Pain 123 (2006) 5363 www.elsevier.

com/locate/pain

Chronic pain couples: Perceived marital interactions and pain behaviours

Toby R. Newton-John
b

a,* ,

Amanda C de C Williams

a Innervate Pain Management, Hunter Specialist Medical Centre, Newcastle NSW 2292, Australia Sub-Department of Clinical Health Psychology, University College London, Gower St, London WC1E 6BT, United Kingdom

Received 31 October 2005; received in revised form 25 January 2006; accepted 6 February 2006

Abstract Patient adjustment to chronic pain is well known to be inuenced by the spouse and his or her response to patient expressions of pain. However, these responses do not occur in a vacuum, and the aim of the present study was to investigate patientspouse interactions in chronic pain in detail. Ninety-ve patientspouse dyads completed questionnaires relating to mood, marital satisfaction and communication, and 80 couples also took part in semi-structured interviews. Data were analysed using quantitative and qualitative methods. Results showed that spouses of chronic pain patients reported engaging in a far wider repertoire of responses to pain behaviours than has been recognised to date. New response categories of hostile-solicitous and observe only were identied. Patients generally interpreted solicitous responses less favourably than spouse responses which encouraged task persistence. Male spouses identied fewer pain-related situations than female spouses but were more likely to report responding solicitously to patient pain behaviours. Marital satisfaction was signicantly higher in patients who rated themselves as talking more frequently about their pain. Spouse perceived frequency of pain talk was not related to spouse marital satisfaction. There were no gender dierences in marital satisfaction. The results of this study challenge some of the assumptions that have been held regarding chronic pain patientspouse interactions. 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
Keywords: Chronic pain; Disability; Marital satisfaction; Solicitous; Spouse responses

1. Introduction The spouse of the chronic pain patient has an important inuence on patient coping and adjustment (Turk et al., 1985; Flor et al., 1987; Schwartz and Edhe, 2000), most commonly sampled from patient or spouse using the Spouse Response Scale from the Multidimensional Pain Inventory (MPI; Kerns et al., 1985; Kerns and Rosenberg, 1995). This identies three kinds of spouse behavioural response to patient expressions of
* Corresponding author. Tel.: +61 2 4969 5309; fax: +61 2 4940 0322. E-mail addresses: tobynj@innervate.com.au (T.R. Newton-John), amanda.williams@ucl.ac.uk (A. CdeC Williams).

pain behaviour: solicitous (e.g., Tries to get me to rest), punishing or negative (Gets angry with me) and distracting (Tries to get me involved in some activity). Operant behavioural theory identies solicitous behaviours as positively reinforcing, so maintaining pain behaviours, and negative responses as aversive, thereby extinguishing them (Fordyce, 1976). While the evidence generally supports this formulation (e.g., Romano et al., 1992, 1995; Fillingim et al., 2003), there are failures to conrm them (Lousberg et al., 1992; Schwartz et al., 1996) and counter-examples, where higher pain levels and greater activity interference were associated with more frequent punishing spouse responses (Burns et al., 1996; Papas et al., 2001). In addition, patient

0304-3959/$32.00 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2006.02.009

54

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363 Table 1 Patient and spouse demographic data Demographic Age (years) Mean (SD) Range Sex Male Female Ethnicity (%) White Black/Afrocaribbean Asian Other Pain site (%) Low back Upper limb Head/neck Other Pain duration (years) Mean (SD) Range N = 95. Patient 48.2 (10.4) 2568 38 57 93 2 4 1 64 17 12 7 9.7 (9.2) 150 Spouse 49.1 (11.5) 2871 57 38 94 2 4

gender is not acknowledged in operant formulations although it has emerged as a moderator of patient spouse interactions (Fillingim et al., 2003; Cano et al., 2004; Smith et al., 2004); for example, spouse solicitousness was associated with greater self-reported disability only for male patients. The cognitive behavioural marital literature oers a somewhat dierent perspective on dyadic interactions. Greater variability in spouse behaviour is recognised than in the operant pain literature (Hahlweg et al., 1984; Fincham, 1994, 1997), with spouse cognitions highlighted as determinants of the variability in spouse behaviour (Cheung, 1996). Couple communication is also considered dierently. From an operant pain perspective pain talk, or patient verbal reports of pain, is classied as pain behaviour and therefore targeted in treatment for extinction. In contrast, the marital/relationships literature conceptualises open disclosure of emotions as generally desirable in close relationships (Kennedy-Moore and Watson, 2001), so encouraging this is a treatment goal to promote intimacy (Baucom et al., 1998). Although marital communication in the context of chronic pain has been discussed to a limited extent (Rowat and Kna, 1985; Osborn and Smith, 1998), the relationship between patient pain talk and marital satisfaction for chronic pain couples has not yet been explored. The present study was designed to address several of these issues, including patient and spouse gender as variables. The rst aim was to determine whether chronic pain patients spouse responses to patient pain behaviours covered a wider repertoire than currently represented in the chronic pain literature. Second, the immediate emotional impact on the patient of the spouses response was investigated in order to test the desirable/aversive assumptions regarding solicitous and punitive behaviours. Third, the role of patient pain talk was investigated in relation to communication within the relationship and marital satisfaction for patient and spouse.
2. Methods

tance, 16% said the cost of travelling was prohibitive, 12% cited the spouses work commitments and 32% said they were not interested in participating in the study. The remaining 8% gave a mixture of other reasons. Eighty-nine percent of the nal sample was married, with a mean relationship duration of 20.9 years (SD = 12.8 years). There were no same-sex couples. 29/95 (30%) of the spouses also reported experiencing a chronic illness, including chronic pain. Of the 95 couples that completed the self-report measures, one refused to take part in the semi-structured interviews and 14 were not available due to time constraints, leaving a total of 80 couples interviewed. 2.2. Ethics The study protocol was approved by St. Thomas Hospital Research Ethics Committee on behalf of the Guys and St Thomas NHS Trust, London. 2.3. Measures

