- Email: [email protected]
Remembering an everyday pain: the role of knowledge and experience in the recall of the quality of dysmenorrhoea
Pain 84 (2000) 89±94 www.elsevier.nl/locate/pain
Remembering an everyday pain: the role of knowledge and experience in the recall of the quality of dysmenorrhoea Eric E. Brodie a,*, Catherine A. Niven b a
Department of Psychology, Glasgow Caledonian University, Glasgow, UK Department of Nursing and Midwifery, University of Stirling, Stirling, UK
b
Received 28 September 1998; received in revised form 2 June 1999; accepted 15 July 1999
Abstract The ability to describe the quality of a previous pain may be thought to be better if one had experienced that particular pain because information stored in episodic and/or semantic memory is available rather than if one had not and could only guess what the pain may be like on the basis of information stored in semantic memory. However research has shown that not only is the quality of labour pain poorly recalled by women who have given birth but also it is no better described by them than by women who have never given birth at all. In order to replicate this effect for an everyday pain, the ability to recall the quantity and the quality of dysmenorrhoea was measured in two groups of women. One group regularly experienced dysmenorrhoea, the other had never experienced it. Analysis of the pain intensity scores revealed that the `pain' group reported signi®cantly less pain 2 weeks later whereas the `no-pain' group did not signi®cantly differ in their rating over time. Analysis of the MPQ Descriptors chosen by subjects using Cohen's kappa resulted in `fair' recall for both groups with no signi®cant advantage for the `pain' group. These results suggest that the episodic memory system plays a limited role in facilitating the recall of the quality of an often experienced pain. However the semantic memory system allows both previous pain sufferers and pain guessers to describe the core qualities of a pain to the same extent. Further research is required to explain why remembering the quality of a pain experience is not advantaged by episodic memory and what facilitates the transfer of a pain experience into semantic event memory. q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. Keywords: Memory; Dysmenorrhoea; Pain intensity; Pain quality
1. Introduction Memory for the intensity of pain has been frequently investigated and although numerous factors have been shown to affect its recall, it has generally been shown to be reasonably accurate (for a review see Erskine et al., 1990). In contrast the memory for the quality of pain has rarely been investigated despite the fact that the McGill Pain Questionnaire (MPQ; Melzack, 1975) provides valid and reliable measures of both pain intensity and pain quality (Dubuisson and Melzack, 1976) and has been utilized in countless studies. The few studies which have investigated memory for the quality of pain by comparing MPQ descriptors selected at the time of a pain experience with those selected at a later time have reported varying degrees of ability. Roche and Gijsbers (1986) found that memory for the most frequently selected descriptors of induced ischaemic pain and of chronic rheumatoid pain was highly accurate leading them * Corresponding author.
to suggest that memory for pain quality was better than for pain intensity. Niven and Brodie (1996) found only reasonably accurate recognition of `core' descriptors for labour pain 3±4 years later and suggested that there may only be certain aspects of the pain quality which are retained in memory. However studies which have utilized the Cohen kappa statistic, which controls for chance selection of MPQ descriptors, have found memory for pain quality to be `poor' or at best `fair' (Beese and Morley, 1993; Niven and Brodie, 1996). These equivocal ®ndings may re¯ect differences in theoretical and methodological assumptions rather than differences between memory for pain intensity and quality. Analysis which uses Cohen's kappa considers the subject's response to every descriptor in the MPQ, whereas the analysis of the most frequently selected descriptors concentrates on the de®ning characteristics of the pain syndrome; the pattern of MPQ responses which distinguishes one pain from another (Dubuisson and Melzack, 1975). Remembering a pain and being given descriptors to report it is a cued recall task and it is a qualitatively different memory task
0304-3959/00/$20.00 q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. PII: S 0304-395 9(99)00186-4
90
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
from a free recall task in which no descriptors are given. This was demonstrated by Morley (1993) who carried out a retrospective study of vivid everyday pains. He found that 41% of subjects were unable to freely recall any sensory quality of their pain whereas 100% were able to recall the intensity of their pain. The range of pain types which have been investigated in previous studies of pain quality has been limited, with current published work only considering dental pain, ischaemic pain, labour pain and rheumatoid pain (Roche and Gijsbers, 1986; Beese and Morley, 1993; Niven and Brodie, 1996). There is a need to extend this sample, since a patient's recollection of pain quality is frequently used by physicians as a basis for diagnosis and by doctors, nurses and others for evaluating treatment. Furthermore, although the memory for the quality of a pain has generally been shown to be inaccurate when MPQ descriptors are compared, individual subjects have occasionally demonstrated remarkable accuracy (Niven and Brodie, 1996). Therefore the factors underlying memory differences between pain types and between individuals needs to be further investigated. The memory systems involved in remembering a pain's quality may be more diverse and complex than those involved in remembering a pain's intensity. Niven and Brodie (1996) suggested that the ability to remember the quality of a past pain should result from accessing the semantic and episodic memory systems both of which can retain information about the previous personal pain experience (Tulving, 1977, 1985). Semantic memory stores any fact about the world including those facts that directly relate to the rememberer and episodic memory stores