidence. In spite of the variability in empirical support and the potential influences of illusory phenomena, there is still a fascination with the influence of climate conditions of disease symptoms. Judging from the decades-long duration of this belief, it will probably not be dispelled until definitive contrary evidence is present. Thus, further investigations are absolutely necessary to further elucidate the specific mechanisms and causes for such a relationship.
Pain
Lupus provides a unique challenge for patients and practitioners. The disease experience is highly specific to the individual. The symptoms experienced by one lupus patient may be completely different from those of a second patient, making it difficult to make generalizations about the lupus population as a whole.
Pain. In the past century, psychology has found its way into the medical world as practitioners and researchers are recognizing the cognitive components of pain and other disease states. Biological factors are not enough to explain pain experiences and the associated effects of pain, such as depression and disability. Instead, pain can be better investigated by examining the interplay of biological, psychological, and social variables (Adams, Poole, & Richardson, 2006). Within the biopsychosocial approach to pain, there is a distinct focus on pain behavior and pain management. As opposed to the traditional approach to pain control, in which the ultimate goal is a “cure”, the biopsychosocial approach looks to reduce the daily effects of pain in a patient’s life (Grant & Haverkamp, 1995). Thus, the biopsychosocial conception of pain holds that pain is a “dynamic process that not only is influenced by biological, psychological, and social mechanisms, but also produces biological, psychological, and social changes that, in turn, affect future responses to pain” (Keefe & France, 1999, p. 137).
Pain is a highly personalized experience, thus there is significant variability in patient responses to identical pain-inducing stimuli. Early theories on the psychology of pain stressed the importance of personality, gender, age, and cultural influences on patient perceptions. However, recent research has shown that psychological states may better explain how patients report pain. When presented with a stressful stimulus, like pain, humans naturally respond so as to remove or alleviate the stress. In addition, patients naturally attempt to make sense of their pain experience and achieve an overall sense of understanding. Until this is achieved, the pain experience will continue to disrupt daily function, leading to worry and concern (Eccleston, 2001). A study on SLE patient perspectives on healthcare by Hale et al. (2006) found that female patients struggled when their disease diagnosis was not certain. Women who had significant difficulties obtaining a diagnosis felt a general sense of unease. A definitive diagnosis legitimized these patients’ symptoms, and, according to one patient, “…it was just the relief of knowing they knew what it was…” (p. 386). This study emphasizes the importance of a diagnosis in providing closure and affirmation for a patient.
Measuring pain. Measuring pain is a challenging task. Because pain is a personal experience and an internal sensation, direct observation or measurement is almost impossible. In addition, any measurements of pain are based on subjective data, and these measurements are confounded by the influences of patient emotional responses and societal attitudes toward pain. Therefore, pain reports reflect the integrated influences of the pain trigger, environmental conditions, and individual patient characteristics (McDowell & Newell, 1996). Illusory causations and correlations seem to be inherent to the pain experience.
The McGill Pain Questionnaire (MPG) was developed by Ronald Melzack in 1975 to address the three core psychological components to pain: sensory-discriminative, motivational-affective, and cognitive-evaluative. According to Melzack, pain cannot be merely a sensory experience; instead, pain “has a unique, distinctively unpleasant, affective quality that differentiates it from sensory experiences such as sight, hearing or touch…” (McDowell & Newell, 1996, p.346). Thus, the MPG addresses this issue by addressing such affective dimensions.
The McGill Pain Questionnaire has withstood the test of time, continuing to be widely cited in current pain literature. Research by Reeves et al. (2003) on chronic fatigue syndrome suggests that the McGill Pain Questionnaire is optimal for a characterization of pain. The authors recommend the MPQ because it has demonstrated validity, it is inexpensive and available in several languages, and a shorter form is available. A review on techniques for assessing arthritis knee joint pain states that both the full-length and the short-form MPQ have been used to assess pain in arthritis patients (Neugebauer et al., 2007). In a study of arthritis patients, Wagstaff et al. (1985) found that the pain descriptor words chosen by RA patients were distinct from the pain descriptor words chosen by osteoarthritis patients. The authors concluded that the McGill Pain Questionnaire is sensitive enough to discriminate between similar pain syndromes. This study also indicates that it is appropriate to use the McGill Pain Questionnaire with rheumatoid disease patients.
Pain and fatigue are physical influences on health that have strong psychological components. Individual perceptions of pain and fatigue result in the variability of symptom characteristics. Fatigue and pain are both common SLE symptoms, thus they cannot be ignored by medical practitioners when providing care to these patients. The McGill Pain Questionnaire is a highly regarded quantitative method for measuring pain for various chronic pain conditions, including SLE.
Quality of Life in Systemic Lupus Erythematosus
Why is a patient’s quality of life important in disease management? In adopting a biopsychosocial view of health and wellness, a patient’s quality of life becomes an important concern. Initially, medical practitioners became interested in patient quality of life after the increase in life expectancy. There was a realization that “patients want to live, not merely to survive” (McDowell & Newell, 1996, p. 380). The World Health Organization defined quality of life as “individuals’ perception of their position in the life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns” (as cited in Khanna et al., 2004). In chronic disease conditions, including systemic lupus erythematosus, the physical characteristics of health are not enough to provide the complete scope of one’s disease status. Other social and psychological variables, including pain, apprehension, personal and family relationships, financial responsibilities, and altered cognition, must be considered when assessing health and treatment options (Muldoon et al., 1998).
Influences on quality of life. In systemic lupus erythematosus, a patient’s quality of life is an important concern because of the variable nature of the disease course. The very nature of SLE is inherently stressful. Thus, the patient’s perspective is extremely valuable because of the physical, social, and psychological impact of SLE. When treating SLE, a balance must be established between minimizing the disease symptoms and diminishing the medication side effects. A health-related quality of life measurement allows for patient input in the treatment protocol and aids in open communication with healthcare providers (McElhone et al., 2006).
Social support is a key moderator in health perception. SLE patients often benefit from high levels of social support. Sutcliffe et al. (as cited in Seawell & Danoff-Burg, 2004) found that female SLE patients with greater perceived social support had higher self-reported physical functioning, physical and mental health, and social functioning.
Stress is another potential moderator of health perception and quality of life. There is disagreement among researchers as to whether major or minor life stressors have a greater impact on disease activity. Among SLE patients who experienced disease flares, Pawlak et al. (2003) found that there is an association between greater daily stress (housekeeping, financial duties, family responsibilities, and mood) and the incidence of flares.
Peralta-Ramirez et al. (2004) found that a high percentage of lupus patients perceive a worsening in their clinical disease symptoms due to the effects of daily stress. However, the same perception of symptom worsening was not found for stressful major life events. The authors also found that patients differed in when their symptoms worsened. The effects of stress affect patients on an individual basis. Some patients notice the effects of stress on the same day the stressor occurs, and some may experience a worsening of symptoms up to two days after the occurrence of a stressful event. Still others may never experience a worsening of symptoms after a stressful event. The authors ultimately concluded that a time relationship exists between daily stress and decline in symptoms, yet it must be noted that the disease experience is inherently stressful and a definitive causal link cannot be established.
While these studies confirm the role of stress in disease activity, it cannot be ignored that the participants in these investigations are predominantly female. Because SLE is a “woman’s disease,” affecting a disproportionately larger number of females, this population readily lends itself to empirical studies. However, would the same results be evident if males were participants? The answer to this question is unclear, for the literature on this subject is lacking studies of male SLE patients. Gender differences in stress responses do exist, and these may result in different effects of stress for male SLE patients.
Shelley Taylor, a pioneer in stress research, has proposed that humans possess an affiliative neurocircuitry that promotes affiliation, especially in times of stress. Taylor asserts that humans have a basic, innate need to sustain “protective and rewarding social relationships” (2006, p. 273). The release of the hormone oxytocin in response to some stressors may prompt affiliative behavior, resulting in protective responses toward offspring or seeking meaningful social contact for protection and comfort. These theories have mainly been studied in females and estrogen-treated female animals, so the role of oxytocin in male stress responses is somewhat unclear.
Heinrichs et al. (2003) addressed the question of oxytocin’s effects in males. The authors found that in a stressful condition, male participants who had been administered exogenous oxytocin and had social support exhibited lower cortisol (stress hormone) levels compared to controls. Psychological responses to stress showed similar patterns. Participants given either oxytocin or social support, or both, showed increased calmness and lower anxiety scores compared to controls given a placebo and no social support. Heinrichs et al. concluded that oxytocin may be important for males in moderating the effects of positive social interactions when one is faced with stressors. However, the exogenous administration of oxytocin in this study may weaken the authors’ arguments; the effects found may not be typical of male responses to stress.
Measuring quality of life in SLE. Quality of life measures for SLE patients are very important in medical research, and such measurements are increasingly being included in observational studies and clinical trials. A review of health related quality of life in SLE patients by McElhone et al. (2006) indicates that the generic health-related quality of life (HRQoL) instrument, SF-36, is commonly used. However, the SF-36 is lacking in some areas for assessing key issues faced by many lupus patients, specifically questions relating to sleep and sexual functioning.
According to Wiginton (1999), there are several primary concerns regarding the disease experience of female SLE patients: the uncertainty and unpredictability of the disease, misunderstanding by others, fear, support from family and friends, dependence, and depression. These core concerns cannot be ignored when assessing an SLE patient’s quality of life. Thus, an SLE-specific quality of life questionnaire may better address these concerns, resulting in a more comprehensive gauge of a patient’s disease experience.
