INTERNATIONAL JOURNALOF BEHAVIORAL MEDICINE, 5(Yhlg5 203 Copyright 9 1998. Lawrence Erlbaum Asmciates, Inc.

EMPIRICAL CONTRIBUTIONS

Pain and Stress in Sickle Cell Disease: An Analysis of Daily Pain Records Laura S. Porter, Karen M. Gil, Jan A. Sedway, Jawana Ready, Elizabeth Workman, and Robert J. Thompson, Jr. This study examined daily reports of pain, medication use, heaIth care use, and activity reduction in adults with sickle cell disease, and their association with stress. Participants were 53 adults with sickle cell disease. They completed the Daily Hassles questionnaire at the start of the study, and they kept daily records of pain and pal n response over the following 14 days. On average, patients reported pain on 6.5 days of the 14-day study period. The average pain intensity rating during a painful episode was 4.4 on a 10-point scale. Pain was most often managed at home. Patients took medicat.ion (analgesics and/or narcotics) on 80% of the days they experienced pain. and they were more likely to use medication, particularly narcotics, as pain levels increased. At higher pain levels some patients also utilized a range of health care services. On average, patients also cut back considerably on household and social activities, especially when pain reached a level of over 5 on the 10-point scale. Those who were employed. however, were likely to continue to work, even when in pain. In addition, stress had significant positive associations with average pain intensity as well as reductions in household and social activities. Furthermore. stress predicted activity reductions even after controlling for pain intensity. Stress was unrelated to medication and health care use in this study.

Key words: sickle cell disease. $CD. chronic pain. stress, daily diaries Laura S. Porter. Ks~,n M Gil, Jan A. Sedway, Jaw~a Ready, and ElizabeJh Workman, Department of Psychology. University of North Carolina at Chapel Hill. NC. USA; Robert J. Thomr~on. Jr. Duke Univc~ity Medical Center, Durham. NC, USA. Correspondence concerning this article should be addrt.~sed to Karcn M. Gil, Department of Psychology. CB 3270. University of North Carolina. Chapel H i l l NC 27599-3270.

1 B6

POR1'EREr AL

There is a long history of the use of daily diaries in recording day-to-day health and health-related events. Health diaries have been used to compare retrospective and prospective reporting of events, as memory aids to improve recall of health events in a later retrospective interview, to record responses to experimental treatment in treatment-outcome studies, and as a prima,%, data source (Burman, 1995; Muschet, Miller, Clements, Pait, & Gutterman, 1996; Norman, McFartane, Streiner, & Neale. I982; Verbrugge, 1980). Daily diaries have been particularly useful in studies of chronic pain (Affleck, Tennen, Urrows, & Higgins, 1994; Stone, Broderick, Porter, & Kaell, 1997; U~rows, Affleck, Tennen, & Higgins, 1994). These studies examined patterns and variability in pain as weII as the association between pain, mood, and stress. Recently, daily diaries have been used to describe the course and complications of sickle cell disease in children (Fuggle, Sband, Gill, & Davies, 1996; Gil I. Shand, Fuggle, Dugan, & Davies, 1997; Shapiro et al., 1990; Shapiro, Dinges, Orne, Ohene-Freemong, & Orne, 1990; Shapiro et al., 1995). Sickle cell disease (SCD) is a major health problem, affecting about one in 500 African Americans. SCD is a genetic disorder detected through newborn screening. Patients often begin to experience symptoms and/or prophylactic treatment (e.g., penicillin for infection) during early infancy. The most frequent, intractable problem encountered by patients with SCD is vasocclusive pain. Most patients experience several unpredictable and severe painful episodes each year, and some patients seek frequent emergency treatment for pain and may require repeated and prolonged hospitalization. There is currently no cure or prevention for SCD pain. Shapiro et al.'s (1990, 1995) daily diary studies provided unique information regarding pain and functional impairment in children with SCD. Children in these studies completed daily diaries about the presence and intensity of their pain, their use of medication, and their school attendance for a period of up to I 0 months. Resuits indicated that pain occurred frequently, on about 25% of the days, with an average pain intensity of 40 on a 100-point scale. Pain was most often managed at home. In addition, children missed, about one fifth of their school days, although absences were not always associated with SCD pain. The authors noted that, based on the pervasiveness of pain in everyday life, these patients may be more accurately described as having chronic rather than acute pain. Similarly, Gill et al. (1997) found that children with SCD reported pain an average of 10 out of 28 days (36%) using a diary method. They noted that the average frequency of pain events reported by parents and children in retrospective interviews was only half of that reported in the daily diaries. As part of a pain management program including both children and adults with SCD (Dinges et al., t997), patients completed daily diaries during a 4-month baseline phase prior to intervention. The majority of patients in this study (76%) were children and adolescents, and the participants were described as "individuals with high rates of severe pain that cannot be managed entirely at home," (p. 420). Analysis of