speci®c personal experiences. Thus individuals may be able to describe a past pain on the basis of recollecting the particular episode in which their knowledge of the pain was gained (episodic memory) or they may just know about it (semantic memory). Both systems acquire information in the same way and information cannot be acquired by one system without being acquired by the other. However there is the added complication that information stored in semantic memory may have been derived from publicly available information about labour pain not just from speci®c personal pain experience. The Niven and Brodie (1996) study of labour pain found that women, when asked to remember the pain they had experienced in childbirth, selected a number of pain descriptors which they had selected at the time of birth, although they also `forgot' MPQ descriptors and selected descriptors which they had not previously chosen. However, a comparable group of women who had never given birth and therefore had no information derived from experience also chose a number of these `core' labour pain descriptors. Niven and Brodie (1996) argued that when women recall the quality of their labour pain, their episodic memory system may not play as crucial a role as might be expected and suggested that the presence of factual information not derived from experience may need to be taken
account of when measuring the ability to recall the quality of a particular pain. However, Niven and Brodie's (1996) labour pain study examined recall over 3±4 years; an unusually long time period in empirical research and one which may not have much clinical relevance since clinicians are often more concerned with `how your pain was last week', rather than 4 years ago. It also involved a very distinctive, and infrequently experienced, form of pain; while labour pain can be held to be a good model of acute pain (Melzack, 1993) it is atypical in its association with a highly desirable and memorable life event, one which is widely represented in personal and public accounts, stories, myths and legends. If episodic memory does not play a signi®cant part in the recall of a memorable life event what is its contribution to the recall of a more common, mundane pain experience? One such pain is dysmenorrhoea, a frequently experienced pain associated with the unremarkable health event of menstruation which in our society remains private and hidden and to which women display a wide range of different attitudes, from positive to negative to ambivalent (Walker, 1996). This study aimed to examine the accuracy of recall for the quantity and quality of dysmenorrhoeic pain over a 2 week period in order to establish whether the availability of episodic and semantic memory results in better recall than semantic memory alone. A Visual Analogue Scale (VAS) (Scott and Huskisson, 1976) in which neither words nor numbers are available to specify a pain's intensity and the MPQ in which word descriptors are available to specify both the quantity and quality of a pain were administered to two groups of subjects. One group of subjects was experiencing dysmenorrhoeic pain at the time of ®rst responding to the VAS and the MPQ but experiencing no pain at the second time 2 weeks later, and the other group of subjects had never experienced dysmenorrhoea. 2. Methods 2.1. Subjects 49 healthy, regularly menstruating women aged between 16 and 40 (mean 28 years) were recruited from a university setting, all of whom gave informed consent for participation in the study. They comprised two groups: a `Pain' group, 34 women who regularly experienced dysmenorrhoea, and a `No-pain' group, 15 women who had never experienced dysmenorrhoea. Their description of dysmenorrhoea was therefore dependent on semantic information. 2.2. Procedure Both groups were administered the MPQ and VAS at Time 1, during the ®rst 2 days of menstruation when dysmenorrhoea is usually most acute and at Time 2, two weeks later. The MPQ was modi®ed to allow for complete descrip-
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
91
tor availability by the amalgamation of the words from all twenty of its subclasses. Subjects could therefore select more than the maximum of twenty words allowed by the standard MPQ.
descriptors most frequently selected by the `No-pain' group at Time 1 with those of the `pain' group at Time 2. The more closely these resembled each other, the more likely that semantic memory was involved in pain recall.
2.2.1. Instructions
2.3.3. Accuracy of recall This was assessed by comparing the MPQ responses and VAS responses of the Pain group, where subjects had an experience of pain to draw on, with that for the No-pain group, where subjects could only draw upon the memory of the responses previously made.
2.2.1.1. Pain group: Time 1. `Please assess your period pain when it is at its most severe using the Visual Analogue Scale and the McGill Pain Questionnaire.' 2.2.1.2. Pain group: Time 2. `Try to remember how your period pain felt when you assessed it 2 weeks ago. Keeping that in mind, please assess it using the VAS and MPQ.'
3. Results
2.2.1.3. No-pain group: Time 1. `Please assess how painful you think period pain is using the Visual Analogue Scale. Please choose the words from the McGill Pain Questionnaire which you think best describe period pain.'
3.1. Recall of the quality of dysmenorrhoea
2.2.1.4. No-pain group: Time 2. `Try to remember how you assessed period pain 2 weeks ago. Keeping that in mind, please assess it again using the VAS and MPQ.'
Table 1 shows the descriptors most frequently chosen by dysmenorrhoeic subjects at Time 1 and Time 2. There is a considerable overlap of descriptors selected with the majority of descriptors being common to both lists. Analysis by Cohen's kappa yielded a value of k 0:53 (SD 0.24), which is considered to re¯ect `fair' accuracy of recall.
2.3. Analysis
3.2. Effects of semantic memory
2.3.1. Recall of the quality of dysmenorrhoea Subjects' selections of descriptors at Time 1 and 2 were compared in two ways:
Four of the ®ve MPQ descriptors selected most frequently by the Pain group at Time 1 and at recall were also selected by the No-pain group (those who had never experienced dysmenorrhoea). The descriptors `cramping', `tiring', `aching' and `tender' were common to each group's MPQ assessment of dysmenorrhoea.