2.1. Study population Ninety-ve heterogeneous chronic pain patients and their spouses were recruited for the study (see Table 1). Subjects were obtained over a 12-month period from consecutive patient referrals to an inpatient, chronic pain management programme (Williams et al., 1996). Inclusion criteria for the study were as per the Williams et al. (1996) study, and in addition patients were required to be in a primary relationship and to have been cohabiting with that partner for at least 12 months. Of the 162 eligible couples approached to participate, 95 (59%) agreed to take part. Information regarding refusal to participate was requested from each declining couple and obtained from 25 (38%) couples. Eight percentage of this group cited childcare commitments, 24% cited travelling disPain intensity. Following the recommendations of Jensen et al. (1986), pain intensity was measured on an 11-point numerical rating scale (0 = no pain at all, 10 = pain as bad as it could be). Subjects were asked to rate their average pain intensity over the last week, having rst anchored their response by rating their present pain level. Beck Depression Inventory (BDI; Beck et al., 1979). This 21item self-report questionnaire is among the most widely used mood measures in the pain literature. Although its factor structure in chronic pain has been the subject of some debate (Williams and Richardson, 1993; Miles et al., 2001), there is substantial support for its psychometric properties in general (Beck et al., 1988). Given the overlap between somatic items of depression and common symptoms of chronic pain (sleep

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

55

disturbance, impaired concentration, lack of energy, etc.), Geisser et al. (1997) recommended a cut-o score of 21 to indicate depressive symptoms in patients with chronic pain. Dyadic Adjustment Scale (DAS; Spanier, 1976). The DAS is a marital satisfaction scale containing 32 items, the majority of which are scored on a Likert scale from 0 = always disagree to 5 = always agree. The total score range is 0151, with Eddy et al. (1991) recommending a cut-o score of 98 to dierentiate between marital satisfaction and marital discordance. Heyman et al. (1994) reported excellent internal consistency for the DAS (coecient a = 0.95), and a conrmatory factor analysis showed that 81% of the variance in scores could be attributed to a global marital satisfaction factor. Pain Self-ecacy Questionnaire (PSEQ; Asghari and Nicholas, 2001). The PSEQ is a 10-item self-report questionnaire assessing patient condence to carry out various activities despite pain. Adequate psychometric properties have been reported (Gibson and Strong, 1996; Asghari and Nicholas, 2001), and the scale has been shown to predict outcome from pain management interventions (Coughlan et al., 1995), with patients reporting low levels of pain self-ecacy signicantly more likely to drop out of pain treatment than those reporting higher condence levels. Communication. Although standardised measures of marital communication have been developed (e.g., Primary Communication Inventory, Navran, 1967; Communication Skills Test, Floyd and Markman, 1983), these were not selected for the present study as they provide global information about communication strengths and weaknesses rather than specic information about pain-related verbal interactions. Instead, patients and spouses were presented with three 0 10 numerical rating scales with the following three items: (i) How satised are you with the way you and your partner communicate in general (not about the pain problem)? (0 = very unsatised, 10 = very satised), (ii) How satised are you with the way you and your partner communicate about the pain problem in particular? (0 = very unsatised, 10 = very satised) and (iii) How often do you/does your partner talk about your/their pain problem? (0 = never, 10 = constantly). Inter-item reliabilities were strong for the items relating to satisfaction with communication (a = 0.79 for patients and a = 0.80 for spouses). For use as an independent variable, quartile splits were performed on the pain talk frequency item with scores up to the 25th percentile coded as infrequent pain talk and scores above the 75th percentile coded as frequent pain talk. 2.4. Procedure After receiving approval from the local Hospital Research Ethics Committee to carry out the project, subjects meeting the inclusion criteria were sent a letter outlining the study and requesting their participation. They were asked to bring their partner to the pre-admission appointment, at which the couples completed the questionnaire battery (in separate rooms to avoid response bias) and then took part separately in the individual semi-structured interviews lasing approximately thirty minutes per person. When both patient and partner had completed the questionnaires and semi-structured interviews, they were debriefed as to the nature of the study.