PAIN IN SICKLE CELL DISEASE

] 87

the baseline diary data indicated that, on average, patients experienced SCD pain on 20% of the days, with an average pain intensity rating of 47 on a 100-point scale. On average, patients took medication on 59% of the clays they had pain. Age was a significant predictor of pain days and pain intensity, with children reporting fewer pain days and less intense pain than adults. In general, when SCD pain is measured retrospectively or restricted to pain that results in a health care contact, it is described as considerably less frequent. Platt et al. (199I) defined pain episodes as pain lasting at least two hours and requiring a health care contact. Using these criteria, they found that adults with SCD have less than one painful episode each year. Gil, Abrams, Phillips, & Keefe (1989) and Gi[, Abrams, Phillips, and Williams (1992) reported that adult SCD patients had an average of one to two episodes per month. These findings were based on a retrospective Structured Pain Interview (Gil, 1994) in which pain episodes were defined as pain that lasted at least 2 hours and included episodes treated at home as well as at health care facilities. The first purpose of this study was to analyze patterns of pain, health care use. medication use, and functional impairment in adults with SCD. The use of daily diaries provides a number of advantages over both retrospective interuiews and reliance on medical records. There is some evidence that memory tends to be poor for pain and medication use, especially when recall periods are greater than 1-2 weeks (Linton & Melin, 1982; Ready, Sarkis, & Turner, 1982), and that recall may be affected by the current state of the patient (rEich, Jaeger. & Graff-Radford, 1985). Smith and Safer (1993) found that recall of pain intensity was less r~liable than data collected from prospective measures evaluating pain intensity. Use of daily diaries may help reduce memory recall bias in reporting on pain and other factors (Stone, Kessler, & Haythornthwaite, 1991; Verbvagge, I980). Daily diaries also provide more accurate information than hospital records about the duration, frequency, and severity of illness events, as well as the health actions that are taken to cope with such events (Norman, el at., 1982; Verbrugge, 1980; White, Williams, & Greenberg, 1961 ). I nter'~iews and hospital records may only record occurrences of chronic or acuteiliness events if they result in medical attention or activity restriction (Verbrugge, 1980; White el at., 1961). Additionally, many health problems, including those related to SCD, may be self-treated and not reported to health care providers (Norman et al., 1982; Shapiro et al., ] 995: Stoller, Forster, & Portugal, 1993: Verbrugge, 1980: White et at., 1961). The second purpose of this study was to analyze the relation between psychological stress and pain and pain response. Although a link between stress and SCD pain has been suspected (e.g., Midence & Elander, 1994; Nash, 1994), there have been few empirical studies of this question. Nadel and Portadin (1977) reported that, in 50% of the adult patients they studied, pain episodes were preceded by stressful and depressing life events. These stressful events were identified by interview while patients were hospitalized for painful crises. Leavelt and Ford (1983),

188

PORTERET AL

using the Schedule for Recent Events tile change list, found that patienL~ with higher scores tended to be more symptomatic. Although suggestive, both of these studies are limited by their small sample sizes (22 and 16 patients, respectively) and their use of retrospective, global measures of both stress and pain. Thus, a seeond goal of this study was to examine the association between stress and SCD pain, using a well-established empirical measure of stress, the Daily Hassles Scale, along with prospective measures of daily pain and functioning. Given the unpredictable nature of pain episodes, the identification of factors related to pain onset may be helpfuI to both sickle cell patients and health care providers in their attempts to manage SCD pain.

METHOD Participants Participants were recruited from the Duke University-University of North Carolina Comprehensive Sickle Cell Center adult outpatient clinics. Patients from the center w e r e being recruited into a large-scale longitudinal study of coping with SCD pain (Gi] et al., 1989, 1992). A total of 119 adults completed the baseline evaluation. Of the 119 participants, 53 (44.5%) completed daily pain records for the 2-week period after the baseline assessment; they are the participants in this study. This sample consisted of 35 women and 18 men with a mean age of 32.8 years ( S D = 9.2; range = 18 to 58 years). All participants were African American. The mean educational level was 13.4 years ( S D = 2.03 years: range = 10th grade to graduate degree); 41,5% were married, 9.4% were divorced, 47.2% were single, and 1.9% were widowed. 37.7% were working full- or part-time, 11.3% were students, and 50.9% were disabled or unemployed. 84.9% had sickle cell anemia (SS), 7.5% had sickle thalassemia syndrome, and 7.5% had hemoglobin SCD. The participants who completed the daily pain records were compared to those from the larger study whodid not return the pain records on demographic variables (i.e., age, years of education, gender, employment status, and marital status) and disease-related variables (i.e., pbenotype of SCD, presence of an SCD-related condition, frequency/severity index of painful episodes). No significant differences were found, with the exception of a gender difference indicating that more women than men completed and returned the daily pain records (X2 = 4.0; p < .05).

Procedure Informed consent was obtained from the participants according to the procedures approved by the institutional review boards at the University of North Carolina at

PAIN IN SICKLECELLDISEASE

1 89

Chapel Hill and Duke University Medical Center. Participants completed several questionnaires, including a measure of"daily hassles," as part of the larger baseline study (Gil et al., 1989. Thompson, Gil, Abrams, & Phillips, 1992). After the baseline evaluation, participants were given the daily pain record along with instructions in how to complete it. They were asked to mail the complete pain record back to the investigators at the end of the 2-week period. A stamped, addressed envelope was provided. Participants were not prompted in any way to return the pain record, and they did not receive any compensation for doing so.

Measures Daily pain record. Aself-monitodngpalnrccord(Gil, 1994)wasusedtoassess daily pain level, medication usage, activity reduction, and health care use fora 2-week period. Patients were asked to complete the record every day of the 2-week period.

Pain rating. Patients were asked to rate their pain for each day on a scale from 0 (no pain) to |0 (pain as bad as it can be). Pain was summarized in two ways. Pain occurrence was defined as the number of days in the 14-day period that a participant reported any pain (e.g., a pain intensity rating greater that 0). Pain intensity was the average of a participant's pain ratings on those days that he or she experienced pain. Medication use, Medication use was measured by asking participants to indicate whether they had taken analgesics or narcotics each day, and to indicate the type and amount of medication taken. This information was summarized as number of days out of 14 days on which a participant reported taking any medication. We also calculated the number of days on which analgesics or narcotics were taken. Activity reduction. Patients were asked to indicate whether they cut back on work~school activities (for those participants who were either employed or students), social activities, and housework because of pain on each day. Activity reduction was determined for each of these categories of activity by summarizing the number of days that participants cut back on social activities and housework and, for those participants who were students or employed, work/school activities.

Health care use. Patients were asked to indicate whether or not they went to the emergency room, were admitted to the hospital, telephoned their physician (or other health care professional), or visited their physician each day because of pain.

1 go

PORTERET A~,.

This information was summarized as the number of days on which a participant utilized each of these health care services.