1. By comparing the most frequently chosen descriptors from the MPQ within the `Pain' group at Time 1 (during menstruation) and Time 2 (2 weeks later) by 33% or more of the subject group (Roche and Gijsbers, 1986). 2. By using Cohen's kappa. Kappa values are grouped into three evaluations; values above 0.75 are regarded as `excellent', values from 0.4 to 0.6 as `fair', and those below 0.4 as representing `poor accuracy' (Fleiss, 1981).
3.3. Accuracy of recall At Time 2, the `No-pain' group was asked to recall their previous MPQ response. Cohen's kappa was calculated to assess the accuracy of their recall of pain descriptors. The kappa value was 0.43 (SD 0.24). As with the Pain group who had been asked to recall their previous pain experience, this value fell within the `fair' range. A t-test was carried out
2.3.2. Effects of semantic memory These were assessed by comparing the MPQ pain Table 1 Words most frequently chosen to describe dysmenorrhoea ( $ 33% of subjects) `Pain' group
`No-pain' group
During menstruation % Cramping Tiring Aching Tender Miserable Nagging
2 weeks later % 56 47 38 35 35 33
Cramping Aching Tiring Throbbing Tender Miserable Boring
During painless menstruation % 61 47 41 35 33 33 33
Cramping Aching Tiring Intense Tender Taut Nauseating Dull Sore
2 weeks later % 80 53 53 40 33 33 33 33 33
Cramping Aching Tiring Intense Throbbing Nauseating Sickening Sharp Tight Searing
80 53 53 47 40 40 33 33 33 33
92
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
Table 2 Intensity of Dysmenorrhoeic pain Group
Pain Pain No-pain No-pain
Time
1 2 1 2
n
34 34 15 15
Number of MPQ descriptors chosen
VAS pain scores
Mean
SD
Min
Max
Mean
SD
Min
Max
8.06 8.12 10.40 11.47
4.40 5.03 8.46 7.55
2 3 2 3
21 25 35 31
58.03 54.58 54.67 61.33
21.95 21.42 28.69 23.45
6 8 5 26
99 85 90 93
between kappa values for Pain and No-pain groups. No signi®cant difference was found (t 47 1:248, P . 0:05). Table 2 shows the patterns of response to the MPQ and VAS at Time 1 and Time 2 for both Pain and No-pain groups. A repeated measure analysis of variance was carried out upon the number of words chosen from the MPQ at Time 1 and at Time 2 with one between subject factor (group). No signi®cant main effects or interaction were found. Thus the number of words chosen from the MPQ by the Pain and No-pain groups did not differ. A repeated measure analysis of variance was carried out upon the VAS scores at Time 1 and at Time 2 with one between subject factor (group). No signi®cant main effects were found but a signi®cant interaction of group £ time was found (F 1; 47 6:93, P , 0:05). This was due to the intensity of dysmenorrhoea as assessed by the VAS being signi®cantly different at Time 1 from Time 2 for the Pain group (t 33 2:60, P , 0:015) but not for the No-pain group (t 14 2 1:31, P . 0:05). In fact the intensity of dysmenorrhoea was reported to be signi®cantly less at Time 2 (54.58) than at Time 1 (58.0) when they were experiencing the pain. 4. Discussion The fact that dysmenorrhoea is a very common pain experience occurring throughout a woman's reproductive years (Reading, 1979) was con®rmed in this study by the dif®culty in recruiting a group of healthy subjects who had never experienced it. This limited the size of the No-pain group. The dysmenorrhoeic subjects who participated in this study reported pain levels comparable with those of other studies (Reading, 1979). Additionally the ®nding that the reported level of dysmenorrhoeic pain as assessed by a visual analogue scale was signi®cantly less 2 weeks later is in keeping with previous research which found that the accuracy of recall of a pain's intensity is poor when repeated episodes of the pain are experienced (Roche and Gijsbers, 1986). Not only is this further evidence in support of the hypothesis that the intensity of chronic/often experienced pain is less well remembered than acute/novel pain (Erskine et al., 1990), but that there is a tendency to underestimate its intensity. The recall of pain quality by the Pain group was similar to
that found in other studies where recall was over a short period (Roche and Gijsbers, 1986; Beese and Morley, 1993) but superior, on the kappa analysis, to that found in the study of labour pain (Niven and Brodie, 1996). Since the subjects of the latter study were recalling a pain they had experienced 3±4 years previously, this is not surprising. However, this study con®rms that only a small proportion of MPQ descriptors is retained at recall (Roche and Gijsbers, 1986; Niven and Brodie, 1996). The existence of such a description, which dominates recollections of a pain experience, is consistent with a model of pain quality recall which is at best considered `fair' when analysed by Cohen's kappa. The use of the MPQ to assess recall of the quality of pain presents the subject with a very large number of choices to make when articulating a pain experience. It is unsurprising that both omissions and intrusions are made. When subjects are asked to describe a pain using their own words, they typically use three or four words at most (Niven and Gijsbers, 1984), which neither approaches the 20 available in the standard MPQ nor the 78 available with the modi®ed MPQ used in this study. Perhaps, as Fernandez and Towery (1996) have suggested, our natural pain vocabulary is quite limited and a core description of pain quality is all that can be reliably held in semantic memory. Redelmeier and Kahneman (1996) found that subjects experiencing a pain episode associated with lengthy medical treatment did not recall every aspect of pain intensity. Their recall was dominated by peak pain (the pain at its most intense) and end pain. They suggested that ``episodes of pain are extremely complex and storing all the details might be overwhelming'' (Redelmeier and Kahneman, 1996, p. 7), so that retaining only some aspects of the experience makes sense. We would argue that a similar reductive process occurs when subjects recall pain quality. Not every aspect of pain quality will be retained, only its de®ning qualities. Indeed, since the quality of pain will vary with its intensity, the recall of pain quality may be that associated with peak pain or with the pain experienced at the end of the episode, although naturally occurring pains like dysmenorrhoea as opposed to those induced by procedures, do not have an easily identi®able end point. More research is clearly needed to examine the parameters of recall of pain quality over the duration of a pain episode. The description of dysmenorrhoea provided by the Nopain group is remarkably similar to that recorded by Pain