McElhone et al. (2007) developed the LupusQoL, a self-report, lupus-specific measure of quality of life. Questionnaire content was derived from SLE patient perspectives on how their disease and treatment affected their life. The final questionnaire included eight domains: physical health, emotional health, body image, pain, planning, fatigue, intimate relationships, and burden to others. Anecdotal patient reports indicate that the LupusQoL is “more meaningful and relevant” compared to the SF-36, thus it may be more sensitive to measuring disease changes over time. The authors specifically mention that while males were not involved in the initial questionnaire construction, it was reviewed by both males and females in the later stages of design.
At this time, two other SLE-specific questionnaires are mentioned in the literature: the SLE Symptom Checklist (SSC) and the SLE-specific Quality of Life instrument (SLEQOL). The SSC is a self-administered symptom questionnaire in which patients denote the presence or absence and the perceived burden of a particular symptom. Thirty-eight symptoms are included in the SSC, and the final score includes the subjective disease burden (Grootscholten et al. 2003). The SLEQOL is a quality-of-life questionnaire developed by rheumatologists and nurse clinicians, and it includes items believed to be relevant to SLE patients. The developers of the SLEQOL note that this questionnaire sufficiently covers the eight domains of the LupusQoL. In addition, compared to the SSC, the SLEQOL is a more comprehensive measure of quality of life, measuring more than just physical symptoms related to lupus. It is recommended to use the SLEQOL in conjunction with the SF-36 to ensure that all aspects of health-related quality of life are addressed (Leong et al. 2005).
Quality of life is an integral part of the biopsychosocial view of health and wellness. In treating and managing lupus, quality of life becomes an extremely important concern because of influences of social and psychological variables. Stress and social relationships can have strong effects on lupus pain and worsening symptoms. When studying quality of life, it is clear that a disease-specific quality of life inventory will better measure a patient’s quality of life. For SLE, the LupusQoL is a reliable questionnaire for measuring health-related quality of life.
Summary, Implications, and Hypotheses
Systemic lupus erythematosus is an autoimmune disease that affects nearly one million Americans. While most lupus patients complain of joint pain and swelling, skin rashes, and fatigue, the disease symptoms vary significantly among individuals. Lupus patients often report increased sensitivity to weather changes or sunlight, but the evidence for this experience is extremely contradictory. Research with lupus patients deals predominantly with the symptom of photosensitivity, while other characteristic symptoms are seemingly ignored. Studies investigating the impact of meteorological variables on SLE symptoms like pain and fatigue are not prevalent. Investigations with rheumatoid arthritis patients have helped to elucidate a potential correlation, but these studies still fail to reach a definitive conclusion. While work with rheumatoid arthritis patients provides valuable insights, it must be noted that lupus and rheumatoid arthritis are inherently different diseases. Therefore, it would be unwise to generalize rheumatoid arthritis findings to the greater lupus population. Thus, the present study aims to address the void in empirical support for a lupus-weather correlation by examining the influences of various climate variables, including temperature, barometric pressure, humidity, and wind speed, on lupus pain and the cutaneous manifestations in a male SLE patient.
Pain is one of the most common symptoms experienced by lupus patients. Because pain is a highly individual experience, it is difficult to describe this symptom. The McGill Pain Questionnaire provides an objective method for quantifying pain, allowing for a numerical value to be assigned to a subjective sensation. In recent studies, the role of stress in provoking disease flares has been of particular interest. Currently, most authors agree that daily stressors have more of an impact on symptoms as compared to major stress events. However, research on lupus pain and the effects of stress have been conducted primarily on females. It is questionable whether these results can be generalized to the male SLE population, due to gender differences in stress responses. A goal of the present study is to add to the knowledge about male SLE patient experiences by examining the pain experience.
Quality of life is a major concern for lupus patients. Assessing one’s quality of life has proved essential in the management of chronic diseases. Quality of life is influenced by psychological and social influences; thus, it is an integral part of a patient’s disease experience. While generic health-related quality of life questionnaires can provide valid data, an SLE-specific questionnaire, like the LupusQoL, will provide more accurate information. As is common in SLE research, much of the lupus quality of life research is devoted to female patients. Males and females have different goals, responsibilities, and concerns in life, thus the SLE disease experience will inevitably have different effects on the genders. The present study will examine quality of life influences on a male SLE patient through the administration of a disease-specific quality of life questionnaire, the LupusQoL.
Main hypotheses for this investigation are as follows:
1. The lupus symptom pain will correlate with climatologic variables, specifically temperature and dew point (humidity).
Research indicates that increasing temperature and increasing relative humidity will affect rheumatic pain. Thus, it is expected that this similar pattern will be found with lupus pain. A strong correlation between barometric pressure and lupus symptoms is not expected; several authors have concluded that barometric pressure only correlates weakly with such symptoms (Aikman, 1997; Guedj & Weinberger, 1990).
2. Seasonal patterns will exist in the weather-symptom correlations. Pain ratings will be highest in summer; rash ratings will be highest in spring.
Previous analyses (Buller, 2008) revealed said pattern. However, this finding was not statistically significant. The addition of more data is hypothesized to strengthen this correlation, potentially to the point of significance. The seasonal pattern for pain can be explained by the variables temperature and humidity. Kansas summers are notoriously hot and humid, and these characteristics are known to affect rheumatic disease patients (Gorin et al., 1997; Guedj & Weinberger, 1990; Aikman, 1997). The seasonal pattern for spring may be explained by photosensitivity, but this is an incomplete conclusion.
3. Significant, stressful life events will affect pain ratings. The presence of stressful life events will be correlated with higher pain intensity ratings.
Stress affects quality of life and can impact disease activity. Pawlak et al. (2003) concluded that greater daily stress is associated with lupus disease flares. The subject of this case-study has indicated that stressful life events may trigger increases in pain.
4. LupusQoL scores will be lowest (i.e. worse health-related quality of life) in the domains of fatigue, pain, and physical health. Scores will be higher (i.e. better health-related quality of life) in the domains of planning and emotional health.
Based on personal statements from a male lupus patient, the domains of fatigue, pain, and physical health are more apt to affect his quality of life. This subject reports that significant pain experiences can impair his daily functioning. His emotional health scores are expected to be higher; he has a well-developed social support network and this will aid in better overall coping and emotional well-being. Because this individual has been diagnosed with lupus for over twenty years, he is better able to manage his disease. While major disease flares do impair his functioning from time to time, it is not likely that his willingness to plan future events is significantly impaired.
Methods
Case history of C.W.
The participant in the present study was a male systemic lupus erythematosus patient, designated C.W. He was diagnosed with lupus in 1989 after a positive ANA and a negative test for Lyme disease. The onset of SLE was believed to be triggered by acute appendicitis and/or the subsequent appendectomy. During the days after the appendectomy, the patient experienced pain in the muscle area just above the knees and a full body rash. Ongoing symptoms have included muscle pain above the knees, lichen planus (oral sores or rash) in the mouth, which is aggravated by certain foods, and fatigue. These symptoms have varied in intensity over time. C.W. periodically experiences lupus flares, relatively short (approximately twelve to thirty-six hours) experiences of intense, often full-body pain. The participant uses Salsalate and aspirin for pain control, usually for only a few days at a time. Various prescription drugs were also taken during longer periods of intense pain.
C.W. began charting in May 1991. Charts initially included the degree of aching leg pain and the severity of skin rash, recorded daily. Daily barometric pressure measurements and various weather patterns were later added to the charts after the patient noticed a correlation between disease flares and certain weather types. Various life events, including vacations, dental surgery, and major life changes, were included in charts. Prescription drugs and other disease symptoms were also noted. C.W. began charting after reading of the benefits of journaling or charting as a way of maintaining a certain degree of control over the disease pain and promoting a balanced lifestyle. Charting was generally done in the evening, with barometric pressure recorded first followed by pain and rash intensity ratings.
Procedure
C.W.’s hand-drawn charts from July 1993 to September 2007 were scanned and digitized using Engauge digitizing software (). The time plots of barometric pressure, pain, and rash were all digitized. Using the software, axes were defined and numerical values were assigned to the points on the charts. An arbitrary 0-10 scale was used for pain and rash charts; 0 indicated no pain/rash and 10 corresponded with high pain or rash intensity. Values higher than ten occurred and were quantified by the software accordingly. The barometric pressure of central Kansas was measured in millibars and the chart coordinates corresponded to participant measurements of pressure using a Swift Instruments, Inc. Scientist Model #477 barometer. C.W. never had the barometer calibrated post-manufacture, but measurements were assumed to be consistent. The Engauge software created numerical data that were imported into a spreadsheet file. Numerical values for barometric pressure, pain, and rash intensity all corresponded to a specific date. An individual file was created for each chart (approximately six months of data).
Dew point and temperature data were obtained from the World Meteorological Organization (WMO) World Weather Watch Program (Federal Climate Complex). Mean dew point and mean temperature were recorded for each day in degrees Fahrenheit, with tenths of a degree precision. These means were derived from hourly observations, and a minimum of four observations per day were required to calculate a mean. Extensive automated quality control was conducted to eliminate any errors contained in the original data sets. All dew point and temperature means were calculated by the WMO. Data recordings were made at Newton, KS Station #724509, located at latitude 3803N, longitude 09717W, and at an elevation of 467 meters (wetterzentrale.de/klima/stnlst.html). An Excel spreadsheet was used to organize these data.
A composite file was created in Excel to integrate the data obtained from Engauge with the WMO data. This composite file barometric pressure, mean temperature, mean dew point, pain rating, rash rating, season, and the presence or absence of health and life events. Health events included dental surgery and illness. Life events included vacation times and moving to a new home. Seasons were numerically coded; winter was “1”, spring was “2”, and so forth. This composite file was a level one file, which was analyzed using Hierarchical Linear Modeling (HLM 6.0 software, Scientific Software International). Because of the nature of the case-study design, years were treated as individuals in the HLM models.