Stress. Stress was measured by the Daily Hassles Scale (Kanner, Coyne, Schaefer, &Lazarus, 198 I), a 117-item questionnaire that measures daily problems and difficulties, including financial concerns, physical problems, work issues, and domestic difficulties. Even though it is referred to as a measure of"daily" stress, this summary scale has respondents think back over the past month and indicate how often each hassle occurred and rate their severity on a scale of I (somewhatset,ere) to 3 (extremely severe). Several studies have supported the reliability and validity of the Hassles Scale (DeLongis, Coyne, Dakof, Folkrnan, & Lazarus, 1982: Holahan, Holahan, & Belk, 1984; Holm, Holroyd, Hursey, & Penzier. 1986; Kanner et el., 1981; Lazarus & DeLongis, 1983; Weinberger, Hiner, & Tiemey, 1987; Zarski, 1984). There are three summary scores that can be derived from the Hassles Scale (Kanner et al., 1981 ): (l)frequency, which is the simple count of the number o fitems checked (range = 1-117); (2) cumulated severity, the sum of the 3-point severity ratings (range = 0-351); and (3) intensio,, the cumulated severity divided by the frequency, a measure of how strongly or intensely the average hassle was experienced.

Disease severity. Disease severity was measured by phenotype and number of acute and chronic complications. All participants were diagnosed as having SCD by standard laboratory methods, including hemoglobin electrophoresis. Patients were categorized into three phenotypes: sickle cell anemia (Hb SS), hemoglobin SC disease (Hb SC), and sickle beta thalassemia syndromes. Although there is considerable variability across the phenotypes, sickle cell anemia is usually the more clinically severe type (Rucknagel, 1974). The physical course of SCD is often complicated by acute and chronic events, such as seizures and strokes (Charache, Lubin, & Reid, 1994). The number of these events was recorded at the baseline assessment. RESULTS Compliance

The 53 participants completed a total of 735 daily pain records. One participant missed 2 of the 14 days, and a second participant missed 5 days; this represents a compliance rate of 99%. The two participants with missing data were omitted from analyses requiring complete data. In addition, 2 other participants did not provide complete data on the activity variables and were omitted from those analyses.

PAIN IN SICKLECELLDISEASE

191

Overview of Analyses From the daily pain record, we examined each of the four domains of variables: pain, medication use, health care use, and activity reduction. Three sets of analyses were conducted for each domain. First, we calculated descriptive statistics. For medication, health care use, and activity reduction, we determined overall patterns--what the participants reported across the entire 2-week study period--as well as medication use, health care use, and activity reduction on those days on which participants reported pain. In the second set of analyses, we examined the associations between demographic variables and disease severity and pain, medication, health care use, and activity reduction. In the third set of analyses, we examined the extent to which pain intensity predicted patterns of medication use, health care use, and activity reduction, using pooled regression analyses that allowed us to examine within-subject relations utilizing participants' multiple daily reports. Finally, pooled regression analyses were used to examine the relation of daily stress to pain, medication use, health care use, and activity reduction.

Pain During the 14-day period, participants reported experiencing pain on an average of 6.5 days (SD = 5.3). Ten participants (19% of the sample) reported pain every day, and 8 participants (I5% of the sample) reported no pain on any day during the 2-week period. The average pain intensity rating during a painful episode was 4.4 (SD = 2.2) on the 0- to 10-point scale.

Pain, demographics, and disease severity. Chi-square and general linear regression analyses were performed to investigate the relations between demographic variables and pain occurrence and intensity. Predictor variables included gender, age, education, employment status, and marital status. There was one significant result: Marital status predicted average pain intensity [F(3,48) = 5.45, p < .01 ]. A Tukey Studentized Range test indicated that married and single participants had significantly lower pain intensity levels (p < .05) than those who were divorced or widowed. General linear regression analyses were also performed predicting pain occurrence and intensity from the two measures of disease severity--type of sickle cell disease and the number of SCD-related complications. None of these results were significant. Medication Use On average, participants reported taking medication (either analgesics, narcotics, or both) on 5.8 days of the 14-day reporting period (SD = 5.4). In analyzing the

192

PORTBRET AL

types of medications used, participants averaged 2.1 days of analgesic use (SD = 3.8), and 4.4 days of narcotics use (SD = 5.1). Eleven participants (21%) used no medication over the course oftbe reporting period; 32 (60%) did not use analgesics at all and 22 (42%) did not use narcotics. A small proportion of participants reported medication use every day of the study (any medication: 15%; analgesics: 4%; narcotics: 9%). On average, participants took medication on g0% of the days on which they reported pain (SD = 34%). They took analgesics an average 34% of tim days they experienced pain (SD = 42%), and they took narcotics 54% of the time (SD = 44%). Four participants (9%) did not use medication at all for pain; 24 (53%) did not use analgesics when in pain, and 15 (33%) did not use narcotics, Twenty-eight participants (62%) reported medication use on every day that they experienced pain; 9 (20%) used analgesics every day they had pain; and 16 (36%) always used narcotics.

Medication use, demographics, and disease severity. Chi-square and general linear regression analyses were performed to investigate the relations between demographic variables and medication use. Predictor variables included gender, age, education, employment status, and marital status. One result approached significance: Employment predicted the use of medication [F(4,39) = 2.22, p --- .08)]. General linear regression analyses were performed predicting medication use from the two measures of disease severity--type of sickle cell disease and the number of SCD-relatcd complications. None of these results were significant.