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
group during their experience and at recall. A core description emerges which is shared between sufferers and those who have never experienced it but who draw on their general knowledge of pain held in semantic memory. Dysmenorrhoea is cramping, aching, tiring and tender. If semantic memory is both in¯uenced by and in¯uences episodic memory, memory which is concerned with speci®c personally experienced events (Tulving, 1977, 1985), this would suggest that a core description of pain derived from personal experience both in¯uences and is in¯uenced by semantic general knowledge about the pain type. For subjects to gain general knowledge about the nature of a pain, they must encounter its description, as provided by people or media. While labour pain is widely portrayed in the media through its association with the dramatic events of childbirth, period pain is rarely the subject of ®lm or drama, nor is it often portrayed in the TV medical soaps but remains private and hidden (Walker, 1996). The main source of general knowledge about period pain is in women's accounts. If these accounts re¯ect women's recollections of dysmenorrhoea, they may also re¯ect the in¯uence on those of semantic memory. Thus the congruence we have found between the core descriptions of dysmenorrhoea during the pain episode, at recall and when described by subjects who have never experienced it could re¯ect the bi-directional effects of episodic and semantic memory. However the concept of episodic memory has recently changed from the memory for temporally dated events to the conscious re-experiencing of events from one's past (Wheeler et al., 1997). This re-experiencing is dependent upon autonoetic consciousness which allows us to mentally represent past subjective experiences in the present. Thus the memory of a pain episode differs from the memory of pain facts not in terms of the type of information but in terms of the subjective experience at retrieval. If autonoetic consciousness was used by subjects to re-experience dysmenorrhoea then it may be expected that the recall of the pain quality be better than `fair' and the ratings of pain intensity should not have differed signi®cantly. Further evidence for the absence of autonoetic consciousness from pain recall has also been reported by Morley (1993) who found that not one subject out of 136 could re-experience a vivid everyday pain. However the re-experiencing of pain does occur for the pain prior to the amputation of a limb and can be triggered by thoughts and feelings (Katz, 1992) and for the pain of labour if triggered by viewing a video of the second stage of labour (McKay and Lund Barrows, 1992). We would argue that for often repeated pain events the semantic memory system is relied upon to recall a pain quality as it deals with facts about the pain a person has experienced and allows a person to know that a MPQ descriptor applies to a pain without consciously recollecting or re-experiencing the pain episode. In this study an attempt was made to address the issue of accuracy of recall by comparing the Pain subjects' recall of a pain experience as assessed on the MPQ with the No-pain
93
subjects' recall of their previous MPQ responses. In both groups, the recall was similar, with kappas falling in the `fair' range. This suggests that the task of recognising the MPQ descriptors selected 2 weeks previously, is not particularly simple. Although the Pain subjects had higher kappa values, even with the small number of No-pain subjects post-hoc analysis demonstrated that the difference was non-signi®cant. Although the recall of pain intensity was not the focus of this study, it is interesting to note that both No-pain and Pain groups assessed dysmenorrhoea as moderately severe, placing their VAS response at approximately the midpoint of the VAS scale. The use of the MPQ responses to calculate pain intensity, similarly resulted in assessments which did not differ signi®cantly between those who recorded their pain and recalled it and those who drew on semantic knowledge to assess dysmenorrhoea. The congruence between these data is notable re¯ecting that information about the quality and quantity of a pain is freely available in society. Further evidence to support this view has come from research carried out in non medicalised societies where it was found that individuals could readily rank order the intensity of pains of diverse origin (labour pain, the pain of a broken leg, toothache etc.), even when they have no personal experience of the pains (Morse, 1989). However the ®nding that women who have never experienced dysmenorrhoea or labour pain (Niven and Brodie, 1996) are as good at describing those pains as women who have experienced them may not be generalisable to male pain experiences. This is because recent research reported that women are more aware of the pain in others than men (Koutantji et al., 1998). Thus further research is required to disentangle the in¯uence of semantic and episodic memory systems in the recall of pain which takes account of the fact that some pains are gender speci®c and women may be more knowledgeable about the pain experience of others than men.