Throughout C.W.’s disease experience, there were extended periods of time when pain and rash intensity were relatively stable. It was speculated that these periods (often weeks long) could possibly interfere with the influence of periods of elevated pain on pain and rash intensities. To address this issue, a second spreadsheet file was created in which only “peak” pain episodes were considered in analysis. These pain incidents were characterized by pain ratings greater than six. The weather data corresponding to the major pain experiences were also included in the second spreadsheet file. Peak pain episodes were identified by a number, and all episodes were nine days in length. Extremely long peak pain episodes (length of time was greater than eight days) were excluded from this analysis.
C.W.’s charts included gaps of time with no data. These dates were retained, but the corresponding barometric pressure, rash, and pain columns were left empty. Days that had been inadvertently omitted by C.W. from the charts were added to the composite file to ensure that months had the same number of days from year to year. To ensure that all the years were of equal length, the data from February 29th, 1996, February 29th, 2000, and February 29th, 2004, were omitted. Pain and rash intensity ratings were made daily by C.W., resulting in a total of 5,175 days of data for analysis.
HLM was used to the test the predictor variables (season, month, barometric pressure, mean temperature, and mean dew point) and their effects on the outcome variables (pain and rash). This was a correlation analysis. All models were saved, and significant or almost significant interactions were illustrated graphically using R software (R Development Core Team, 2007). The criterion for statistical significance was a p-value of 0.05 or less.
The McGill Pain Questionnaire was administered to C.W. in February 2008. The patient was instructed to fill out the entire questionnaire to reflect his worst pain experience ever (summer 2007). The questionnaire was then shortened to include only questions about present pain. C.W. was instructed to fill out these shorter questionnaires every other day for approximately two weeks. In total, seven short questionnaires were completed. The shorter questionnaires included space for the patient to note present symptoms, the current weather, and barometric pressure. The McGill Pain Questionnaire is included in Appendix A.
The data from the McGill Pain Questionnaires were scored according to the methods outlined in Melzack’s “The McGill Pain Questionnaire: Major Properties and Scoring Methods” (1975). The questionnaires were scored by adding the rank values of pain descriptor words in the four categories (sensory, evaluative, affective, and miscellaneous). The scores for the individual categories were summed to obtain a total score, the pain rating index (PRI-R). The Pain Rating Intensity Score (PRI-S) was be calculated by summing the scale values for all the words chosen in each of the four categories. The total number of words chosen (NWC) was calculated. Average test-retest consistency for the McGill Pain Questionnaire is 70.3%. A high correlation was found between the scale- and rank-value methods for scoring the McGill Pain Questionnaire, indicating the validity of each scoring method (PRI-R, PRI-S, NWC, and PPI). The Pearson correlations among the four scoring procedures were greater than 0.9. Thus, the different scoring procedures provide complementary data regarding a pain experience (McDowell & Newell, 1996).
The LupusQoL quality of life questionnaire was self-administered to C.W. in January 2009. The questionnaire was in paper form and an instruction handout accompanied the questionnaire. The LupusQoL can be completed in less than ten minutes (McElhone et al., 2007). The experimenter was not present during the completion of the questionnaire. C.W. was specifically asked for his comments regarding the questionnaire. The LupusQoL is included in Appendix B.
The LupusQoL questionnaire contains thirty-four questions and an open-ended section for comments. The questionnaire comprises eight domains: physical health (questions 1-8), pain (questions 9-11), planning (questions 12-14), intimate relationships (questions 15-16), burden to others (questions 17-19), emotional health (questions 20-25), body images (questions 26-30), and fatigue (questions 31-34). The questionnaire has been found to demonstrate good reliability and validity. The eight domains have good internal reliability, with Cronbach’s alpha coefficients ranging from 0.88 to 0.95. The questionnaire was found to have good discriminant validity; post hoc analysis found that it discriminated between different degrees of disease severity. Good concurrent validity was demonstrated with comparable domains correlating with the SF-36; r-values ranged from 0.71 to 0.79. Test-retest reliability was good for all domains, with interclass correlations ranging from 0.72 to 0.93 (McElhone et al., 2007).
The questionnaire responses were scored according to the protocol outlined in McElhone et al. (2007). The response format for the LupusQoL is a 5-point Likert scale: 0 = all of the time, 1 = most of the time, 2 = a good bit of the time, 3 = occasionally, and 4 = never. For each domain, item response scores are summed and a mean raw domain score is calculated. This mean raw domain score is then converted to a transformed domain score, which has a value ranging from 0 (worst health-related quality of life) to 100 (best health-related quality of life). This calculation is done according to the following formula:
If 50% of the items are left unanswered, a transformed domain score cannot be calculated. Instead, the item responses are totaled, and a mean raw domain score is calculated. An item response of “not applicable” is treated as an unanswered item.
The data obtained from the LupusQoL was compared to the data cited by McElhone et al. (2007). In developing the questionnaire, the authors utilized a sample of 269 female lupus patients. Summary data and statistics are available for this sample. C.W.’s transformed domain scores were compared to mean domain scores reported by McElhone et al.
Results
Analysis of influences on pain and rash
HLM models were constructed to include Level 1 and Level 2 variables. Figure 1 displays one of the models created during data analysis; this model is characteristic of the analysis. This model investigated the statistical significance of barometric pressure on pain intensity ratings. In this model, pain is the outcome variable and barometric pressure is the predictor variable. This model generated output that is displayed in Table 1. When outliers were removed, barometric pressure was found to be a significant predictor of pain, see Table 1, (INTRCPT2, B10). Outliers were defined as barometric pressure values below 980 millibars; nine outliers fit this criterion.
[pic]
Figure 1. HLM model with pain as the outcome variable
Table 1. HLM output pertaining to the model in Figure 1
Final estimation of fixed effects:
---------------------------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
---------------------------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 3.970827 0.285527 13.907 14 0.000
For BAR slope, P1
INTRCPT2, B10 -0.015477 0.006909 -2.240 14 0.042
---------------------------------------------------------------------------------------------
The scatterplot below (Figure 2) displays barometric pressure as a predictor of pain. To minimize the outliers in the pain variable, the logarithm of pain was calculated. In the following scatterplot, the logarithm of pain is used as the outcome variable. Greater pain intensity ratings seem to cluster around mid-range values for barometric pressure.
[pic]
Figure 2. Scatterplot of barometric pressure vs. log of pain for all data points (excluding barometric pressure outliers)
When peak pain episodes (pain rating was greater than or equal to six) were isolated and analyzed, barometric pressure was not a significant predictor of pain, model coefficient = -0.0203, t(88) = -1.785, p = 0.077. By using the transformed pain rating (i.e. the logarithm of pain), this relationship is strengthened, model coefficient = -0.00172, t(88) = -1.860, p = 0.066. According to the criteria established for outliers, four data points in the peak pain data set had barometric pressure values less than 980 mmHg. These outliers were not removed from analysis, but it is believed that the correlation between barometric pressure and peak pain intensity ratings would be strengthened if these outliers were removed (for complete HLM outputs, refer to Appendix C, Figures C1-C2).
Mean daily temperature was tested as a predictor of pain. When all data were included in the analysis, temperature was a statistically significant predictor of pain, t(14) = 2.290, p = 0.038. The model coefficient for the relationship between pain and temperature was positive (0.0135), indicating that higher temperatures are associated with greater pain ratings. The scatterplot below (Figure 3) displays temperature as a predictor of log pain. Although there is considerable scatter around the regression line, there is a clear positive slope to the regression line (for complete HLM outputs, refer to Appendix C, Figure C3).
[pic]
Figure 3. Scatterplot of temperature vs. log pain for all pain ratings
Analysis of peak pain episodes revealed a strengthening of the correlation between pain and temperature. Again, temperature was a statistically significant predictor of pain, model coefficient = 0.0179, t(88) = 2.543, p = 0.013 (for complete HLM outputs, refer to Appendix C, Figure C4).
Dew point was also tested as a predictor of pain. Dew point was almost a statistically significant predictor of pain when all data were included in analysis, model coefficient = 0.0144, t(14) = 2.120, p = 0.052. The coefficient for the relationship between dew point and pain was positive, indicating that greater pain ratings are associated with a higher dew point. High dew point is associated with high relative humidity. Relative humidity is a more universal concept than dew point, thus it is easier to conceptualize the implications of a higher relative humidity. Therefore, the positive coefficient for the relationship between dew point and pain can also be interpreted as the association of higher pain ratings with days of higher relative humidity. Figure 4 is a scatterplot of dew point and log pain ratings. A positive regression line is visible among the dense scatter, reflecting the overall trend of higher pain ratings associated with higher dew points (for complete HLM outputs, refer to Appendix C, Figure C5).
[pic]
Figure 4. Scatterplot of dew point vs. log pain for all data points
In the analysis of peak pain episodes, the relationship between dew point and pain was weakened, model coefficient = 0.0135, t(88) = 1.875, p = 0.064. Again, dew point was not a significant predictor of pain (for complete HLM outputs, refer to Appendix C, Figure C6).