Patterns ofpain and medication use. In order to investigate the relation between pain and medication use, we first performed within-subject pooled regression analyses predicting the use of medi cation from the concurrent day' s pain i ntensity rating. This is a form of least squares regression that has demonstrated utility in the analysis of multilevel data (Kecfe et al., 1997). Two levels exist in this data: between-subjects and within-days associations. To remove the potentially confounding effects of differential reporting level by participants, all between-subjects variance is removed from both independent and dependent variables prior to exploration of within-person associations (Schwartz & Stone, 1993; Stone, Kennedy-Moore, & Neale, 1995). This type of ipsative analysis addresses the question of whether medication use is associated with pain intensity across an individual's reporting. Pooled regression analyses can be done using two types of models. Fixed effects models specify differences among individuals in separate intercept terms that arc considered fixed parameters. Random effects models allow for differences among individuals by including an additional error component in the model (Schwartz & Stone, 1998; Ward & Leigh, 1993). The choice between fixed and

PAIN IN SICKLE CELL DISEASE

"193

random effects models is based upon both conceptual and statistical cOnsiderations. Random effects models are generally preferred because their results are more readily generalized from the study sample to the population of interest and because the estimated coefficients usually have smaller standard errors. However, the random effects approach makes the strong assumption that the intercepts are uncorrelated with all the measured variables in the model. A statistical test developed by Hausman can be used to help determine whether this assumption is valid, and thus whether the fixed or random effects model is more appropriate (Ward & Leigh, 1993). We chose to use a fixed effects model in our analyses of the relations between pain, medication use, health care use, and activity reduction, after performing a Hausman test. The Hausman test was significant for a number of the models in question, indicating that the results of the fixed effects model were preferred (Ward & Leigh, 1993). 1 Results of these analyses indicated that pain intensity was a significant predictor of the use of medication (see Table 1). As expected, participants were more likely to take medication on days on which they had higher levels of pain. When considering type of medication (narcotics versus analgesics), the results indicated that pain intensity was a significant predictor of narcotic use. Again, as expected, participants were more likely to take narcotics on days on which they had higher levels of pain. Pain intensity did not predict analgesic use. We next examined participants' pain intensity ratings on days they took medication and compared them to days on which no medication was taken. Mean pain intensity ratings on days on which participants took analgesics was 3.9 (SD = 2.5). On days they took narcotics, the average pain rating was 5.7 (SD = 2.2). The average pain rating on days on which no medication was taken was 3.0 (SD = 1.9). To test whether people were more likely to take medication when their pain level was higher, we performed a repeated measures analysis of variance comparing the three mean pain intensity ratings: days with narcotic use, days with analgesic use, and days with no medication use. The result was significant [F(3,68) = 31.1, p < .0001]. A Tukey Studentized Range test indicated that all four means were significantly different from each other (p's<.05) (see Table 2). Thus, it appears that participants were more likely to use medication, particularly narcotics, on days that they experienced higher pain intensity levels. Health Care Use

During the 14-day study period, 40% (n = 21) of the participants had health care contacts: 24% (n = 13) went to the emergency room; 13 % (n = 7) were h ospitaliTcd: TFora subset of the analyses, the Hausman test was not significant, indicating that the random cffect~ model was appropriate. For thr analyses, ~sulls from the random effects models were not meaningfully different from tho~ of the fixed effects model.

194

PORTER E " AL

TABLE 1 Summary of Pooled Regression Analyses Predicting Medication Use, Health Care Use, and Activity Reduction From Pain Intensity Variable Medication Any medication Narcotics Analgesics Health Care Use Emergency room Hospitalizations Physician visits Phone calls Activity Reduclion Housework Social activities Work

Total R:

Unique R:

/' rant)

~/I-

~ c,efficient

.70 .71 .61

.04 .04 .OO

2g.0* 42.3 = 0.61

1,337 1,337 1,337

.04 .05 -.01

.37 .65 .27 .42

.14 .04 .04 ,0g

67. I " 32.4" 15.4" 38.7*

1,337 1,337 1,335 1,337

.05 .02 .02 .04

.64 .68 .33

.10 .12 .I 0

75.2* I 11.7" 20.9"

1,320 1,324 1,158

.08 .(19 .05

Note. Unique R2 n:fets to the amount of variane~ attributed to the variable of intere~. ToTal RZrefers to the amount of variance attributed to the variable of interest plus the amount of variance attributed Io individual differences. *p < .13001.

TABLE 2

Average Pain Intensity Levels

Days with narcotic mR Days with analgesic use Days with no medication Note.

N

Mean I)mn Intensity

SI)

30 21 17

57 39 ~0

22 2.5 I9

All means differed significantly from each other (/~ < .05)

21% (n = 1 I ) visited tbei r phy sician; and 24 % (n = 13 ) made a telephone call to a health care professional. When we examined only those days on which participants experienced pain. health care use was infrequent. For the 45 participants who experienced pain during the study period, on average they went to the emergency room on 6% of the days they reported pain. They were in the hospital on 5% of the days with pain, they visited their physician 5% of those days, and they called their physician or other health care professional on 8% of the days they had pain.

Health care use, demographics, and disease severity.

Chi-square and

general linear regression analyses were performed to investigate the relations between demographic variables and health care use. Predictor variables incIuded

PAIN IN SICKLECELLDISEASE

1 {}5

gender, age, education, employment status, and marital status. There was one significant result: Age predicted the frequency with which participants made telephone calls to their health care providers when in pain [F(1,43) = 5.84, p < .05], with older participants making more telephonecalls. General linear regression analyses were performed predicting health care use from the two measures of disease severity--type of sickle eel! disease and the number of SCD-related complications. None of these results were significant.

Patterns of pain and health care use. Pattemsofhealthcareuseandpain levels were analyzed, using within-subject pooled regression analyses. Pain intensity was a significant predictor of emergency room visits, hospitalization, physician visits, and phone calls (see Table 1). As expected, higher pain ratings were associated with greater health care use.