2.5. Semi-structured interviews In order to investigate the research questions stated above, semi-structured interviews were carried out separately with patients and their spouses. The spouses were always interviewed before their patient-partner, as their responses determined the items presented to the patient. Spouses were given a series of 14 written vignettes (see Appendix A) depicting a variety of everyday situations (e.g., socialising, watching television and doing household chores) in which a chronic pain patient and his or her spouse are present. Derived from clinical experience and observation, each vignette describes the patient as exhibiting pain behaviour of some kind (e.g., grimacing, refusing to continue with an activity because of pain and complaining about pain). In order to ensure that the material was personally relevant, the spouses were rst asked to identify which of the vignettes they had experienced themselves. These situations were then selected for analysis and the spouses asked, In this situation, how would you respond? What would you normally do? Their responses to each item were recorded verbatim on protocol sheets. Following this, the patients independently read the same 14 vignettes and identied those that they had personally experienced. For the purposes of consistency within the couple, only those vignettes previously selected by the spouse were included in further patient questioning. Having nominated those situations that were familiar to them, the patients were asked, In this situation, how would your spouse respond? What would he/she do? After giving their response, the patients were asked the supplementary question, When your spouse responds in that way, how does it make you feel? If a patient had diculty identifying their aective reaction to the spouses response, they were given one prompt: Does that kind of response make you feel good, bad or neutral? They were not asked any further questions. 2.6. Data analysis qualitative All data were transcribed verbatim onto recording sheets and then into a word processor for analysis. Content analysis (Krippendorf, 1980; Weber, 1990) was the method chosen to analyse the qualitative data. This form of qualitative analysis has been used widely in health services research and has been applied to chronic illness populations including vertigo suerers (Yardley and Beech, 1998) and spinal cord injury patients (King and Kennedy, 1999). Content analysis involves transforming a data set into groups or categories of homogeneous responses (Lee and Peterson, 1997). The aim is to code the data until saturation of the data set is reached; in other words, the point at which no further categories or groups can be generated. The existing set of categories are then reviewed for any duplication or signicant overlap and reduced where possible into the nal core set of categories. The entire data set is then analysed according to the categories obtained, and the frequency of occurrence of each category is counted. Content analysis stipulates that all categories be mutually exclusive (Krippendorf, 1980), such that a particular response may not be coded as more than one category. An integral part of the content analytic method is the examination of intercoder reliability, in order to ensure that the categories are objectively veriable.

56

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

Three doctoral-level clinical psychologists experienced in assessing and treating chronic pain patients were used as coders in the study. Training consisted of a half-day workshop in which a standardised set of instructions was presented to the three coders outlining the process of a content analysis and the nature of the present study. In the rst practical training section, the rst author explained the 12 coding categories and their rules, and gave a number of examples of spouse responses appropriate to each category. Then in an open discussion format, 20 example spouse responses were presented and a general discussion held as to why they would be coded in one category rather than another. Coders were encouraged to ask questions or seek clarication of the coding rules in order to improve their accuracy. The nal training task involved each coder privately coding 20 example spouse responses. The results of their coding were then examined, and 91% agreement across all three coders was obtained from this initial trial which was considered satisfactory prior to embarking on the actual data set.

3. Results 3.1. Quantitative data There were no signicant dierences between the spouses reporting chronic illness and those not reporting chronic illness on the variables of age or gender, nor were there any group dierences on mood or marital satisfaction scores. Table 2 shows the results of the questionnaire data completed by patients and their spouses. Overall, these data indicate that the patient group is similar to that described in the Williams et al. (1996) study in terms of pain levels, mood and self-ecacy, and could be therefore considered representative of other pain clinic samples. The mean BDI score for the patient sample was below the cut-o recommended by Geisser et al. (1997), indicating no signicant mood disturbance for the sample as a whole. Compared to norms for the general population given by Beck et al. (1988), the spouse sample fell in the mild range for depression. On the DAS, group means for the patients and their spouses were approaching one standard deviation above the

cut-o score for marital distress, indicating generally high levels of marital satisfaction in both groups. A series of two-way ANOVAs were carried out with group (patient vs spouse) and gender (male versus female) as the independent variables, and marital satisfaction, depression and the communication variables as the dependent variables. A main eect for depression was observed, with patients reporting signicantly higher depression scores than spouses [F (1, 182) = 85.0, p < 0.001], but no gender dierences within either patient or spouse groups. Marital satisfaction was not signicantly dierent for either group or gender. There were no main eects or interaction eects for the satisfaction with general communication variable; however, satisfaction with pain-related communication showed a signicant interaction eect, such that female spouses reported signicantly lower levels of satisfaction than male spouses [F (1, 186) = 4.17.0, p < 0.04]. Table 2 also shows that there was no signicant difference between patients and spouses in the perceived frequency of pain talk by the patient. However, a strong gender eect was observed, with males perceiving a greater frequency of pain talk than females [F (1, 184) = 8.08, p < 0.005]. There was also a highly signicant interaction eect (p < 0.002) which showed that male spouses reported signicantly more pain talk (from their female patient-partners) than their female counterparts or either patient group. The analysis of the communication data revealed agreement across all groups that being satised with general communication within the relationship is associated with better communication about pain (Table 3). For female patients only, talking more often about pain was associated with a more positive perception of overall communication. But when considering satisfaction with pain communications specically, both patient groups saw pain talk frequency as signicantly positively associated with satisfaction, whereas neither spouse group did. Two-way ANOVAs were then carried out on the patient and spouse data separately to explore the relationships between gender and pain talk on marital satis-

Table 2 Means and standard deviations (in parentheses) for pain, pain self-ecacy, mood, marital satisfaction and communication ratings for male and female patients and spouses Patient Male Pain intensity (0100 VAS) Pain Self-Ecacy Questionnaire Beck Depression Inventory Dyadic Adjustment Scale Satisfaction with communication in general (010 VAS) Satisfaction with communication about pain (010 VAS) Perceived frequency of pain talk (010 VAS) N = 95. 69.57 24.36 18.03 116.03 7.55 7.13 5.28 (16.65) (9.97) (7.99) (20.34) (2.45) (2.71) (2.74) Female 72.37 24.32 18.47 113.34 7.78 7.24 5.46 (17.19) (11.94) (8.15) (18.54) (2.69) (2.71) (2.34) Spouse Male 7.33 112.96 8.07 7.35 6.71 (4.89) (15.38) (1.81) (2.05) (2.30) Female 10.11 109.84 7.29 5.97 4.55

(6.37) (23.08) (2.70) (2.34) (2.47)