The analysis of peak pain episodes and the analysis of the entire data set revealed similar correlations. To clarify and better illustrate these correlations, the individual days of the peak pain data were studied. Daily means were calculated for the four variables (pain intensity, temperature, barometric pressure, and dew point). There were nine days (numbered one through nine) for each peak pain episode, and each day corresponded to a mean value for pain intensity, temperature, barometric pressure, and dew point. Figure 5 displays the trends of the daily means. Mean pain intensity increases during the nine days, culminating in a high pain rating for day nine. Mean temperature and mean dew point both increase during the nine day period, and a slightly lower mean temperature and mean dew point are evident on day nine. Barometric pressure means decrease during the nine day period. These trends are reflected in the analyses of both the peak pain episodes and the entire data set. Higher pain ratings were significantly correlated with higher temperatures. Dew point was positively correlated with pain intensity ratings, and although this correlation was not significant it indicates that higher dew points are associated with increased pain. When outliers were removed, barometric pressure was significantly correlated with pain intensity, and this was a negative relationship.
[pic]
Figure 5. Means for pain intensity, temperature, barometric pressure, and dew point for the nine days of the peak pain episodes
The statistical tests presented provide valuable information regarding the influence of climate variables on pain symptoms. The significant interactions are evidence for the predictive value of each climate variable. However, there is some hesitance regarding the predictive power of these individual variables (barometric pressure, temperature, and dew point). Small effects may be labeled as significant due to the enormous amount of data included in analysis. By examining the validity of different models, a conclusion may be drawn about what influences on pain intensity are most important. By including different combinations of climate variables, a particular model’s ability to predict pain intensity may be enhanced. Analysis of deviance calculations allows for the efficacy of a particular model to be compared to that of another model. Because temperature was a statistically significant predictor of pain, this model was used as a starting point. In the final estimation of variance components, temperature had a deviance of 19051.186540 with 4 estimated parameters. This model was compared to a model including both temperature and dew point as predictors of pain. When included together, this model was significantly better at predicting pain rating intensity than a model with temperature alone, Χ²(3) = 66.13405, p < 0.000. Finally, this model was compared to a model that included barometric pressure, temperature, and dew point as predictors of pain. The combination of these three climate variables resulted in a model that was significantly better at predicting pain intensity compared to the model with temperature and dew point, Χ²(4) = 505.64350, p < 0.000. Thus, individual climate variables may offer less insight into pain intensity ratings, contrary to initial conclusions. It seems as though a model consisting of all three climate variables is the most accurate predictor of pain intensity ratings (for complete HLM outputs, refer to Appendix C, Figures C7-C9).
Time analyses were conducted to determine if month or season were predictive of pain ratings. Months were assigned a numerical value between one and twelve (January corresponded to 1, February to 2, and so forth). Months were grouped into seasons, and each season was given a numerical value. Winter was denoted “1” and included December, January, and February. Spring was denoted “2” and included March, April, and May. Summer was denoted “3” and included June, July, and August. Autumn was denoted “4” and included September, October, and November. While these divisions do not reflect the seasonal divisions of the vernal and autumnal equinoxes, the divisions adequately reflect the seasonal fluctuations of central Kansas. When all data were included in analysis, month was not predictive of the variance in pain ratings, model coefficient = -0.0343, t(14) = -0.948, p = 0.360. From this analysis, it appeared that pain intensity ratings were higher during the summer months. The season variable was renumbered to reflect this trend, assigning summer the highest value (“4”) and autumn the lowest value (“1”). Analysis with this new season variable revealed a stronger correlation between season and pain ratings, although the correlation was still not significant, model coefficient = 0.011041, t(14) = 1.438, p = 0.172 (for complete HLM outputs, refer to Appendix C, Figures C10-C11).
Analysis of peak pain events revealed that season and month were not significant predictors of periods of high pain intensity ratings, month model coefficient = 0.0428, t(88) = 0.907, p = 0.367, season model coefficient = 0.0545, t(88) = 0.296, p = 0.768 (for complete HLM outputs, refer to Appendix C, Figures C12-C13).
Skin rash was the second lupus symptom charted by C.W. According to C.W., skin rash severity was not believed to be associated with climate variables. Statistical analysis supported this inference. Barometric pressure was not a significant predictor of the variation in rash intensity, model coefficient = -0.002089, t(14) = -0.555, p = 0.587. Temperature was not a significant predictor of rash, model coefficient = 0.000978, t(14) = 0.283, p = 0.781. Similarly, dew point was not a significant predictor of rash intensity, model coefficient = 0.000847, t(14) = 0.243, p = 0.811. Models consisting of combinations of these three climate variables further supported the conclusion that the variation in rash intensity is unable to be predicted by the climate variables of barometric pressure, temperature, and dew point (for complete HLM outputs, refer to Appendix C, Figures C14-C16).
The same logarithmic transformation was conducted for rash intensity ratings to minimize the impact of outliers on analysis. Time variables were found to have a greater impact on skin rash intensity ratings. Month was a significant predictor of rash intensity (log-scale), model coefficient = -0.006454, t(14) = -2.314, p = 0.036. The negative model coefficient indicates that rash intensity ratings were higher in months with lower numerical values. Figure 6 depicts rash intensity ratings by month. The spring months (labeled 3, 4, and 5) have a relatively higher number of outliers compared to the other months, indicating the presence of more variable and intense rash experiences. It is important to note that there is a visible difference in mean rash intensity ratings during the various months. Rash intensity rating means are visibly lower in the autumn months compared to in the spring months (for complete HLM outputs, refer to Appendix C, Figure C17).
[pic]
Figure 6. Rash intensity ratings (log scale) by month
Season was a significant predictor of rash intensity, model coefficient = -0.152, t(14) = -2.279, p = 0.039. Figure 7 depicts this relationship. Again, a greater number of outliers are visible during spring (season 2) (for complete HLM outputs, refer to Appendix C, Figure C18).
[pic]
Figure 7. Rash intensity ratings (log scale) by season
The disproportionate number of outliers among seasons results in a quadratic trend in the boxplot. For this reason, the season variable was squared and a second model was run. When season was squared (i.e. winter corresponded to “1”, spring to “4” summer to “9”, and autumn to “16”), it was a more significant predictor of rash intensity, model coefficient = -0.0034, t(14) = -2.426, p = 0.030. The boxplot of this relationship is identical to Figure 7 (for complete HLM outputs, refer to Appendix C, Figure C19).
To further investigate the influence of month and season on rash intensity, peak rash events were isolated. Rash events with intensity ratings greater than or equal to eight were selected and designated “peak rash events”. The number of peak rash events per month and per season was calculated. A Chi-squared statistical analysis was conducted to determine if there was significant variation in the number of peak rash events per month and per season. The total number of peak rash events was 158, resulting in a null hypothesis stating that there would be no month-to-month variation in the number of peak rash events, resulting in an expected 13.17 peak rash events per month. Chi-squared analysis of month revealed that there was a significant difference in the number of peak rash events per month, Χ²(11) = 19.675, p < 0.005. For the analysis of variation in the number of peak rash events per season, the null hypothesis stated that there would be an expected 39.5 peak rash events per season. Chi-squared analysis of season revealed that there was a significant difference in the number of peak rash events per season, Χ²(3) = 12.838, p < 0.005.
In his charts, C.W. made notations of significant life-changing events, including vacations and work-related changes, and major health-related events, including surgeries, illnesses, and changes in medication. Two qualitative variables (Health and Life) were created to reflect these events. The presence of a health or life event was indicated by the numerical value “1”; the absence of such events was indicated by a “0”. The health variable was not found to be a significant predictor of pain, model coefficient = 0.189, t(14) = 0.716, p = 0.485. The life variable was also not a significant predictor of pain intensity, model coefficient = -0.142, t(14) = -0.432, p = 0.672. The life variable was not particularly accurate; many life events occurred during periods when C.W. was not charting. Thus, it was difficult to determine whether this variable truly influenced pain intensity ratings (for complete HLM outputs, refer to Appendix C, Figures C20-C21).
When only peak pain experiences were considered, the presence of life or health events had different effects on pain. The presence of a health event was a significant predictor of pain, model coefficient = 0.956, t(88) = 2.098, p = 0.038. The absence of a life event was a significant predictor of pain, model coefficient = -0.926, t(88) = -2.597, p = 0.011. This finding is not particularly revealing. The peak pain events generally were not associated with major life events, as evident in the spreadsheet file. When both health and life events were included in a model to predict pain, only health events proved to be significant predictors, model coefficient = 0.967, t(88) = 2.136, p = 0.035. This indicates that health and life events are carrying different information, confirming the integrity of experimenter coding of such variables. These analyses reveal that when C.W. is experiencing unusually intense pain, other non-lupus health factors may be involved as well (for complete HLM outputs, refer to Appendix C, Figure C22-C24).
McGill Pain Questionnaire
The results from the McGill Pain Questionnaires indicate that pain experiences are quite variable form day to day. C.W. completed the McGill Pain Questionnaire protocol during February 2008. The questionnaire was also completed in a retrospective manner for summer 2007, believed to be the period when C.W. experienced the worst lupus pain of his life.
Table 2 displays the results from the McGill Pain Questionnaires. The numerical values provide a quantitative index of pain, and this allows mild pain experiences to be better compared with more intense pain experiences. The words in the top row are pain categories designated by the McGill Pain Questionnaire. The number of words chosen does not necessarily correlate with the total pain rating index. For example, the questionnaire for summer 2007 had nine words chosen and the questionnaire for February 14th, 2008, also had nine words chosen. However, the PRI-R score for summer 2007 is almost three times higher than the PRI-R score for February 14th. For February 8th, 2008, through February 22nd, 2008, the mean PRI-R score was seven, the mean PRI-S score was 11.50, and the mean number of words chosen was 5.7 words.