Activity Reduction We examined the rates at which participants cut back from their usual housework, social activities, and work activities during the 2-week study period. This was accomplished in two ways. First, we looked at overall patterns of activity reduction across the study period. Secondly, we focused exclusively on activity reduction on those days that participants reported pain. The latter analyses included only the 45 participants who experienced pain during the study. We found variations in reductions across activities, with participants most likely to cut back on housework, then social activities, and only infrequently on work. For the overall pattern, participants reported cutting back from housework on an average of 3.2 days out of the M-day study period (SD = 4.2); 20 participants (38% of the sample) did not cut back at all, and 3 participants (6%) cut back every day. Regarding social activities, participants cut back on an average of 2.8 of the 14 days (SD = 3.8); 24 participants (46% of the sample.) did not cut back at all; and 2 participants (4%) cut back every day. Of the 26 participants who were either students or employed, on average they cut hack from work or school half of one waking day (SD = 1.2 days) during the 14-day study period. Twenty participants (77%) did not cut back from work at all, and none of the participants cut back every day. Reduction in housework and social activities were highly correlated ( r = .83, p < .0001). Reduction in work showed much lower correlations with reductions in both housework ( r = .25)and social activities (r---.22), although these latter correlations (done within the subgroup of 26 participants who were students or employed) were also significant (p < .000l). Thus, when participants cut back from either housework or social activities, they were likely to cut back in both. However, those participants who were students or employed did not necessarily cut back from work activities, even if they did reduce their participation in housework or social activities.

196

PORTERET AL

On days on which they reported pain, participants cut back from housework 45%, or almost half of the time. Twelve participants (27% of the sample) did not cut back from housework at all while in pain, and ten participants (23%) cut back every day that they experienced pain. With regard to social activities, on average, participants cut back 41% of the time that they were experiencing pain. Sixteen participants (36%) did not cut back from social activities at all while in pain, and nine participants (20%) cut back every day that they experienced pain. Of the 2t participants who were employed or students and experienced pain, on average they cut back from work 5% of the days they had pain. Sixteen participants (76%) did not cut back from work at all while in pain. None of the participants cut back every day that they experienced pain.

Activity reduction, demographics, and disease severity, Chi-square and general linear regression analyses were performed to investigate the relations between demographic variables and activity reduction. Predictor variables included gender, age, education, employment status, and marital status. There were three significant results: Employment status predicted reduction in social activities [F(4.42) = 3.13, p < .05] and reduction in housework [F(4,42) = 4.43, p < .01l. Tukey Studentized Range tests indicated that participants who were disabled were more likely to cut back from social activities and housework than those who were employed (p's < .05). Education also predicted reduction in housework IF(1,42) = 4.23, p < .05]; participants with less education were more likely to cut back on housework when in pain. General linear regression analyses were performed predicting activity reduction from the two measures of disease severity--type of sickle cell disease and the number of SCD-related complications. None of these results were significant. Patterns of activity reduction and pain. Within-subject pooled regression analyses were conducted predicting activity reduction from the concurrent day's pain intensity rating. Pain intensity was a significant predictor of reductions in housework, social activities, and work, with participants more likely to cut back from their usual activities on days that they experienced higher levels of pain (see Table 1). We also examined average pain intensity ratings on days that participants cut back from their usual activities. Compared to the overall pain intensity level of 4.4, the average pain intensity rating on days that participants reduced their activities was 5.5 (housework: 5.37, SD = 2.26; social activities: 5.78, SD = 2.61; work: 5.23, SD = 3.16).

PAIN IN SICKLECELL DISEASE

197

The Relation of Daily Stress to Pain, Medication Use, Health Care Use, and Activity Reduction Pooled regression analyses were performed to investigate the associations between stress, as measured by the Hassles Scale, and pain levels, medication use, health care use, and activity reduction. Because stress was measured with a single summary scale, for these analyses we used a random effeets model, which is the appropriate model for investigating the effects of between-subjects factors (e.g., variables that do not change over time). The three Hassles Scale scores for f r e q u e n c y of hassles, the stressfulness of those hassles (intensity), and a combination of frequency and stressfulneas (cumulated severity) were used. Stress intensity ratings were a significant predictor of pain intensity levels (B = 1.1, p=.05). Stress intensity also predicted reductions in housework (B = 7.1, p < .05) and social activities (B = 7.1, p < .05). The relation between stress intensity and reductions in work approached significance (B = 2.7, p = .06) among those participants who were students or employed. The c u m u l a t e d severity scale was also a significant predictor of reductions [n housework (B = 0.1, p < .01 ), social activities (B = 0. l, p < .05), and work (B = 0.4, p < ,05). There were no significant associations found with the freq u e n c y scale; associations between stress and medication use or health Care use were also not significant (see Table 3). Because of the significant associations between stress intensity, cumulated severity, and activity reduction, we conducted follow-up analyses to determine whether stress was predictive of activity reduction after controlling for pain intensity. Additional pooled time series analyses were performed, entering pain level followed by the stress scales. The cumulated severity scale rumained a significant predictor of reductions in housework (B = .11, p < .05) and social activities (B = .95, p < .05) after controlling for pain. The relation between cumulated severity

TABLE 3 Summan/of Pooled Ragression Analyses Pmdic'dng Pain and Activity Reduction From Stress Hassle Intensity

Hassle Frequency

Variable

B

Standard Error

Pain Activily Reduction

.16

.17

z

B

.94

I,I

Standard Error .56

Cumulated Severi~

z

B

Standard Error

1.9"

.77

81

z .94

Housework

.17

.I0

1.7

"7.1

3.5

2.0*

.12

.48

2.5" *

Social activities Work/school

.15 .59

.I 1 .44

1,4 1.3

7.1 27

3.5 1.5

2.0" .I l 1.81" .41

.4q

2.3*

.21

2.0*

*p < .05: **p < 01. tp < .06.

198

PORTEREl" AL.

and reductions in work was no longer significant, and relations between stress intensity and reductions in activity were no longer significant.