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363 Table 3 Bivariate correlations between marital communication variables Variables correlated Satisfaction with communication in general and satisfaction with communication about pain Satisfaction with communication in general and frequency of pain talk Satisfaction with communication about pain and frequency of pain talk
* **

57

Female spouse 0.72 0.11 0.31

**

Female patient 0.68

**

Male spouse 0.60 0.07 0.07

**

Male patient 0.62** 0.15 0.35*

0.33* 0.45**

p < 0.05. p < 0.01.

faction. A signicant main eect was observed for the analysis of patient DAS score and pain talk frequency. Patients who perceived that they talked more often about their pain reported signicantly higher marital satisfaction than patients who rated themselves as talking less often about their pain [F (1, 69) = 5.87, p < 0.01]. There was no dierence in marital satisfaction scores for spouses according to whether they perceived patients talking frequently or infrequently about their pain. There were no signicant gender dierences in any of the comparisons. 3.2. Qualitative data spouses

3.

4. The 80 spouses generated a total of 540 responses to the 14 vignettes (mean 6.6 responses per spouse, range 313 responses). Saturation was reached in the initial coding run after 36 categories had been generated. However, subsequent cross-comparison of the categories reduced this total to a nal set of 12 core responses. Intercoder reliability checks were carried out on a random sample of 120 responses (20% of the total). j coefcients ranged from j = 0.84 to j = 0.89, which indicates excellent intercoder reliability (Everitt, 1996). A brief description of each spouse response category follows, in descending order of frequency of reported use. The labels for each category are based upon the spouses intention rather than from an a priori judgement. 1. Provide Help (35% of total responses): this category is the most closely aligned with traditional notions of solicitousness. The behaviours represented here are specically aimed at either relieving pain (e.g., giving a massage, getting pain medications and taking over the patients activity) or decreasing pain-related distress (e.g., expressing sympathy or concern). 2. Observe Only (15% of total responses): this category represents an absence of overt behavioural responding by the spouses. They reported that after witnessing the patients pain behaviour, their reaction was simply to observe and monitor the situation rather than act immediately. A variety of dierent reasons were given for this response,

5.

6.

7.

8.

including You cant always be jumping up to help them, I like to see how he gets on himself and She doesnt like me fussing, but I want to know whats happening. Oer Help (11% of total responses): this response category reects an inquiry on the part of the spouse (e.g., Do you need any help? Is there anything I can do?). This is contrasted with the Provide Help category, in which the spouse simply performs the pain-relieving activity. The distinction was drawn because the Oer Help response allows the patient to refuse the inquiry and thereby maintain a self-management stance. Discourage Pain Talk (8% of total responses): on the occasions where the target pain behaviour was verbal, this category of response refers to the spouses attempts to inhibit discussion of the pain. This was usually by achieved by interruption or deliberately changing the topic of conversation away from pain. Encourage Pain Talk (7% of total responses): the opposite of the above category, here the spouse attempts to promote discussion of the patients pain. Encourage Task Persistence (6% of total responses): this category refers to the spouses attempt to encourage the patient to continue with an activity, despite acknowledging the pain behaviour. The Encourage Task Persistence category may relate to a specic activity such as socialising (e.g., We dont need to return home just yet, youll be OK in a short while) or domestic chores (e.g., You look like you can cope with that); or it may represent more general examples of encouragement (e.g., Come on, you can do it). Shield (6% of total responses): the spouse tries to avoid upsetting the patient by either protecting him or her from further stress (e.g., withholding bad news) or by terminating an argument or disagreement when pain behaviour becomes apparent. Express frustration (4% of total responses): this response is an equivalent to the MPI category of punishing/negative responses. Here the spouse

58

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

directly expresses anger or irritation with the patient as a result of pain behaviour (e.g., For goodness sake, stop complaining about your back!). 9. Ignore (4% of total responses): the spouse pretends that the pain behaviour had not occurred and intentionally disregards the patients display of discomfort. This is dierent to the Observe Only category, where the spouse doesnt try to ignore or disregard the patient but instead defers making any behavioural response at the time. 10. Problem-Solve (3% of total responses): this category reects the spouses specic attempt to help the patient maintain a given activity or achieve a particular goal despite the pain. For example, the spouse might suggest that the patient cooks the evening meal, but does it sitting down rather than standing. It diers from the Encourage Task Persistence category in that it contains specic suggestions for changing behaviour in order to achieve an outcome. 11. Hostile-Solicitous (2% of total responses): this response category encompasses the customary notion of solicitousness, insofar as the spouse is overtly trying to relieve pain or distress, but here the behaviour is accompanied by an aggressive

or irritated manner (e.g., Ill get the pain killers for her, but only when I am good and ready, or Ill shout at her and How many times have I told you not to carry the bags put them down right now). 12. Distraction (1%): very rarely did spouses report specically attempting to distract the patient from the pain (e.g., I start talking to her about anything I can think of). This category is the equivalent of the MPI subscale of distracting responses.

3.3. Qualitative data patients The patient sample generated a total of 353 responses to the 14 vignettes (mean 3.45 responses per patient, range 18 responses). The lower mean reects the inclusion in the analysis of only those vignettes that the spouse had also previously selected. A random sample of 70 patient responses (20% of the total) was also examined for intercoder reliability and again excellent coecients were obtained (j = 0.85 to j = 0.87). Not only does this indicate high levels of agreement between raters, it also demonstrates that as a group, patients reports of their spouses behaviour closely matched the reports of the spouses themselves.