References Beese A, Morley S. Memory for acute pain experience is speci®cally inaccurate but generally reliable. Pain 1993;53:183±189. Dubuisson D, Melzack R. Classi®cation of clinical pain descriptors by multiple group discriminant analysis. Exp Neurol 1976;51(2):480±487. Erskine A, Morley S, Pearce S. Memory for pain: a review. Pain 1990;41:255±265. Fernandez E, Towery S. A parsimonious set of verbal descriptors of pain sensation derived from the McGill Pain Questionnaire. Pain 1996;66:31±37. Fleiss J. Statistical methods for rates and proportions, New York: Wiley, 1981. Katz J. Psychophysiological contributions to phantom limbs. Can. J. Psychiatry 1992;37(5):282±298. Koutantji M, Pearce SA, Oakley DA. The relationship between gender and family history of pain with current pain experience and awareness of pain in others. Pain 1998;77:25±31. McKay S, Lund Barrows T. Reliving birth: maternal responses to viewing
94
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
video tapes of their second stage labors. Image J Nursing Scholarship 1992;24:27±33. Melzack R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975;1:277±299. Melzack R. Labor pain as a model of acute pain. Pain 1993;53:117±121. Morley S. Vivid memory for everyday pains. Pain 1993;55:55±62. Morse JM. Cultural variation in behavioural responses to parturition: childbirth in Fiji. Med Anthropol 1989;12:35±54. Niven CA, Brodie EE. Memory for labor pain: Context and quality. Pain 1996;64:387±392. Niven CA, Gijsbers K. Obstetric & non obstetric factors related to labor pain. J Reproduct Infant Psychol 1984;2:61±78. Reading A. The internal structure of the McGill Pain questionnaire in dysmenorrhoea patients. Pain 1979;7:353±358. Redelmeier DA, Kahneman D. Patients' memories of painful medical treat-
ments: real time and retrospective evaluations of two minimally invasive procedures. Pain 1996;66:3±9. Roche P, Gijsbers K. A comparison for memory for induced ischemic pain and chronic rheumatoid pain. Pain 1986;24:337±343. Scott J, Huskisson EC. Graphic representation of pain. Pain 1976;2:175± 184. Tulving E. Episodic and semantic memory. In: Tulving E, Donaldson W, editors. Organization of memory, New York: Academic Press, 1977. Tulving E. How many memory systems are there? Am. Psychol. 1985;40:385±398. Walker A. Menstruation and the premenstrual syndrome. In: Niven CA, Walker A, editors. Reproductive potential and fertility control, Oxford: Butterworth Heineman, 1996. pp. 39±58. Wheeler MA, Stuss DT, Tulving E. Towards a theory of episodic memory. Psychol. Bull. 1997;121(3):331±354.
Remembering an everyday pain: the role of knowledge and experience in the recall of the quality of dysmenorrhoea Eric E. Brodie a,*, Catherine A. Niven b a
Department of Psychology, Glasgow Caledonian University, Glasgow, UK Department of Nursing and Midwifery, University of Stirling, Stirling, UK
b
Received 28 September 1998; received in revised form 2 June 1999; accepted 15 July 1999
Abstract The ability to describe the quality of a previous pain may be thought to be better if one had experienced that particular pain because information stored in episodic and/or semantic memory is available rather than if one had not and could only guess what the pain may be like on the basis of information stored in semantic memory. However research has shown that not only is the quality of labour pain poorly recalled by women who have given birth but also it is no better described by them than by women who have never given birth at all. In order to replicate this effect for an everyday pain, the ability to recall the quantity and the quality of dysmenorrhoea was measured in two groups of women. One group regularly experienced dysmenorrhoea, the other had never experienced it. Analysis of the pain intensity scores revealed that the `pain' group reported signi®cantly less pain 2 weeks later whereas the `no-pain' group did not signi®cantly differ in their rating over time. Analysis of the MPQ Descriptors chosen by subjects using Cohen's kappa resulted in `fair' recall for both groups with no signi®cant advantage for the `pain' group. These results suggest that the episodic memory system plays a limited role in facilitating the recall of the quality of an often experienced pain. However the semantic memory system allows both previous pain sufferers and pain guessers to describe the core qualities of a pain to the same extent. Further research is required to explain why remembering the quality of a pain experience is not advantaged by episodic memory and what facilitates the transfer of a pain experience into semantic event memory. q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. Keywords: Memory; Dysmenorrhoea; Pain intensity; Pain quality
1. Introduction Memory for the intensity of pain has been frequently investigated and although numerous factors have been shown to affect its recall, it has generally been shown to be reasonably accurate (for a review see Erskine et al., 1990). In contrast the memory for the quality of pain has rarely been investigated despite the fact that the McGill Pain Questionnaire (MPQ; Melzack, 1975) provides valid and reliable measures of both pain intensity and pain quality (Dubuisson and Melzack, 1976) and has been utilized in countless studies. The few studies which have investigated memory for the quality of pain by comparing MPQ descriptors selected at the time of a pain experience with those selected at a later time have reported varying degrees of ability. Roche and Gijsbers (1986) found that memory for the most frequently selected descriptors of induced ischaemic pain and of chronic rheumatoid pain was highly accurate leading them * Corresponding author.