DISCUSSION

From this prospective daily record study, we determined that adults with sickle cell disease reported frequent pain, with pain occurring on average on about half of the 14 study days. This pain was often low-level pain that was managed at home, sometimes without any medications. As pain levels increased, individuals tended to use medication, particularly narcotics, to manage pain. At higher pain levels, some patients also utilized a range of health care services, although overall utilization tended to be infrequent. On average, patients also cut back considerably on household and social activities, especially when pain reached a level of over 5 on the 10-point scale. Those who worked, however, were likely to continue to work, even when in pain. In contrast to previous studies of pain in adults with SCD, the results of this study indicate that the daily course of sickle cell disease may be characterized as an ongoing background of low-level pain with periods of marked increases in pain. Previous diary studies (Dinges et al., 1997; Fuggle et al., 1996; Gill et al., 1997; Shapiro et al., 1990, ! 995) reported somewhat lower rates of pain, ranging from 25 to 36% of the days. However, participants in these studies were predominately children, who tend to have less frequent pain than adults (Dinges et al., 1997). Other studies of adults with SCD (Gil et al., 1989, 1992; Plattet al., 1991) have described relatively infrequent and severe painful episodes. The discrepancy between our findings and those of previous studies is likely due to the daily, prospective design utilized in this study, as well as the absence of any predefined criteria for pain reports. As the definition of pain be.comes more inclusive, and as assessment methods become more intensive, more frequent and less severe episodes of SCD pain are captured. One possible explanation for this is that patients may be more likely to remember low-intensity pain, as well as pain that is managed at home, with shorter recall periods. Given the memory biases that have been demonstrated in the recall of pain and medication use (Eich et al., 1985; Linton & Melin, 1982; Ready et a]., 1982; Smith & Safer, 1993), daily assessment of these variables may provide more accurate information regarding patients' symptomatology and functioning. This type of information is critical in determining patients who are in need of intervention, as well as in the evaluation of treatment outcomes (Shapiro et al., 1990). In addition to reducing recall biases, daily pain records are a simple and cost-effective means of collecting prospective data. They may be especially useful in monitoring symptoms (such as pain) that vary frequently and arc difficult to

PA1N IN SICKLECELLDISEASE

199

summarize over long periods (Stone & Shiffman, 1992). Despite these advantages, daily recording methods are not without their drawbacks. One limitation is in determining adherence to the protocol. Although participants were instructed to complete the pain record on a daily basis, we cannot be certain that they foIlowed these insa'uctions or fi{[ed in the entire diary at one time before return. These types of problems can be addressed through more sophisticated data colIection techniques, such as the u ~ of palm-top computers (see, e.g., Stone et al., 199g), although these can be prohibitively expensive to implement. Another finding thai deserves mention is that regarding disease severity. Although health care providers generally assume that patients with the most severe disease experience the most frequent and intense pain and are the highest health care users, our data do not support this assumption. Disease severity, as measured by phenotype and number of acute and chronic complications, was not related to any of the pain, medication, or health care use variables. These findings are similar to prior studies of patients with SCD (Gil et at., I989, 1992). Demographics were also only weakly linked to pain and pain response, with noconsistcnt patterns emerging. Thus, it is especially important to identify other individual variables that are predictors of poor outcome, such as stress and coping (Gil et al., 1989,1992). The second major finding of this study concerns the relation between stress and pain and pain responses. From our analyses, we determined that stress had significant positive associations with average pain intensity as well as reductions in household, work, and socia} activities. Furlhermore, stress predicted reductions in housework and social activities even after controlling for pain intensity. These findings are important in that they are, to our knowledge, the f-L"Stto document the relation between stress and sickle cell disease pain and functioning using a well-established measure of stress along with a prospective measure of pain and pain response. Somewhat surprisingly, stress was unrelated to medication and health care use in this study. This finding should, however, be interpreted with caution. Health care use on days with pain was infrequent in this study. Moreover, a retrospective global self-report questionnaire was used to assess stress level. Future studies should include a longer monitoring period as well as daily recordings of personally relevant stressors that have been associated with increased chronic pain in other patient populations (Affleck el at., 1994; Stone et at., 1997, Urrows et al., ! 994). In addition, future studies should consider the potential mediating effect of mood on the stress-pain relation. Taken together, our findings with adults parallel those that have been reported for children with SCD (Fuggle e( aL, 1996~ Gill et aL, i 997; Shapiro eta]., 1990). Many painful episodes am managed at home and lead to major reductions in daily activities. However, the adult patients in this study appeared to be selective in their response to pain. They tended to cut back household and social activities but not work, and they abstained from narcotic medication on many days with pain. in addition, both their pain and their functioning were affected by the a m o u n t of stress

200

PORTERET~,L

they had experienced in the previous month, with more stress leading to both greater pain and greater functional impairment. There are several limitations of this study that need to be mentioned. First, as already noted, the recording period was only 2 weeks. A longer time frame may be needed to capture more accurate patterns of painful episodes. This may be especially true for health care utilization, Second, less than half of the eligible participants returned the pain record. Although our analyses of demographic and disease severity variables indicated that those who returned the pain record did not differ from those who did not (with the exception that women were more likely to return the pain record than men), there may be other important differences between the groups that were not detected. Thus, our findings might not generalize to the larger population of adults with SCD, particularly men. Future studies using daily monitoring should include strategies such as telephone reminders (Gill et al., 1997; Shapiro et al., 1990"} to enhance compliance with daily record keeping. Finally, in analyzing the relations between pain and pain response, we used a fixed effects model, which may also limit the ability to generalize these results to a larger population. In conclusion, this daily study of adults, combined with the Shapiro (1990, 1995) and Gill (1997) studies of children, indicate that pain is a pervasive aspect of SCD, and that it is most often managed at home. Systems of measuring disease impact that only track episodes leading to a health care contact may miss important information on the course of pain in patients with this disease. This realization may necessitate a change in, or at least a standardization of, one major outcome variable of interest. Most studies of SCD pain measure "pain episodes," but they employ different criteria for defining an episode (e.g., number of hours of pain necessary, or whether it resulted in a health care contact). A less ambiguous outcome measure might be pain days. Number of days with pain can be assessed easily and accurately by use of a daily recording system. Diary studies can also be useful in evaluating the need for either psychological or medical intervention, as well as the efficacy of intervention. Charache and colleagues (1995), for example, effectively utilized daily pain diaries to evaluate the success of hydroxyurea. Moreover, because pain is so often managed at home, psychological or behavioral interventions that enhance home management skills may be particularly advantageous. Additional evidence of the stress-pain relation would point to the possible efficacy of intervention efforts directed at stress management. Future studies should include a larger number of patients followed over longer periods and possibly utilize more frequent recordings in order to examine within-day fluctuations in pain and pain response. Finally, further studies of the association between stress and SCD pain are needed. Daily monitoring of stressful events has shown great promise in studies examining the impact of stress on other pain problems (Affleck et all., 1994; Stone et al., 199"7: Urrows et al., 1994).