Table 2. Pain rating index values for MPQ
| Date |Sensory |Affective |Evaluative |Miscellaneous |PRI-R |PRI-S |NWC |
|Summer 2007 |11 |7 |5 |9 |32 |30.33 |9 |
|2/8/2008 |7 |0 |1 |1 |9 |10.95 |5 |
|2/10/2008 |1 |0 |1 |3 |5 |9.34 |5 |
|2/12/2008 |3 |0 |0 |2 |5 |10.65 |5 |
|2/14/2008 |7 |1 |1 |2 |11 |18.84 |9 |
|2/16/2008 |7 |1 |1 |3 |12 |18.67 |9 |
|2/20/2008 |3 |0 |0 |2 |5 |9.09 |5 |
|2/22/2008 |1 |0 |0 |1 |2 |2.95 |2 |
Examination of this data revealed that days characterized by more intense pain have higher PRI-R, PRI-S, and NWC scores. Summer 2007, the period in which C.W. experienced his worst pain ever, has a much higher PRI-R and PRI-S compared to the scores of February 2008.
For the purposes of this study, the numerical output of the McGill Pain Questionnaire is of secondary importance to the actual descriptor words chosen by C.W. These words attempt to describe C.W.’s pain experience in a way that is not shown in his charts. During summer 2007, C.W.’s charts revealed pain intensity ratings above 10. However, these charts only express that C.W. was experiencing worse pain than usual. The type of pain experienced is unclear without the data of the McGill Pain Questionnaire. The pain descriptor words chosen by C.W. to describe his pain in summer 2007 and in February 2008 were often from the same category. The pain words describing summer 2007 are words of greater intensity. This indicates that C.W.’s pain during intense disease flares is not a different kind of pain but a much more intense pain. For example, pain in summer 2007 was evaluated as “unbearable”, a descriptor with a rank value of five. Pain during February 2008 was evaluated as “annoying” (rank value of one). This data is displayed in Appendix D.
Lupus Quality of Life Inventory
C.W. was self-administered the Lupus Quality of Life Questionnaire (LupusQoL). In addition to completing the questionnaire, C.W. made extensive comments regarding the nature of the questionnaire. The data from the questionnaire are displayed in Table 3. Missing responses occurred only in the body image domain; therefore, this domain could not be scored.
Table 3. Domain scores for the LupusQoL inventory
|LupusQoL domains |Transformed domain |Missing responses |
|(no. of items in domain) |score | |
|Physical health (8 items) |81.25 |- |
|Emotional health (6 items) |95.83 |- |
|Body image (5 items) |- |Yes |
|Pain (3 items) |66.67 |- |
|Planning (3 items) |75 |- |
|Fatigue (4 items) |68.75 |- |
|Intimate relationships (2 items) |87.5 |- |
|Burden to others (3 items) |75 |- |
With the exception of emotional health, C.W.’s LupusQoL domain scores fell within the normal curve for the scores cited by McElhone et al. (2007). For most domains, C.W. had scores higher than the mean and closer to one standard deviation from the mean. Higher domain scores represent better health-related quality of life. The best quality of life, according to the LupusQoL, is a domain score of 100.
C.W. noted that the questionnaire’s conceptual framework depicted the lupus disease experience in a “negative” manner. He noted the predominant use of negative words to phrase questions; such words included “unable”, “need help”, “lost”, “difficulty”, et cetera. He stated that this negative conceptual frame could be balanced by a complementary questionnaire designed to test the effect of the negative frame on participants’ responses. In his disease experience, C.W. has learned to adopt a positive outlook toward life, and this greatly influences his attitude toward the effects of lupus on his daily functioning. For example, C.W. confronts lupus-created challenges with creative approaches, developing new skills and designing innovative solutions. He is able to translate these problem-solving techniques to other areas of his life. Thus, lupus has proved to be a learning tool for C.W., and he would probably not regard the experience as entirely negative. In addition, because of the constraints of lupus, C.W. has adopted a slower paced lifestyle. Such a lifestyle could be beneficial to everyone, and C.W. believes that the advantages of lessening one’s load far outweigh the negatives.
Discussion
Interactions between climate variables (barometric pressure, temperature, and dew point) and lupus symptoms (pain and rash) proved to be complex. The correlations discovered through analysis of the entire dataset are remarkable. The significant or nearly significant results indicate that climate variables have some influence on lupus pain symptoms.
The first experimental hypothesis was both supported and contradicted. Climatologic variables proved to be significant predictors of pain intensity ratings. Temperature and barometric pressure were statistically significant predictors of pain intensity, and dew point was an almost statistically significant predictor. Positive model coefficients for temperature and pain intensity and dew point and pain intensity support the hypothesized correlations. The results of these analyses confirm the hypotheses of C.W. Several interviews with C.W. have revealed that he experiences worse pain during the summer months, which are characterized by high heat and humidity. The climate variables of barometric pressure, temperature, and dew point did not prove to be significant predictors of skin rash intensity ratings. This finding is not surprising. Pain and rash symptoms are inherently different, and it can be concluded that different endogenous and exogenous factors are present for these symptoms. Furthermore, the literature on SLE cutaneous disease manifestations does not correlate changes in climate variables with aggravated skin symptoms. These symptoms are generally attributed to ultraviolet light exposure (Hasan et al., 2004, Amit et al., 1997).
The finding that greater temperatures are correlated with higher pain intensity ratings corresponds to research with rheumatoid arthritis patients (Guedj & Weinberger, 1990). While rheumatoid arthritis and lupus are inherently different, they still fall under the blanket category of rheumatic diseases. Temperature may influence those with different rheumatoid diseases through a similar mechanism, modulating the effects of disease pain in an analogous manner. However, it cannot be concluded that high temperatures will always be correlated with increased pain experiences. Other studies on rheumatoid arthritis have indicated that lower temperatures are associated with greater pain (Aikman, 1997; Gorin et al., 1997).
Dew point was an almost significant predictor of pain intensity. The empirical support for this finding is contradictory. Studies with rheumatoid arthritis and osteoarthritis patients found that high relative humidity may be associated with greater pain (Aikman, 1997). However, low relative humidity has also been found to correlate with greater rheumatoid arthritis pain experiences (Gorin et al., 1997).
When outliers were removed, barometric pressure was a statistically significant predictor of pain intensity. C.W. was skeptical that a clear relationship existed between pain symptoms and barometric pressure. When he began charting, C.W. hoped to use his records as a predicting tool. However, he has been unable to use barometric pressure data to predict possible pain flares. He did indicate that the greater variation in pressure during the winter months may contribute to decreases in pain intensity.
Aikman’s study of rheumatoid arthritis and osteoarthritis patients concluded that barometric pressure was significantly correlated with pain (1997). A scatterplot of mean pain ratings (n = 25) versus barometric pressure revealed a clustering of points around mid-range pressure values. This pattern is similar to the data of C.W. However, Aikman’s study examined correlations within individuals, making these results difficult to compare to the present study. In accordance with Aikman, Guedj and Weinberger (1990) cited a positive correlation between barometric pressure and arthritic pain.
Studies of lupus patients have yet to confirm similar relationships between pain intensity and climate variables. However, hypotheses regarding climate influences on SLE pain have yet to be rejected. Amit et al. (1997) found no seasonal variation in lupus symptoms across a sample of 105 patients. However, it was concluded that seasonal patterns in disease activity may occur for individual lupus patients. Variability among individuals may pose problems for investigations of universal relationships regarding disease symptoms. Thus, case-by-case analyses may provide a more accurate picture of the disease experience compared to means across many SLE patients.
The experimental hypothesis regarding seasonal patterns in disease manifestations is both supported and contradicted. Pain intensity ratings did not show a seasonal pattern. A previous study including approximately half of C.W.’s data revealed a seasonal pattern in pain intensity (Buller, 2008), but such a pattern was not supported by the relevant literature (Amit et al., 1997). Therefore, the conclusion that pain intensity does not vary by season is not surprising.
Seasonal and monthly patterns for rash intensity were present. Higher rash intensity ratings were characteristic of the spring months (March, April, and May) as compared to the other months. Studies with lupus patients report a seasonal pattern in photosensitivity, a cutaneous disease symptom. Haga et al. (1999) and Amit et al. (1997) both found greater skin symptoms during the summer months. In these studies, ultraviolet light exposure was believed to be involved in this seasonal aggravation of disease activity.
For C.W., the rash variable most likely does not reflect photosensitivity. It is a measure of a skin rash localized to the arms and legs, areas not prone to extended sun exposure. In addition, C.W. is extremely conscientious about sun exposure, limiting time in the sun and wearing protective clothing. Determining the etiology of this symptom may require examination of immunological factors. Seasonal variations in immunological functions and adrenocortical function have been reported (Amit et al., 1997), but without this data for C.W., it is uncertain what biological changes are implicated in his rash manifestations.
The seasonal pattern for C.W.’s rash symptoms is different from the conclusions of Haga et al. (1999) and Amit et al. (1997). Interestingly, Amit et al. examined several other cutaneous manifestions, including the butterfly rash and bodily rash. Neither of these symptoms showed a seasonal pattern. This seasonal pattern may be unique to C.W., further proof of the variability in disease experiences among lupus patients. Amit et al. conclude their research with reference to this variability, stating that seasonal patterns in disease activity may exist for individual lupus patients.
The presence of life and health events proved to be significantly correlated with peak pain experiences. The presence of a health event (i.e. dental surgery, influenza, shingles, etc.) was a significant predictor of pain. Inherently painful or distressing health situations are likely to exacerbate any lupus pain already being experienced. In addition, the disease process can be triggered by viruses or bacteria if the individual is genetically predisposed to lupus. It is not uncommon for lupus flares to be associated with other types of infections (Wallace, 2000). While it is unknown whether C.W. carries the lupus “gene”, it is probable that other infectious agents may induce lupus flares. However, the etiology of all flares cannot be attributed to infections. Other causative agents are at work, and disease triggers include environmental agents, medications, excessive ultraviolet light exposure, and physical or emotional stress (Wallace, 2000).