50 45 40 35 30 % total responses 25 20 15 10 5 0
ov Pr id e H O bs ve er p el O ffe rH el p y nl O D is En ur co Pa in l Ta k e ag En ur co Sh ie Ex es pr no Ig ob Pr H t os D is ur co tra

Spouse report Patient report

re

ld

ile

le m

ct

Fig. 1. Patient and spouse reports of spouse responses. Graph shows the frequency of each spouse response as a percentage of the total responses.

e ag Pa

e ag s Ta k Pe k i rs in l Ta

lic So

u Fr st tio ra n st en ce

So lv e

ito us

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

59

3.4. Between-groups comparisons: patient and spouse ratings of spouse responses Fig. 1 shows the frequency counts for the various response categories given by both patients and spouses. Providing Help was the most common form of spouse response to patient pain behaviours, with Oering Help also being rated highly by both members. Overall, the congruence between patients and spouse groups was moderate to high, with Friedmans test failing to reveal any signicant dierence in rank orderings between the two groups (Fr = 15.40, p = .16). However, at an individual item level there were several discrepancies deserving comment. For example, spouses rated themselves as far less likely to respond with Express Frustration (ranked 8th) than the patients perception of this response, which was rated as the equal 3rd most common. Equally, the Hostile-Solicitous category was endorsed more frequently by patients than spouses. As might be expected, spouses rating of the Observe Only condition was more frequent than patients, who may not always be aware that this behaviour is occurring. Interestingly, the patients perception of Encouraging Pain Talk was greater than the spouses report of their behaviour here. There was a high level of agreement regarding the Distraction category however, with both viewing this as an uncommon response. 3.4.1. Gender dierences in spouse responses Although there were more similarities than dierences between male and female spouses in the ways in which they reported responding to patient pain behaviours, two important dierences were noted. First, the mean number of responses to the 14 vignettes by female spouses was 6.76, whereas for males it was only 4.96. The fewer responses given by male spouses indicate that they identied fewer pain-related interactions compared to female spouses. In terms of the categories of response, 41% of all responses by male spouses were Providing Help the traditionally solicitous type of response, whereas only 28% of the female spouse responses were coded this way. This suggests that male spouses are more likely to respond solicitously than female spouses. 3.4.2. Gender dierences in patient perceptions of spouse responses Very few dierences were noted between male and female patients in terms of their perceptions of spouse responses. For example, the Provide Help category accounted for 36% of all male patient perceived responses and 39% of female patient perceived responses. The only discrepancy of note was that female patients reported a slightly greater frequency of Oer Help responses than male patients (14% versus 9%), which

is consistent with the greater frequency of these responses reported by male spouses. 3.5. Patient aective responses The responses to the patient question, When your spouse responds in that way, how does it make you feel? were coded into three categories of aect: positive, negative or neutral. However, unlike the 12 response categories described above, these aective categories were not mutually exclusive. Thus patients could report feeling cared for in response to a Provide Help response, but also guilty for receiving the assistance. Equally, a number of patients described feeling upset in reaction to the Express Frustration response category, but also relieved as it meant that the spouse had noticed their discomfort. On the occasions such as this, the response was coded as having both a positive and a negative aective value. 3.5.1. Patient aective responses results The data presented in Fig. 2 show patients aective reactions to the spouses responses, coded as positive, negative and neutral. No response was uniformly rated positively or negatively, and most responses had some impact upon the patients emotional state, i.e., neutral ratings were relatively infrequent. In general, the most positively rated responses were those associated with active coping strategies, rather than with solicitous behaviours. The categories receiving the highest number of positive endorsements were those in which the spouse encouraged the patient to persist with an activity despite pain or where the spouse simply observed the patient without responding. The third most positively rated response for patients was the Problem Solve category, where the spouse works together with the patient to try to achieve a certain goal or activity despite the pain. In contrast, the Oer Help response was ranked 7th out of 12, and the Provide Help response traditionally associated with solicitousness was placed fourth last in the list of positive responses. These categories were also rated negatively on 36% of occasions. The negative aective reactions to these responses were typically reported to be feelings of guilt, uselessness and being a burden. The patients reports of the spouse responses they considered to be most positive contrasted strongly with the spouses reports of their own behaviours. While nearly half 46% of all spouse-rated responses involved oering help or giving it without asking, these responses were often perceived negatively by patients. Furthermore, although the Problem Solve response was highly rated by patients (3rd most positive), only 3% of all spouse responses were of this category. There were no clear gender dierences in the aective reactions to spouse responses.

60

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

100
Frequency of response rated positively (%) Frequency of response rated negatively (%) Frequency of response rated neutrally (%)

90 80 70 60 50 40 30 20 10 0
re no Ig lk p el Ta in rH ffe Pa O e ag ur co lk En Ta in ld ie Pa e Sh ag ur co ve is ol D S e m nc le te ob ly is n Pr rs O Pe ve sk er bs Ta O e ag ur co En n ct tio ra tra st is ru D F s us es pr ito lic Ex So e til p os el H H e id ov Pr

Fig. 2. Patient ratings of their aective reactions to spouse responses. Graph shows the percentages of positive, negative and neutral ratings for each spouse response.