PAIN IN SICKLE CELL DISEASE

201

Future studies of stress and pain in patients with SCD may benefit from daily measures of stress and coping rather than relying on more global, retrospective measures.

ACKNOWLEDGMENTS This work was supported by Grant ROI HL46953--06, by Project VI.B.2 in the Duke University-University of North Carolina Sickle Cell Center Grant in P60HL2839-12, and by the University of North Carolina at Chapel Hill GCRC Grant RR00046. The authors acknowledge the contribution of the late George Phillips, who provided invaluable input into the design and execution of this research as well as an endless devotion to the care of patients with sickle cell disease. We also express appreciation to all of the staff of the Duke University-University of North Carolina Sickle Cell Center including Mary Abrams, Adrena Johnson-Telfair, Susan Jones, Thomas Kinney, Eugene Orringer, Daria Peace, Dell Strayhorn, Wendy O'Kelley, Patricia Thomas, Bill Schultz, and Wendell Rosse. We thank the UNC Clinical Research Center nursing staff. We also thank Kelly Kosobucki for help with data management and manuscript preparation.

REFERENCES Affleck,G., Tcnncn, H., Urrows, S,,& HJggins,P. (1994).Person and contextualfcaturc,~of dailystress reactivity:Individualdifferencesin mlalionsof undesirabledailycvenlswith mood disturb~ce and chronic pain intensity. Journal o/Personality & Social Psychology, 66(2), 329-340. Burma.n, M. E. (199.5). Hcahh diaries in nursing research and practice. Image--the Journalr~Nursing Scholarship. 27(2), 147-152. Charache, S., Lubin, B., & Reid, C. D. ( 1994}. Management and therapy of sickle celldi.~ease (NIH Publication No. 94--2117). Washington, DC: National Institute of Health. Charm:he, S., Tcmn, M. L., Moore, R. D., Dover, G. J., Barton, F. B.0 Eckert, $. V., McMahon, R. P., & Bonds, D. R. (1995). Effect o f hydroxyurca on the frequency of painful crises in sickle cell anemia. New England Journal of Medicine, 332(20), 1317-1322. DeLongis, A.. Coyae, I., Dakof, G., Folkman, S., & Lazarus, R. S (1982}. Relationship of daily hassles, uplifts, and major life events to health .status. Health Psychology,/(2), 119-136. Dingcs, D. F., Whitehouse, W. G., Ome, E. C., Bloom, P. B., Caflin, M. M., Bauer, N. K., Gillen, K. A., Shapiro, B. S., Ohene-Fmmpoog, K., Dampicr, C., & Ornc, M. T. (1997). Self-hypnosis training as an adjunctive weatment in the management of pain associated with sickle cell di.~asc. International Journal t~ Clinical and Experimental Hypnosis, 45(4), 417--4.'t0. Eich, E., Reeves, J. L.. Jaeger, B., & Gra.ff-Radford, S. B. (1985). Memory for pain: Relation between past and pre,~nt pain intensity. Pain. 23, 375-379~ Fuggle, P., Shand, P. A., Gill, L. L, & Davies, S. C. (1996). Pain, quality of life, and coping in sickle cell disease. ArcAives of Disease in Childhood. 75(3), 199-203. Gil, K. M. (1994). Behavioral assessment of sickle cell di.seasc pain. Jaurnal of Health and Social Policy. 5(3/4). 19--38.