to suggest that memory for pain quality was better than for pain intensity. Niven and Brodie (1996) found only reasonably accurate recognition of `core' descriptors for labour pain 3±4 years later and suggested that there may only be certain aspects of the pain quality which are retained in memory. However studies which have utilized the Cohen kappa statistic, which controls for chance selection of MPQ descriptors, have found memory for pain quality to be `poor' or at best `fair' (Beese and Morley, 1993; Niven and Brodie, 1996). These equivocal ®ndings may re¯ect differences in theoretical and methodological assumptions rather than differences between memory for pain intensity and quality. Analysis which uses Cohen's kappa considers the subject's response to every descriptor in the MPQ, whereas the analysis of the most frequently selected descriptors concentrates on the de®ning characteristics of the pain syndrome; the pattern of MPQ responses which distinguishes one pain from another (Dubuisson and Melzack, 1975). Remembering a pain and being given descriptors to report it is a cued recall task and it is a qualitatively different memory task
0304-3959/00/$20.00 q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. PII: S 0304-395 9(99)00186-4
90
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
from a free recall task in which no descriptors are given. This was demonstrated by Morley (1993) who carried out a retrospective study of vivid everyday pains. He found that 41% of subjects were unable to freely recall any sensory quality of their pain whereas 100% were able to recall the intensity of their pain. The range of pain types which have been investigated in previous studies of pain quality has been limited, with current published work only considering dental pain, ischaemic pain, labour pain and rheumatoid pain (Roche and Gijsbers, 1986; Beese and Morley, 1993; Niven and Brodie, 1996). There is a need to extend this sample, since a patient's recollection of pain quality is frequently used by physicians as a basis for diagnosis and by doctors, nurses and others for evaluating treatment. Furthermore, although the memory for the quality of a pain has generally been shown to be inaccurate when MPQ descriptors are compared, individual subjects have occasionally demonstrated remarkable accuracy (Niven and Brodie, 1996). Therefore the factors underlying memory differences between pain types and between individuals needs to be further investigated. The memory systems involved in remembering a pain's quality may be more diverse and complex than those involved in remembering a pain's intensity. Niven and Brodie (1996) suggested that the ability to remember the quality of a past pain should result from accessing the semantic and episodic memory systems both of which can retain information about the previous personal pain experience (Tulving, 1977, 1985). Semantic memory stores any fact about the world including those facts that directly relate to the rememberer and episodic memory stores speci®c personal experiences. Thus individuals may be able to describe a past pain on the basis of recollecting the particular episode in which their knowledge of the pain was gained (episodic memory) or they may just know about it (semantic memory). Both systems acquire information in the same way and information cannot be acquired by one system without being acquired by the other. However there is the added complication that information stored in semantic memory may have been derived from publicly available information about labour pain not just from speci®c personal pain experience. The Niven and Brodie (1996) study of labour pain found that women, when asked to remember the pain they had experienced in childbirth, selected a number of pain descriptors which they had selected at the time of birth, although they also `forgot' MPQ descriptors and selected descriptors which they had not previously chosen. However, a comparable group of women who had never given birth and therefore had no information derived from experience also chose a number of these `core' labour pain descriptors. Niven and Brodie (1996) argued that when women recall the quality of their labour pain, their episodic memory system may not play as crucial a role as might be expected and suggested that the presence of factual information not derived from experience may need to be taken
account of when measuring the ability to recall the quality of a particular pain. However, Niven and Brodie's (1996) labour pain study examined recall over 3±4 years; an unusually long time period in empirical research and one which may not have much clinical relevance since clinicians are often more concerned with `how your pain was last week', rather than 4 years ago. It also involved a very distinctive, and infrequently experienced, form of pain; while labour pain can be held to be a good model of acute pain (Melzack, 1993) it is atypical in its association with a highly desirable and memorable life event, one which is widely represented in personal and public accounts, stories, myths and legends. If episodic memory does not play a signi®cant part in the recall of a memorable life event what is its contribution to the recall of a more common, mundane pain experience? One such pain is dysmenorrhoea, a frequently experienced pain associated with the unremarkable health event of menstruation which in our society remains private and hidden and to which women display a wide range of different attitudes, from positive to negative to ambivalent (Walker, 1996). This study aimed to examine the accuracy of recall for the quantity and quality of dysmenorrhoeic pain over a 2 week period in order to establish whether the availability of episodic and semantic memory results in better recall than semantic memory alone. A Visual Analogue Scale (VAS) (Scott and Huskisson, 1976) in which neither words nor numbers are available to specify a pain's intensity and the MPQ in which word descriptors are available to specify both the quantity and quality of a pain were administered to two groups of subjects. One group of subjects was experiencing dysmenorrhoeic pain at the time of ®rst responding to the VAS and the MPQ but experiencing no pain at the second time 2 weeks later, and the other group of subjects had never experienced dysmenorrhoea. 2. Methods 2.1. Subjects 49 healthy, regularly menstruating women aged between 16 and 40 (mean 28 years) were recruited from a university setting, all of whom gave informed consent for participation in the study. They comprised two groups: a `Pain' group, 34 women who regularly experienced dysmenorrhoea, and a `No-pain' group, 15 women who had never experienced dysmenorrhoea. Their description of dysmenorrhoea was therefore dependent on semantic information. 2.2. Procedure Both groups were administered the MPQ and VAS at Time 1, during the ®rst 2 days of menstruation when dysmenorrhoea is usually most acute and at Time 2, two weeks later. The MPQ was modi®ed to allow for complete descrip-
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
91
tor availability by the amalgamation of the words from all twenty of its subclasses. Subjects could therefore select more than the maximum of twenty words allowed by the standard MPQ.