4. Discussion There were three primary aims of the present study. First, to determine whether spouses of chronic pain patients demonstrate greater variability in their responses to pain behaviours than has previously been identied in the pain literature. Second, to assess the aective impact of these responses upon the patient in order to test some of the assumptions relating to solicitousness in chronic pain couples. And nally, to examine the association between marital satisfaction and the frequency with which patients talk about pain to their spouses. The relationship between the above factors and gender within the patientspouse dyads was also explored. In contrast to the prevailing view that solicitous, punishing or distracting behaviours are the main ways in which spouses respond to patient displays of pain, our content analysis revealed a much broader response repertoire. Twelve dierent categories of spouse response were identied which included the three existing response categories and two new categories, labelled Observe Only and Hostile-Solicitous, which had not previously been discussed in the pain literature. Intercoder reliability checks conrmed that the categories were highly reliable and were not generated according to idiosyncratic coding rules. It is apparent that, rather than drawing upon one of three mutually exclusive types of response, spouses of chronic pain patients behave in far more varied ways than has been represented by the MPI scales. Schwartz et al. (2005) have recently pub-

lished a 39-item Spouse Response Inventory (SRI) which distinguishes between spouse responses to pain behaviours and spouse responses to well behaviours. There are a number of areas of overlap between the SRI and the 12 response categories obtained here (e.g., both distinguish between oering help and initiating help, both have shielding and encourage task persistence responses, both refer to expressions of frustration by the spouse, etc.). However, as was acknowledged by Schwartz and colleagues, both the MPI and the SRI make a priori assumptions about the reinforcing or punishing qualities of a given patientspouse interaction. The results of the present study suggest that some of these assumptions may be inaccurate. The qualitative data obtained in the present study challenge the view that solicitous responses are always perceived positively by chronic pain patients. Patients rated spouse responses that are consistent with cognitive behavioural models of chronic pain self-management more favourably than those responses traditionally associated with the solicitousness construct (Turk, 2002). These data may help to explain some of the inconsistencies in this literature (e.g., Lousberg et al., 1992; Burns et al., 1996; Schwartz et al., 1996; Papas et al., 2001), where no relationship between patient disability and spouse solicitousness has been found. In these cases, although the spouses responses were classied as solicitous on an a priori basis, they may not have been perceived positively by the patients and therefore had no consistent pain behaviour-reinforcing qualities.

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

61

There was a relatively high level of agreement between patient and spouse groups regarding the ways in which spouses respond to pain behaviours. Between-groups comparisons were carried out in this study in order to maximise the generalisability of these initial ndings, but this result is consistent with research exploring within-couple congruence. Cano et al. (2004) showed that non-depressed patient couples are more congruent in their ratings of patient pain and physical disability levels than depressed patient couples, and patients in this sample were predominantly in the nondepressed range. An important question for future research would be to examine congruence in patient and spouse perceptions of these 12 response codes. However, there was a major discrepancy between the most frequently reported spouse responses and those rated most positively by patients. Both patients and spouses agreed that Encourage Task Persistence and Problem-Solve occurred rarely, yet patients rated these responses favourably. Conversely, the traditionally solicitous responses were by far the most commonly reported, yet they frequently engendered negative feelings in the patients. This result highlights the importance of including the spouse in pain management interventions (e.g., Keefe et al., 1996, 2004) in order to ensure that they are familiar with the principles of self-management and are not inadvertently undermining the patients eorts to develop better coping skills. Patient and spouse gender was also found to be an important factor in these dyadic interactions. Male spouses identied fewer pain-related situations from the study vignettes than did female spouses, but reported signicantly more pain talk than either female group. The person perception literature indicates that males are less accurate perceivers of disability than females (Ickes et al., 2000), and this nding may reect this. The male spouses in this study may have been less aware of the pain behaviours of their partners, and may have overestimated the frequency of discussions about pain, but without objective evidence of these behaviours it is not possible to determine this. However, male spouses were more likely to report responding solicitously than were female spouses. Rowat and Kna (1985) showed that spouses often report feelings of helplessness when their partners are in pain, and a limited response repertoire may be a factor in that helplessness. Our data indicated that male spouses knew of few alternatives other than responding solicitously, which underlines the importance of including the spouse in pain management training. The results concerning communication and marital satisfaction also have some bearing upon couples-based pain interventions. A positive association was found between the frequency of pain talk and marital satisfaction, which might be assumed to reect a general communication style in which the couple converse frequently on many topics of importance to them,

and that it is the latter characteristic which underpins their marital satisfaction. However, the association was only signicant for patients and not their spouses, so the relationship between the two variables is not a shared phenomenon. The nding suggests that there is something particularly supportive about being able to freely discuss ones pain problem with ones partner. The results also showed that partners did not necessarily nd a high frequency of pain-related communication aversive, insofar as a signicant negative correlation between frequency of pain talk and marital satisfaction for spouses was not observed. A limitation of the measure of pain talk used in this study was its generality it did not quantify the content of the pain talk, who initiated it, or the context in which it occurred. There is clearly a dierence between complaining about pain and discussing how it is being managed, what strategies are eective, what progress is being made, and so on. Nevertheless, these data suggest that rather than trying to extinguish all communications about pain, patients should be encouraged to discuss their pain management with their spouses as part of communication skills training (Baucom et al., 1998; Kennedy-Moore and Watson, 2001). The qualitative and quantitative data were not integrated in this study due to concerns about the non-normality of the qualitative data distributions. We felt that it was premature to carry out data transformation techniques and combine the data sets before the coding categories have been replicated in a subsequent study. However, further research to explore the relationships between these response categories and patient coping variables would be illuminating. Of particular interest is the relationship between high frequency Hostile-Solicitous responses and patient disability is the association consistent with traditional solicitous responses (i.e., high patient disability) or is there a predominantly punishing eect from this response which leads to reduced patient disability levels? The study has several acknowledged limitations. The sample was comprised of predominantly white, lower back pain suerers with lengthy pain histories and non-clinical levels of depression. Marital satisfaction was generally high, while the mean relationship duration was more than 20 years. Whether this set of results would be obtained with a less maritally satised, more distressed, more culturally diverse patient sample is open to question. Further research is also required to determine how spouse responses are inuenced by contextual factors such as the environment in which the pain behaviour occurs or the current state of the relationship. As noted by Cheung (1996), the interpretation of partner behaviour is inuenced by such variables, and where a given spouse response may be acceptable within the home, it may be viewed dierently by the patient in social situations outside the home. Finally, as the study