202

PORTER El" AL

Gil. K. M.. Abrams, M R.. Phillips, G., & K~fc, F..I. (I989). Sickle cell di~ca.sr pain.: Relation of copin.g ~ralcgies to adju.~tmcat. Journal of Cnnsuhing & Clinical Psychology. 57, 725-73 I. Gil, K. M., Abram.`=.M R.. Phillips, G., & Williams, D. A. (1992). Sickle cell d i m pain: 2. Predicting health care use and activity level at 9-month follow-up. Journal r162 & Clinical Pxy.chol. ogy, 60(2), 267-273. Gill, L. J., Shand, P. A., Fugglr P., Dugan, B., & Davies, S. C. (I 997). Pain asses.~men( for children with sickle cell discr162 Improved validity of diary keeping versus interview ratings. British Journal oJ Health P.Wchnlogy, 2, 131 140. Holahan, C. K., Holahan, C. J., & Bclk, $. S. (1984). Adj u.~trncnt in aging: The roles of life stre.~s, ha.~sles, and sell-efficacy. Health P.rychology, 3(4). 315-328. Holm. ]. E.. Holroyd, K. A., Hursey, K. G., & Penzien.. D. B. ['1986). The role of stress in [en.`=ionheadaches. Headache, 26, 160-t67. tanner, A. D., Coyne, J. C., Schaefer, C., & Lazarus, R. S. ( 1981 ). Comparison of two modes of stre~s mcasureraent: Daily ha.c.sle~and uplifL'~versus major life evcnt.~. Joarnal nfBehavioml Medicine. 4(I), 1~39. Keefe, F. J., Affleck, G., Lefehvm, J. C., Start, K., Caldwell, D., & Ten.hen, H. (1997). Pain coping strategies and coping efficacy in rheumatoid arthritis: A daily procc`=~analysis. Pain, 69(] 2), 35-42. Lazarus, R. S., and DeLongis. A. (1983). Psychological stress and coping in aging. American Psychologist, 38(3), 245-2.~4. Leavell, S. R., & Ford, C. V. (1983). Psychopathology in patients with sickle cell di~a,sc. Pyychosomatics, 24(I), 2.t-37. I_,Jnton. S..l.. & Melin. L. (1982). The accuracy of remembering chronic pain. Pain. 13, 281-285. Midencr K.. & Elandcr, J. (1994). Sickle celldisease: A p~vchosociat approach. Oxford, UK: Raddiff Medical Pm~`= Muschet, G. R.. Miller. D., Clemc, ts, B.. Pail, G.. & Gatterman, D. L. (1996). Impact ofsumatriptan on workplace productivity, nonwork activities, and health-related quality of life among hospilal employees with migraine. Headache, 36(3), 137-143. Nadel, C,. & Ponadin, G. (1977). Sickle cell crises: Psychological factors a.ssoeiatedwith onset New Fork State Journal of Medicine, 77, 1075=1078. Nush, K. B. (Ed.). (1994). Ps)~'ho.r a,rpect.r of,tickle cell disea.re: Pa.r present, and furore directions of research. Binghamlon, NY: Haworth Press. Norman. G. R., McFarlane, A. H., Stteiner. D. L., & Nea]e, K. (1982). Health diaries: Strategies for compZiance and relation to other measu~s. Medical Care, 20(6). 623-629. Plan, O. $., Thorington, B. D., Btambilla, D. J., Milner, P. F.. Rosse. W. F., Vichinsky, E.. & Kinney. T. R. (1991). Pain in sickle cell disease. Rates and risk faclors. New England Journal qfMedicine, 325(I), I ]-16. Ready. L. B., SarkL~, F-, & Tamer, J. A. (1982). Sel f-reported versus actual use of medications in chronic pain patients. Pare, 12, 285-294. Rucknagcl, D. L. (19"/4). 31~ genetics of sickle cell anemia and related syndromes. Archive.r of Internal Medicine. 133, 595--606. Schwartz. J. E., & Stone, A. A. (199.1). Coping with daily work problcn~: Contributions of problem content, appraisals, and person factors. Special Issue: Coping with sims,,=at work. Work & Stress. 7(I), 47.-62. Schwartz. J. E., & Stone, A. A. (1998). Strategies fat analyzing ecological momentary assessment data. Health Psychology. 17(7). 6-16. Shapiro, B. S., Din.ges. D. F., Orne, E. C., Bauer, N., Rcilly, L. B., Whitehouse, W. G , Ohene-Frcmpong, K.. & Ome. M. T. (1995). Home management of sickle cell-related pain in children and adoIescents: Natural history and impact on school atlendancc. Pain. tfl(I). 139-144. Shapiro. B. S., Dinges, D. F., Ome, E. C., Ohenc-Fremong, K., & Ome, M. T. (1990). Recording of crisis pain in sickle cell disease. Advances in Pain Research Therapy, l.'i, 313-321.

PAIN IN SICKLE C E L L DISEASE

203

Smith, W. B., & Safer,M. A. (1993).Efr~.~ of presentpain levelon recallofr'hronicpain and medication use. Pain, 55(3). 355---361 Stoller. E. P.. Forslcr. L. E., & Portugal, S. (I 993). ,~l f-care responsesto symptoms by older people: A health diary study of illness behavior. Medical Care. 31(1), 24-42. Stone, A. A., Brodenck, J. E., Porter, L. S., & Ka=ll, A. T. (1997). The experience ofrheunmtoid arthritis pain and faligue: Examining momentary ~ports and correlates over one wo~k. Arthritis Care and Research, 10(3), 185-193. Stone, A. A., Kennedy-Moore, E., & Ncale, J. M. 0995). Association between daily coping and end-of-day mood. Health Psychology, 14. 341-349. Stone, A. A., Kessler, R. C., & Haythornthwaite, J. A. ( 1991 ). Measuring daily events and experiences: Decisions for the researcher. Journal of Perxonality, 59, 575-607. Stone, A. A., Schwa~z, J. S., Neale, J. M., Shiffman, S. 5., Marco, C. A., Hickcox, M., Paty, J., Porter, L. S., & Cruise, L. J. (1998). A comparison of coping asse.~sing by ecological momentary assessment and ~urospective recall. Journal of Personalim. and Psychology, 74(6). Stone, A. A., & 5hiffman, S. (1992). Refleclions on the intensive measurement of stress, coping, and mood, with an emphasis on daily nw,asums. P.w.chology & Health, 7(2), 115-129. Thompson, R. I. J., Gil. K. M., Abrams, M. R., & Phillips, G. 0992). 5tress, coping, and psychological adjustment of adults with sickle cell disease. JcJurnal of Consulting & Clinical P.vychology, 60f3L 433---440. Urrows, S.,Affleck,G., Tennen, H., & Higgins,P. (1994).Unique clinicaland psychologicalcormlate.~ of fibromyalgiatenderpointsand jointTendernessin rheumatoid arthritis.Arthritix & Rheumatism, 37, 1513-1520. Verbrugge. L. M. (1980). Health diaries.Medical Care, 18(I),73-95. Ward, M. M., & Leigh, J. P. (1993). Pooled time seriesregressionanalysisin longitudinalstudies. Journal of Clinical Epidemiology, 46(7), 645-659. Weinbcrger. M., Hiner, 5. L., & "Henley, W. M. (1987). In support of hassles as a mca.~um of stress in predicting health outcomes. Journal of Behm, ioral Medicine./0(! ), 19-31 White, K. L.. Williams, T. F., & Greenberg, B. G. (1961). The ecology of medical ca~. New England Journal of Medicine, 265, 885-g92. Zar~ki. J. J. (1984). Ha.~les and health: A replication. Health P.w.ehology, 3(3), 243-25 I.