descriptors most frequently selected by the `No-pain' group at Time 1 with those of the `pain' group at Time 2. The more closely these resembled each other, the more likely that semantic memory was involved in pain recall.
2.2.1. Instructions
2.3.3. Accuracy of recall This was assessed by comparing the MPQ responses and VAS responses of the Pain group, where subjects had an experience of pain to draw on, with that for the No-pain group, where subjects could only draw upon the memory of the responses previously made.
2.2.1.1. Pain group: Time 1. `Please assess your period pain when it is at its most severe using the Visual Analogue Scale and the McGill Pain Questionnaire.' 2.2.1.2. Pain group: Time 2. `Try to remember how your period pain felt when you assessed it 2 weeks ago. Keeping that in mind, please assess it using the VAS and MPQ.'
3. Results
2.2.1.3. No-pain group: Time 1. `Please assess how painful you think period pain is using the Visual Analogue Scale. Please choose the words from the McGill Pain Questionnaire which you think best describe period pain.'
3.1. Recall of the quality of dysmenorrhoea
2.2.1.4. No-pain group: Time 2. `Try to remember how you assessed period pain 2 weeks ago. Keeping that in mind, please assess it again using the VAS and MPQ.'
Table 1 shows the descriptors most frequently chosen by dysmenorrhoeic subjects at Time 1 and Time 2. There is a considerable overlap of descriptors selected with the majority of descriptors being common to both lists. Analysis by Cohen's kappa yielded a value of k 0:53 (SD 0.24), which is considered to re¯ect `fair' accuracy of recall.
2.3. Analysis
3.2. Effects of semantic memory
2.3.1. Recall of the quality of dysmenorrhoea Subjects' selections of descriptors at Time 1 and 2 were compared in two ways:
Four of the ®ve MPQ descriptors selected most frequently by the Pain group at Time 1 and at recall were also selected by the No-pain group (those who had never experienced dysmenorrhoea). The descriptors `cramping', `tiring', `aching' and `tender' were common to each group's MPQ assessment of dysmenorrhoea.
1. By comparing the most frequently chosen descriptors from the MPQ within the `Pain' group at Time 1 (during menstruation) and Time 2 (2 weeks later) by 33% or more of the subject group (Roche and Gijsbers, 1986). 2. By using Cohen's kappa. Kappa values are grouped into three evaluations; values above 0.75 are regarded as `excellent', values from 0.4 to 0.6 as `fair', and those below 0.4 as representing `poor accuracy' (Fleiss, 1981).
3.3. Accuracy of recall At Time 2, the `No-pain' group was asked to recall their previous MPQ response. Cohen's kappa was calculated to assess the accuracy of their recall of pain descriptors. The kappa value was 0.43 (SD 0.24). As with the Pain group who had been asked to recall their previous pain experience, this value fell within the `fair' range. A t-test was carried out
2.3.2. Effects of semantic memory These were assessed by comparing the MPQ pain Table 1 Words most frequently chosen to describe dysmenorrhoea ( $ 33% of subjects) `Pain' group
`No-pain' group
During menstruation % Cramping Tiring Aching Tender Miserable Nagging
2 weeks later % 56 47 38 35 35 33
Cramping Aching Tiring Throbbing Tender Miserable Boring
During painless menstruation % 61 47 41 35 33 33 33
Cramping Aching Tiring Intense Tender Taut Nauseating Dull Sore
2 weeks later % 80 53 53 40 33 33 33 33 33
Cramping Aching Tiring Intense Throbbing Nauseating Sickening Sharp Tight Searing
80 53 53 47 40 40 33 33 33 33
92
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
Table 2 Intensity of Dysmenorrhoeic pain Group
Pain Pain No-pain No-pain
Time
1 2 1 2
n
34 34 15 15
Number of MPQ descriptors chosen
VAS pain scores
Mean
SD
Min
Max
Mean
SD
Min
Max
8.06 8.12 10.40 11.47
4.40 5.03 8.46 7.55
2 3 2 3
21 25 35 31
58.03 54.58 54.67 61.33
21.95 21.42 28.69 23.45
6 8 5 26
99 85 90 93
between kappa values for Pain and No-pain groups. No signi®cant difference was found (t 47 1:248, P . 0:05). Table 2 shows the patterns of response to the MPQ and VAS at Time 1 and Time 2 for both Pain and No-pain groups. A repeated measure analysis of variance was carried out upon the number of words chosen from the MPQ at Time 1 and at Time 2 with one between subject factor (group). No signi®cant main effects or interaction were found. Thus the number of words chosen from the MPQ by the Pain and No-pain groups did not differ. A repeated measure analysis of variance was carried out upon the VAS scores at Time 1 and at Time 2 with one between subject factor (group). No signi®cant main effects were found but a signi®cant interaction of group £ time was found (F 1; 47 6:93, P , 0:05). This was due to the intensity of dysmenorrhoea as assessed by the VAS being signi®cantly different at Time 1 from Time 2 for the Pain group (t 33 2:60, P , 0:015) but not for the No-pain group (t 14 2 1:31, P . 