62

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363

was based on subjects recall of their pain-related interactions, it is possible that inaccurate recall or social desirability eects inuenced their responses. Future observational studies based on the coding system reported here would help to determine the degree of consistency between couples reports of their behaviour and their actual behaviour. In summary, it appears that while the operant behavioural model of chronic pain has advanced our understanding of patientspouse interactions enormously, it has also under-represented the complexity of these dyadic interactions. Spouses vary signicantly in how they respond to patient pain behaviours, although tend to rely on solicitous behaviours most frequently. Patients, on the other hand, perceived solicitous responses less favourably than the spouses eorts to support their self-management of pain. Furthermore, data presented here highlighted the importance of communication within the couple as a means of enhancing and maintaining marital harmony.

11. Your spouse is in the middle of a are-up of pain, when you learn of some disappointing news that concerns both of you. 12. You are talking with your spouse, and they begin telling you how bad their pain is at that moment. 13. You and your spouse are lying in bed at night. Your spouse is tossing and turning as they do when in pain, and it is keeping you awake. 14. Your spouse tells you that they are feeling very romantic and asks whether you would like to retire early this evening.

References
Asghari A, Nicholas MK. Pain self-ecacy beliefs and pain behaviour: a prospective study. Pain 2001;94:85100. Beck AT, Steer RA, Garbin MG. Psychometric properties of the Beck Depression Inventory: twenty-ve years of evaluation. Clin Psychol Rev 1988;8:77100. Beck AT, Rush AJ, Shaw BF, Emery G. Cognitive therapy of depression. New York: Guilford Press; 1979. Baucom DH, Shoham V, Mueser KT, Daiuto AD, Stickle TR. Empirically supported couple and family interventions for marital distress and adult mental health problems. J Consult Clin Psychol 1998;66:5388. Burns JW, Johnson BJ, Mahoney N, Devine J, Pawl R. Anger management style, hostility and spouse responses: gender dierences in predictors of adjustment among chronic pain patients. Pain 1996;64:44553. Cano A, Gillis M, Heinz W, Geisser M, Foran H. Marital functioning, chronic pain and psychological distress. Pain 2004;107:99106. Cano A, Johansen AB, Geisser M. Spousal congruence on disability, pain and spouse responses to pain. Pain 2004;109:25865. Cheung S. Cognitive behaviour therapy for marital conict: rening the concept of attribution. J Family Ther 1996;18:183203. Coughlan GM, Ridout KL, Williams A CdeC, Richardson PH. Attrition from a pain management programme. Br J Clin Psychol 1995;34:4719. Eddy JM, Heyman RE, Weiss RL. An empirical evaluation of the Dyadic Adjustment Scale: exploring the dierences between marital satisfaction and adjustment. Behav Assess 1991;13:199220. Everitt BS. The assessment of reliability. In: Everitt BS, editor. Making Sense of Statistics in Psychology: A Second Level Course. Oxford: Oxford University Press; 1996. Fillingim RB, Doleys DM, Edwards RR, Lowery D. Spousal responses are dierentially associated with clinical variables in women and men with chronic pain. Clin J Pain 2003;19:21724. Fincham FD. Cognition in marriage: current status and future challenges. Appl Preventative Psychol 1994;3:18598. Fincham FD. Understanding marriage: from sh scales to milliseconds. Psychologist 1997;10:5437. Flor H, Turk DC, Rudy T. Pain and families: II. Assessment and treatment. Pain 1987;30:2945. Floyd FJ, Markman HJ. Observational biases in spouse observation: toward a cognitive-behavioral model of marriage. J Consult Clin Psychol 1983;51:4507. Fordyce WE. Behavioral methods in chronic pain and illness. St. Louis: Mosby; 1976. Geisser ME, Roth RS, Robinson ME. Assessing depression among persons with chronic pain using the Center for Epidemiological Studies-Depression Scale and the Beck Depression Inventory: a comparative analysis. Clin J Pain 1997;13:16370.

Appendix A. Vignettes used as prompts for recall of spouse responses to pain behaviour 1. You are sitting quietly with your spouse when you notice that they are in pain. They then say I could do with a hot bath/hot water bottle. 2. You are getting shopping out of the car, and you see your spouse starting to lift a heavy bag of shopping. 3. You are with some friends or family, and the conversation turns to your spouses pain problem. 4. You are at a social gathering with your spouse. They are really enjoying themselves, but they tell you that they must have overdone it and now they are really hurting. They tell you they would like to go home. 5. You and your spouse are watching television. Your spouse is shifting about, trying to get comfortable, but it is distracting you from the program you are watching. 6. You are feeling aectionate towards your spouse. As you give them a cuddle, they suddenly wince and grab their painful area. 7. You and your spouse are having a disagreement about something, when you become aware that your spouses pain has increased. 8. You have some friends or family over to your place. During the visit, your spouse says that they are going to lie down because of the pain. 9. You and your spouse have accepted an invitation to go out, but on the day your spouse says that their pain is bad and they cant go. 10. It is your spouses turn to cook this evening. But in the morning before you leave to go out for the day, your spouse says I am having a bad day today. I wont be able to cook this evening.