The absence of a major life event was a significant predictor of pain. This is a counterintuitive finding but is easily explained. Life events were numerically coded as “present” (1) or “absent” (0). Life events included vacations, moving, and other stressful times. Unfortunately, C.W. did not consistently chart during these periods. Vacation periods, for example, never had pain, barometric pressure, or rash data available. Thus, the inherent flaws in the life variable prevented a reliable or significant prediction of disease pain. If C.W. had continued to chart during such periods, a different conclusion may have been reached. However, prior research indicates that daily stress, not major stressful life events, is a greater predictor of lupus flares (Pawlak et al., 2003; Peralta-Ramirez et al., 2004). In interviews, C.W. indicated that prolonged stressful situations (for example, deadlines at work or changes in supervisors) affect his disease experience. However, such stressors are not always noted in his charts, resulting in limited predictive power of the life events variable. Thus, the experimental hypothesis regarding the influence of stressful life events on pain intensity cannot be confirmed.
In this investigation, statistical significance may not be as revealing as the validity of the statistical models. Analysis of deviance was used to evaluate the predictive power of the models. Even though temperature was a statistically significant predictor of pain intensity, this model did not prove to be the very best model for predicting pain. The model including barometric pressure, temperature, and dew point proved to be the best predictor of pain intensity. All three climate variables are essential to understanding pain intensity, and these variables interact in complex ways to influence pain symptoms. Yet again, the complex nature of lupus is illustrated.
The consideration of only “peak” pain experiences drastically reduced that amount of data used in analysis. By examining specific pain events, periods of time in which pain intensity ratings remained relatively static were eliminated from the dataset. This prevented any climate-pain correlations from being obscured by long periods of relatively minimal change. Correlations between pain and barometric pressure, temperature, and dew point were all strengthened when such invariable periods were removed. This more focused analysis raises an interesting question: can temperature exert a greater influence during some periods of time and not others? Logically, the answer to such a question is “no.” If climate variables have a bearing on disease symptoms, their influence over time should be consistent. However, statistical analysis reveals that such climate-pain relationships are stronger during these specific periods, resulting in an affirmative answer to the above question. Thus, it cannot be concluded that climate variables are the primary cause of variability in pain experiences. Climate variables more likely will interact with existing disease states, either aggravating or possibly alleviating symptoms. Such variables may modify pain through the interaction with other factors. The interplay between exogenous (i.e. climate) and endogenous factors is incredibly complex, and the scope of the present study is unable to address such an interaction.
The McGill Pain Questionnaire data provided a different account of lupus pain. As stated previously, the lupus disease experience is highly individual. The symptoms experienced by one patient may be drastically different from those experienced by another. The data provided by the McGill Pain Questionnaire attempt to illustrate the complex nature of lupus symptoms. In C.W.’s case, the McGill Pain Questionnaire data were more descriptive than the pain data of the charts, providing specific and varying adjectives to qualify the pain experience. This allowed for a deeper understanding of the C.W.’s charts.
During the two-week period that C.W. completed the questionnaires, his pain was self-described as fairly stable and mild. The adjective words chosen to describe this pain matched the initial description. There was minimal variation in word choice, and words from the same categories were repeatedly picked. When this relatively mild period was compared to summer 2007, a period of intense pain, it was evident that the descriptor words chosen were of a greater intensity. However, the words chosen to describe the intense pain were still generally within the same categories as pain words used to describe mild pain. Thus, the quality of the pain remains relatively stable while the intensity in pain can change dramatically over time.
The LupusQoL questionnaire provides a broader frame of reference for C.W.’s disease experience. Living with lupus is all-consuming. The disease experience is a part of each and every day. While some days are “good”, other days are absolutely unbearable. The variable nature of the disease can be quite stressful and inevitably affects every aspect of one’s life. Physical health, including symptom management, is a primary focus of living with lupus. However, psychological and social health is a considerable concern as well.
As indicated in his LupusQoL domain scores, C.W. has an overall good quality of life. In his comments regarding the questionnaire, it is apparent that C.W. has learned to not let lupus control his life. While he has had to adopt new techniques for seemingly simple tasks, he regards such adjustments in a positive light. He approaches the challenges of lupus with a positive mindset, viewing such trials as an opportunity for creative problem solving. In addition, he has learned to appreciate the merits of a slower-paced lifestyle. The variability in the disease course of lupus has provided opportunities for C.W. to appreciate the unpredictability of life.
One of C.W.’s major assets is his inquisitive attitude toward his disease experience. The desire to gain information about his condition has allowed C.W. to exert some control over his disease. The original goal of charting was to achieve a degree of control over his pain while encouraging a healthy lifestyle. By searching for tools and knowledge, C.W. has been able to successfully manage his lupus for eighteen plus years. This is quite an accomplishment, for it is evident that the variable nature of the disease does not make it easy to live with.
The data provided by C.W. are incredibly unique and valuable. A dataset of this magnitude is not typical, resulting in the distinct nature of this investigation. Cited literature utilizes a similar self-report method, asking patients to keep records of symptoms and severity (Guedj and Weinberger, 1990; Aikman, 1997; Gorin et al., 1997). However, the scope of these studies pales in comparison to the records kept by C.W. Eighteen years of data are currently available, and C.W. has no intentions of ceasing his recordings.
The studies listed above have some advantages over the charts of C.W. The self-report method of these studies is more focused, and climate variables are obtained by the experimenters, not the participants. Thus, participants’ response biases are diminished. In addition, disease activity measurements were not subjective. Studies involving lupus patients utilized the SLE Disease Activity Index (Amit et al., 1997; Haga et al. 1999) or the European Consensus Lupus Activity Measurement score (Hasan et al., 2004). These measurements were devoid of patient ratings of disease activity. When subjective disease activity reports were utilized, the “diaries” were focused, including explicit instructions regarding when to complete the surveys. Standard measurement scales or Likert scales were implemented in pain ratings, and detailed instructions accompanied such scales (Gorin et al., 1999; Aikman, 1997).
Because C.W. was self-motivated to begin data collection, his charts lack the integrity present in the cited laboratory investigations. While C.W. attempted to be consistent in his ratings, he did not assign numerical values to pain or rash intensity ratings. Thus, the data collection and subsequent analysis rest heavily on the assignment of an arbitrary 0-10 scale for C.W.’s ratings. There was assumed to be consistency from chart to chart, but this consistency could be questionable. In a written overview of his charting process, C.W. states that his pain ratings are somewhat relative. Spring of 2007 (April through June) represented an ultimate peak in pain, and previous points noted at the same height of the spring 2007 curve probably represented a lower level of pain. Thus, C.W.’s pain ratings may not be absolutely reliable.
Time of measurement is also a concern with C.W.’s charts. C.W. tries to make recordings one time per day, usually during the evenings. However, he admits that during periods of lesser pain, he will not chart everyday and make retrospective ratings every couple of days. In times of elevated pain, as during spring 2007, C.W. charts every day. The order of recording was not always consistent. Between 1993 and 2003, the barometer was not in a convenient location for charting. During this period, C.W. did not always chart barometric pressure, pain, and rash measurements in the same order. This lack of consistency is advantageous, providing an element of randomization to the charting process. The data recording process of 2003 to the present is more consistent. C.W. charts barometric pressure first, followed by pain and rash ratings. Despite this consistency, there is no way to know if this order has been violated from time to time. By charting barometric pressure first, there may be a risk for a self-fulfilling prophesy phenomenon. The knowledge of the day’s pressure may influence the subjective pain rating. The inconsistency in charting order during 1993-2003 may act as a buffer against this potential bias. Even though he does not actively chart temperature and humidity values, C.W. has his own hypotheses regarding these variables. Beliefs about the effects of temperature and humidity may also influence pain ratings. While temperature and humidity data were obtained from an external source, these variables can be easily experienced. By simply venturing outside on a hot and humid day, C.W.’s appraisal of pain may be biased.
A major limitation of the present study is the case-study design. It is unlikely that the disease experience of C.W. is identical to the experience of another lupus patient. Consequently, external validity in this study was severely compromised. However, this limitation may also be one of the greatest strengths of this study. An in-depth, meticulous analysis was devoted to a single, male lupus patient, resulting in high internal validity. By considering only the experience of one individual, relationships between symptoms and other variables were not obscured by the inclusion of other participant data. The larger samples sizes of studies involving lupus and other rheumatic disease patients may actually hinder the validity of such studies. Lupus research seldom includes male patients, as they are the minority of the afflicted. Thus, the male disease experience is not well understood or well documented. While this study cannot be generalized to the male lupus population as a whole, it does provide a comprehensive picture of what the male disease experience could be like.
In the future, the climate, pain, and rash data could be supplemented with various laboratory variables. The addition of laboratory variables would allow for a more complete examination of the disease experience. In addition, such variables are objective, unlike the pain and rash ratings of C.W. Disease activation could be monitored by both objective and subjective measures, resulting in a more comprehensive description of lupus activity. Immunological function may be influenced by seasonal variation. A longitudinal study on in vitro immune response by Van Rood et al. (1991) found significant monthly variation in several immune functions. Percentages of CD4 and CD8 lymphocytes showed monthly variation. In addition, the proliferative immune response to mitogens demonstrated monthly variation. Mitogens are polypeptide (protein) growth factors that promote cell proliferation, ultimately affecting DNA transcription and cell-cycle control (Lodish et al., 2008). A more complete analysis of lupus symptoms would be possible with the availability of both climate and laboratory variables.
It is important to remember the value of adopting a biopsychosocial approach toward health and wellness. In this study, the biological and psychological components are represented by the climate, pain, and rash analysis and McGill Pain Questionnaire data, respectively. Both of these elements seem to fit within the greater social-psychological context of one’s perceived quality of life. All three are constantly interacting: biological factors are moderated by psychological interpretations and perceived social support and well-being, and vice versa. Thus, one measure alone is not enough to address the lupus disease experience. All three components are vital pieces in the lupus puzzle.
The present study is an important addition to the existing knowledge on lupus. Correlations between lupus symptoms and climate variables were not empirically supported, yet patients and practitioners still believe in such relationships. By confirming that these correlations do indeed exist, medical professionals can be more confident in assertions regarding the influences of weather on lupus. Differences between individuals regarding climate influences are expected. However, the case of C.W. provides a foundation for possible patterns of climate influences. Lupus patients and their medical support team can use this study to better clarify and understand individual sensitivities to weather changes. The present study will provide C.W. with valuable information regarding his own disease experience. His personal hypotheses regarding the influences of climate variables on pain symptoms were confirmed. The knowledge that higher temperatures and humidity levels are correlated with more intense pain may allow for superior planning in the future. In addition, knowledge of the monthly and seasonal patterns for rash symptoms may also aid in future planning.
An incredible amount of data was included in the present study. It is highly unlikely that a similar set of data exists. The detailed tracking of lupus symptoms over time provides valuable insights into the disease experience of a male lupus sufferer. The majority of lupus research is devoted to female patients, due to the disproportionate number of female SLE sufferers. As a result, literature involving primarily male lupus subjects is not available. The present study addresses this issue, providing information regarding almost the entire disease experience of a male lupus sufferer. Ultimately, there is a hope that this study may bring about a renewed sense of knowledge and control for C.W. and for similar lupus patients.
References
Adams, N., Poole, H., & Richardson, C. (2006) Psychological approaches to chronic pain management: Part 1. Journal of Clinical Nursing, 15(3). Retrieved February 12, 2008, form PsycINFO.
Aikman, H. (1997). The association between arthritis and the weather. International Journal of Biometerology, 40, 192-199.
Amit, M., Molad, Y., Kiss, S., & Wysenbeek, A.J. (1997). Seasonal variations in manifestations and activity of systemic lupus erythematosus. British Journal of Rheumatology, 36, 449-452.
Aronson, E. (2004). The social animal. New York: Worth Publishers.
Buller, S. (2008). Seasonal variation in daily ratings of pain and rash of a lupus patient. In R.D. Yearout (Ed.), Proceedings of the National Conference on Undergraduate Research. University of North Carolina at Asheville.
Chapman, L.J., & Chapman, J.P. (1967). Genesis of popular but erroneous psychodiagnostic observations. Journal of Abnormal Psychology, 72(3), 193-204.
Dobkin, P.L., Da Costa, D., Dritsa, M., et al. (1999). Quality of life in systemic lupus erythematosus patients during more and less active disease states: Differential contributors to mental and physical health. Arthritis Care and Research: The Official Journal of the Arthritis Health Professions Association, 12(6), 401-410.
Eccleston, C. (2001). Role of psychology in pain management. British Journal of Anaesthesia, 87(1), 144-152.
Federal Climate Complex. Global surface summary of day data [Data file]. Retrieved from
Gorin, A.A., Smyth, J.M., Weisberg, J.N., et al. (1999). Rheumatoid arthritis patients show weather sensitivity in daily life, but the relationship is not clinically significant. Pain, 81, 173-177.
Grant L.D., & Haverkamp, B.E. (1995). A cognitive-behavioral approach to chronic pain management. Journal of Counseling and Development, 74(1). Retrieved February 12, 2008, from PsycINFO.
Grootscholten, C., Ligtenberg, G., Derksen, R., et al. (2003). Health-related quality of life in patients with systemic lupus erythematosus: Development and validation of a lupus specific symptom checklist. Quality of Life Research: An International Journal of Quality of Life Aspects of Treatment, Care & Rehabilitation, 12(6), 635-644. Retrieved October 1, 2008.
Grossman, J.M., & Kalunian, K.C. (2002). Definition, classification, activity, and damage indices. In D.J. Wallace & B.H. Hahn (Eds.), Dubois’ lupus erythematosus (19-31). Philadelphia: Lippincott Williams and Wilkins.
Guedj, D., & Weinberger, A. (1990). Effect of weather conditions on rheumatic patients. Annals of the Rheumatic Diseases, 49, 138-139.
Haga, H.J., Brun, J.G., Rekvig, O.P., & Wetterberg, L. (1999). Seasonal variations in activity of systemic lupus erythematosus in a subarctic region. Lupus, 8, 269-273.
Hale, E.D., Teharne, G.J., Lyons, A.C., et al. (2006). “Joining the dots” for patients with systemic lupus erythematosus: Personal perspectives of health care from a qualitative study. Annals of the Rheumatic Diseases, 65, 585-589.
Hamilton, D.L., & Rose, T.L. (1980). Illusory correlation and the maintenance of stereotypic beliefs. Journal of Personality and Social Psychology, 39(5), 832-845.
Hasan, T., Pertovaara, M., Yli-Kerttula, U., et al. (2004). Seasonal variation of disease activity of systemic lupus erythematosus in Finland: A 1 year follow up study. Annals of the Rheumatic Diseases, 63, 1498-1500.
Heinrichs, M., Baumgartner, T., Kirschbaum, C., & Ehlert, U. (2003). Social support and oxytocin interact to suppress cortisol and subjective responses to psychosocial stress. Biological Psychiatry, 54, 1389-1398.
Keefe, F.J., & France, C.R. (1999). Pain: Biopsychosocial mechanisms and management. Current Directions in Psychological Science, 6(5), 137-141.
Khanna, S., Pal, H., Pandey, R.M., & Handa, R. (2004). The relationship between disease activity and quality of life in systemic lupus erythematosus. Rheumatology, 43, 1536-1540.
Lassiter, G.D., Geers, A.L., Munhall, P.J., et al. (2002) Illusory causation: Why it occurs. Psychological Science, 13(4), 299-305.
Leong, K.P., Kong, K.O., Thong, B.Y.H., et al. (2005). Development and preliminary validation of a systemic lupus erythematosus-specific quality-of-life instrument. Rheumatology, 44, 1267-1276.
Lodish, H., Berk, A., Kaiser, C.A., et al. (2008). Molecular cell biology. New York: W.H. Freeman and Company.
Melzack, R. (1975). The McGill pain questionnaire: Major properties and scoring methods. Pain, 1, 277-299.
McAlindon, T., Formica, M., Schmid, C.H., & Fletcher, J. (2007). Changes in barometric pressure and ambient temperature influence osteoarthritis pain. The American Journal of Medicine, 120, 429-434.
McDowell, I. & Newell, C. (Eds.) (1996). Measuring health: A guide to rating scales and questionnaires. New York: Oxford University Press.
McElhone, K., Abbott, J., & Teh, L-S. (2006). A review of health related quality of life in systemic lupus erythematosus. Lupus, 15, 633-643.
McElhone, K., Abbott, J., Shelmerdine, J., et al. (2007). Development and validation of a disease-specific health-related quality of life measure, the LupusQoL, for adults with systemic lupus erythematosus. Arthritis & Rheumatism, 57(6), 972-979.
Muldoon, M.F., Barger, S.D., Flory, J.D., & Manuck, S.B. (1998). What are quality of life measurements measuring? BMJ, 316, 542-545. Retrieved October 31, 2008, from
Neugebauer, V., Han, J.S., Adwanikar, H., et al. (2007). Techniques for assessing knee joint pain in arthritis. Molecular Pain, 3(8). Retrieved October 15, 2008, from
Nickerson, R.S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175-220.
Pawlak, C.R., Witte, T., Heiken, H., et al. (2003). Flares in patients with systemic lupus erythematosus are associated with daily psychological stress. Psychotherapy and Psychosomatics, 72(3), 159-165.
Peralta-Ramirez, M.I., Jiménez-Alonso, J., Godoy-García, J.F., & Perez-García, M. (2004). The effects of daily stress and stressful life events on the clinical symptomatology of patients with lupus erythematosus. Psychosomatic Medicine, 66, 788-794.
R Development Core Team (2007). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL .
Redelmeier, D.A., & Tversky, A. (1996). On the belief that arthritis pain is related to the weather. Proceedings of the National Academy of Science USA, 93, 2895-2896.
Reeves, W.C., Lloyd, A., Vernon, S.D., et al. (2003). Identification of ambiguities in the 1994 chronic fatigue syndrome research case definition and recommendations for resolution. BMC Health Services Research, 3(25). Retrieved October 15, 2008, from
Seawell, A.H., & Danoff-Burg, S. (2004). Psychosocial research on systemic lupus erythematosus: A literature review. Lupus, 13, 891-899.
Taylor, S.E. (2006). Tend and befriend: Biobehavioral bases of affiliation under stress. Current Directions in Psychological Science, 15(6), 273-277.
Trethewey, P. (2004). Systemic lupus erythematosus. Dimensions of Critical Care Nursing, 23(3). Retrieved January 22, 2008, from Academic Search Premier.
Van Rood, Y., Goulmy, E., Blokland, E., et al. (1991). Month-related variability in immunological test results; Implications for immunological follow-up studies. Clinical and Experimental Immunology, 86, 349-354.
Wallace, D.J. (2000). The lupus book: A guide for patients and their families. New York: Oxford University Press.