T.R. Newton-John, A. CdeC Williams / Pain 123 (2006) 5363 Gibson L, Strong J. The reliability and validity of a measure of perceived functional capacity for work in chronic back pain. J Occup Rehabil 1996;6:15975. Hahlweg K, Reisner L, Kohli J, Volmmer M, Schindler L, Revenstort D. KPI category system for partnership interaction. In: Hahlweg K, Jacobson N, editors. Marital interaction; analysis and modication. New York: Guilford Press; 1984. Heyman RE, Sayers SL, Bellack AS. Global marital satisfaction versus marital adjustment: an empirical comparison of three measures. J Fam Psychol 1994;8:43246. Ickes W, Gesn PR, Graham T. Gender dierences in empathic accuracy: dierential ability or dierential motivation? Pers Relat 2000;7:95109. Jensen MP, Karoly P, Huger R. The measurement of clinical pain intensity: a comparison of six methods. Pain 1986;27:11726. Keefe FJ, Blumenthal J, Baucom D, Aeck G, Waugh R, Caldwell DS, et al. Eects of spouse-assisted coping skills training and exercise training in patients with osteoarthritic knee pain: a randomised controlled study. Pain 2004;110:53949. Keefe FJ, Caldwell DS, Baucom DH, Salley A, Robinson E, Timmons K, et al. Spouse-assisted coping skills training in the management of osteoarthritic knee pain. Arthritis Care Res 1996;9:27991. Kennedy-Moore E, Watson JC. How and when does emotional expression help? Rev Gen Psychol 2001;5:187212. Kerns RD, Turk DC, Rudy TE. The West Haven-Yale Multidimensional Pain Inventory (WHYMPI). Pain 1985;23:34556. Kerns RD, Rosenberg R. Pain-relevant responses from signicant others: development of a signicant-other version of the WHYMPI scales. Pain 1995;61:2459. King C, Kennedy P. Coping eectiveness training for people with spinal cord injury: preliminary results of a controlled trial. Br J Clin Psychol 1999;38:514. Krippendorf K. Content analysis: an introduction to its methodology. Beverly Hills: Sage Publications; 1980. Lee F, Peterson C. Content analysis of archival data. J Consult Clin Psychol 1997;65:95969. Lousberg R, Schmidt AJM, Groenman NH. The relationship between spouse solicitousness and pain behavior: searching for more experimental evidence. Pain 1992;51:759. Miles A, McManus C, Feinmann C, Glover L, Harrison S, Pearce S. The factor structure of the BDI in facial pain and other chronic pain patients: A comparison of two models using conrmatory factor analysis. Br J Health Psychol 2001;6:17996. Navran L. Communication and adjustment in marriage. Fam Process 1967;6:17384.

63

Osborn M, Smith JA. The personal experience of chronic benign lower back pain: an interpretative phenomenological analysis. Br J Health Psychol 1998;3:6584. Papas RK, Robinson ME, Riley J. Perceived spouse responsiveness to chronic pain: three empirical subgroups. J Pain 2001;2:2629. Romano JM, Turner JA, Friedman LS, Bulcroft RA, Jensen MP, Hops H, et al. Sequential analysis of chronic pain behavior and spouse responses. J Consult Clin Psychol 1992;60:77782. Romano JM, Turner JA, Friedman LS, Bulcroft RA, Jensen MP, Hops H, et al. Chronic pain patientspouse behavioural interactions predict patient disability. Pain 1995;63:35360. Rowat KM, Kna KA. Living with chronic pain: the spouses perspective. Pain 1985;23:25971. Schwartz L, Jensen MP, Romano JM. The development and psychometric evaluation of an instrument to assess spouse responses to pain and well behaviour in patients with chronic pain: The Spouse Response Inventory. J Pain 2005;6:24352. Schwartz L, Slater MA, Birchler GR. The role of pain behaviors in the modulation of marital conict in chronic pain couples. Pain 1996;65:22733. Schwartz L, Edhe DM. Couples and chronic pain. In: Schmaling KB, Goldman Sher T, editors. The psychology of couples and illness. Washington, DC: American Psychological Association; 2000. Smith SJ, Keefe FJ, Caldwell DS, Romano J, Baucom D. Gender dierences in patientspouse interactions: a sequential analysis of behavioral interactions in patients having osteoarthritic knee pain. Pain 2004;112:1837. Spanier GB. Measuring dyadic adjustment: new scales for assessing the quality of marriage and similar dyads. J Marriage Fam 1976;38:1528. Turk DC. A cognitive-behavioral perspective on treatment of chronic pain patients. In: Turk DC, Gatchel RJ, editors. Psychological approaches to pain management: a practitioners handbook. New York: Guilford Press; 2002. Turk DC, Rudy TE, Flor H. Why a family perspective for pain? Int J Fam Ther 1985;7:22334. Weber RP. Basic content analysis. Newbury Park, CA: Sage Publications; 1990. Williams AC, Richardson PH. What does the BDI measure in chronic pain? Pain 1993;55:25966. Williams A CdeC, Richardson PH, Nicholas MK, Pither CE, Harding VR, Ridout KL, et al. Inpatient vs. outpatient pain management: results of a randomised controlled trial. Pain 1996;66:1322. Yardley L, Beech S. Im not a doctor: deconstructing accounts of coping, causes and control of dizziness. J Health Psychol 1998;3:31327.