Wiginton, K.L. (1999). Illness representations: Mapping the experience of lupus. Health Education & Behavior, 26, 443-453.
Wingstrand, H., Wingstrand, A., & Krantz, P. (1990). Intracapsular and atmospheric pressure in the dynamics and stability of the hip: A biomedical study. Acta Orthopaedica, 63(3), 231-235.
Appendix A
McGill Pain Questionnaire (Melzack, 1975)
[pic]
[pic]
[pic]
[pic]
[pic]
[pic]
Appendix B
LupusQoL Questionnaire (McElhone et al., 2007)
[pic]
[pic]
Appendix C
Figure C1. HLM output for interaction of pain intensity ratings and barometric pressure (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.023515 0.154614 32.491 88 0.000
For BAR slope, P1
INTRCPT2, B10 -0.020281 0.011365 -1.785 88 0.077
----------------------------------------------------------------------------
Figure C2. HLM output for interaction of log pain intensity ratings and barometric pressure (peak pain episodes only)
The outcome variable is LOGPAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 0.653140 0.012455 52.439 88 0.000
For BAR slope, P1
INTRCPT2, B10 -0.001717 0.000923 -1.860 88 0.066
----------------------------------------------------------------------------
Figure C3. HLM output for interaction of pain intensity ratings and temperature
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 3.957004 0.260391 15.196 14 0.000
For TEMP slope, P1
INTRCPT2, B10 0.013784 0.006019 2.290 14 0.038
----------------------------------------------------------------------------
Figure C4. HLM output for interaction of pain intensity ratings and temperature (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.999258 0.156452 31.954 88 0.000
For TEMP slope, P1
INTRCPT2, B10 0.017928 0.007050 2.543 88 0.013
----------------------------------------------------------------------------
Figure C5. HLM output for interaction of pain intensity ratings and dew point
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 3.945256 0.256276 15.395 14 0.000
For DEW slope, P1
INTRCPT2, B10 0.014357 0.006772 2.120 14 0.052
----------------------------------------------------------------------------
Figure C6. HLM output for interaction of pain intensity ratings and dew point (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.998812 0.155570 32.132 88 0.000
For DEW slope, P1
INTRCPT2, B10 0.013525 0.007212 1.875 88 0.064
----------------------------------------------------------------------------
Figure C7. HLM output for analysis of deviance of model for the interaction of pain and temperature
Final estimation of variance components:
-----------------------------------------------------------------------------
Random Effect Standard Variance df Chi-square P-value
Deviation Component
-----------------------------------------------------------------------------
INTRCPT1, R0 1.03863 1.07875 14 1542.94127 0.000
TEMP slope, R1 0.02349 0.00055 14 302.16168 0.000
level-1, E 1.80722 3.26603
-----------------------------------------------------------------------------
Statistics for current covariance components model
--------------------------------------------------
Deviance = 19051.186540
Number of estimated parameters = 4
Figure C8. HLM output for analysis of deviance between a model for the interaction of pain and temperature and a model for the interaction of pain, temperature, and dew point
Final estimation of variance components:
-----------------------------------------------------------------------------
Random Effect Standard Variance df Chi-square P-value
Deviation Component
-----------------------------------------------------------------------------
INTRCPT1, R0 1.01623 1.03272 14 1465.84464 0.000
TEMP slope, R1 0.02957 0.00087 14 60.12078 0.000
DEW slope, R2 0.04450 0.00198 14 104.81771 0.000
level-1, E 1.78939 3.20191
-----------------------------------------------------------------------------
Statistics for current covariance components model
--------------------------------------------------
Deviance = 18985.052486
Number of estimated parameters = 7
Variance-Covariance components test
-----------------------------------
Chi-square statistic = 66.13405
Number of degrees of freedom = 3
P-value = 0.000
Figure C9. HLM output for analysis of deviance between a model for the interaction of pain, temperature, and dew point and a model for the interaction of pain, temperature, dew point, and barometric pressure (barometric pressure outliers removed)
Final estimation of variance components:
-----------------------------------------------------------------------------
Random Effect Standard Variance df Chi-square P-value
Deviation Component
-----------------------------------------------------------------------------
INTRCPT1, R0 1.02320 1.04695 14 1187.70342 0.000
BAR slope, R1 0.02841 0.00081 14 82.17244 0.000
TEMP slope, R2 0.03055 0.00093 14 62.05510 0.000
DEW slope, R3 0.04154 0.00173 14 92.69539 0.000
level-1, E 1.76052 3.09943
-----------------------------------------------------------------------------
Statistics for current covariance components model
--------------------------------------------------
Deviance = 18479.408988
Number of estimated parameters = 11
Variance-Covariance components test
-----------------------------------
Chi-square statistic = 505.64350
Number of degrees of freedom = 4
P-value = 0.000
Figure C10. HLM output for interaction of pain intensity ratings and month
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.000970 0.273663 14.620 14 0.000
For MONTH slope, P1
INTRCPT2, B10 -0.034258 0.036134 -0.948 14 0.360
----------------------------------------------------------------------------
Figure C11. HLM output for interaction of pain intensity ratings and season2 (autumn=1)
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.348715 0.224354 23.840 14 0.000
For SEASON2 slope, P1
INTRCPT2, B10 0.107939 0.075046 1.438 14 0.172
----------------------------------------------------------------------------
Figure C12. HLM output for interaction of pain intensity ratings and month (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.996470 0.154271 32.388 88 0.000
For MONTH slope, P1
INTRCPT2, B10 0.042795 0.047201 0.907 88 0.367
----------------------------------------------------------------------------
Figure C13. HLM output for interaction of pain intensity ratings and season (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.015867 0.157412 31.865 88 0.000
For SEASON slope, P1
INTRCPT2, B10 0.050495 0.170520 0.296 88 0.768
----------------------------------------------------------------------------
Figure C14. HLM output for interaction of rash intensity ratings and barometric pressure
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.342257 0.214913 24.858 14 0.000
For BAR slope, P1
INTRCPT2, B10 -0.002089 0.003766 -0.555 14 0.587
----------------------------------------------------------------------------
Figure C15. HLM output for interaction of rash intensity ratings and temperature
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.327704 0.220287 24.185 14 0.000
For TEMP slope, P1
INTRCPT2, B10 0.000978 0.003458 0.283 14 0.781
----------------------------------------------------------------------------
Figure C16. HLM output for interaction of rash intensity ratings and dew point
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.324072 0.219935 24.207 14 0.000
For DEW slope, P1
INTRCPT2, B10 0.000847 0.003480 0.243 14 0.811
----------------------------------------------------------------------------
Figure C17. HLM output for interaction of rash intensity ratings and month
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.383977 0.252513 21.322 14 0.000
For MONTH slope, P1
INTRCPT2, B10 -0.071853 0.035015 -2.052 14 0.059
----------------------------------------------------------------------------
Figure C18. HLM output for interaction of rash intensity ratings and season
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.323234 0.221864 23.993 14 0.000
For SEASON slope, P1
INTRCPT2, B10 -0.154840 0.079918 -1.938 14 0.073
----------------------------------------------------------------------------
Figure C19. HLM output for interaction of rash intensity ratings and season2
The outcome variable is RASH
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 5.326970 0.223286 23.857 14 0.000
For SEASON_2 slope, P1
INTRCPT2, B10 -0.034844 0.016633 -2.095 14 0.055
----------------------------------------------------------------------------
Figure C20. HLM output for interaction of pain intensity ratings and health events
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 3.955090 0.266147 14.861 14 0.000
For HEALTH slope, P1
INTRCPT2, B10 0.188826 0.263588 0.716 14 0.485
----------------------------------------------------------------------------
Figure C21. HLM output for interaction of pain intensity ratings and life events
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 3.957017 0.252926 15.645 14 0.000
For LIFE slope, P1
INTRCPT2, B10 -0.141875 0.328288 -0.432 14 0.672
----------------------------------------------------------------------------
Figure C22. HLM output for interaction of pain intensity ratings and health events (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.959948 0.148422 33.418 88 0.000
For HEALTH slope, P1
INTRCPT2, B10 0.955918 0.455540 2.098 88 0.038
----------------------------------------------------------------------------
Figure C23. HLM output for interaction of pain intensity ratings and life events (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.998370 0.154605 32.330 88 0.000
For LIFE slope, P1
INTRCPT2, B10 -0.926344 0.356716 -2.597 88 0.011
----------------------------------------------------------------------------
Figure C24. HLM output for interaction of pain intensity ratings, health events, and life events (peak pain episodes only)
The outcome variable is PAIN
Final estimation of fixed effects (with robust standard errors)
----------------------------------------------------------------------------
Standard Approx.
Fixed Effect Coefficient Error T-ratio d.f. P-value
----------------------------------------------------------------------------
For INTRCPT1, P0
INTRCPT2, B00 4.960757 0.148208 33.472 88 0.000
For HEALTH slope, P1
INTRCPT2, B10 0.967939 0.453141 2.136 88 0.035
For LIFE slope, P2
INTRCPT2, B20 -0.518169 0.723205 -0.716 88 0.475
----------------------------------------------------------------------------
Appendix D
This table depicts the words chosen to describe pain. The pain rank index score (PRI-R) is shown in parentheses for each word.
[pic]
-----------------------
mean raw domain score
× 100 = transformed score for domain
4
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- bethel medical patient portal
- bethel family medicine patient portal
- bethel medical family practice
- bethel medical group
- bethel family medical
- bethel family practice
- bethel family practice bethel maine
- bethel medical practice
- bethel ct school ratings
- bethel family medicine
- bethel medical portal
- bethel medical care