0:05). In fact the intensity of dysmenorrhoea was reported to be signi®cantly less at Time 2 (54.58) than at Time 1 (58.0) when they were experiencing the pain. 4. Discussion The fact that dysmenorrhoea is a very common pain experience occurring throughout a woman's reproductive years (Reading, 1979) was con®rmed in this study by the dif®culty in recruiting a group of healthy subjects who had never experienced it. This limited the size of the No-pain group. The dysmenorrhoeic subjects who participated in this study reported pain levels comparable with those of other studies (Reading, 1979). Additionally the ®nding that the reported level of dysmenorrhoeic pain as assessed by a visual analogue scale was signi®cantly less 2 weeks later is in keeping with previous research which found that the accuracy of recall of a pain's intensity is poor when repeated episodes of the pain are experienced (Roche and Gijsbers, 1986). Not only is this further evidence in support of the hypothesis that the intensity of chronic/often experienced pain is less well remembered than acute/novel pain (Erskine et al., 1990), but that there is a tendency to underestimate its intensity. The recall of pain quality by the Pain group was similar to
that found in other studies where recall was over a short period (Roche and Gijsbers, 1986; Beese and Morley, 1993) but superior, on the kappa analysis, to that found in the study of labour pain (Niven and Brodie, 1996). Since the subjects of the latter study were recalling a pain they had experienced 3±4 years previously, this is not surprising. However, this study con®rms that only a small proportion of MPQ descriptors is retained at recall (Roche and Gijsbers, 1986; Niven and Brodie, 1996). The existence of such a description, which dominates recollections of a pain experience, is consistent with a model of pain quality recall which is at best considered `fair' when analysed by Cohen's kappa. The use of the MPQ to assess recall of the quality of pain presents the subject with a very large number of choices to make when articulating a pain experience. It is unsurprising that both omissions and intrusions are made. When subjects are asked to describe a pain using their own words, they typically use three or four words at most (Niven and Gijsbers, 1984), which neither approaches the 20 available in the standard MPQ nor the 78 available with the modi®ed MPQ used in this study. Perhaps, as Fernandez and Towery (1996) have suggested, our natural pain vocabulary is quite limited and a core description of pain quality is all that can be reliably held in semantic memory. Redelmeier and Kahneman (1996) found that subjects experiencing a pain episode associated with lengthy medical treatment did not recall every aspect of pain intensity. Their recall was dominated by peak pain (the pain at its most intense) and end pain. They suggested that ``episodes of pain are extremely complex and storing all the details might be overwhelming'' (Redelmeier and Kahneman, 1996, p. 7), so that retaining only some aspects of the experience makes sense. We would argue that a similar reductive process occurs when subjects recall pain quality. Not every aspect of pain quality will be retained, only its de®ning qualities. Indeed, since the quality of pain will vary with its intensity, the recall of pain quality may be that associated with peak pain or with the pain experienced at the end of the episode, although naturally occurring pains like dysmenorrhoea as opposed to those induced by procedures, do not have an easily identi®able end point. More research is clearly needed to examine the parameters of recall of pain quality over the duration of a pain episode. The description of dysmenorrhoea provided by the Nopain group is remarkably similar to that recorded by Pain
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
group during their experience and at recall. A core description emerges which is shared between sufferers and those who have never experienced it but who draw on their general knowledge of pain held in semantic memory. Dysmenorrhoea is cramping, aching, tiring and tender. If semantic memory is both in¯uenced by and in¯uences episodic memory, memory which is concerned with speci®c personally experienced events (Tulving, 1977, 1985), this would suggest that a core description of pain derived from personal experience both in¯uences and is in¯uenced by semantic general knowledge about the pain type. For subjects to gain general knowledge about the nature of a pain, they must encounter its description, as provided by people or media. While labour pain is widely portrayed in the media through its association with the dramatic events of childbirth, period pain is rarely the subject of ®lm or drama, nor is it often portrayed in the TV medical soaps but remains private and hidden (Walker, 1996). The main source of general knowledge about period pain is in women's accounts. If these accounts re¯ect women's recollections of dysmenorrhoea, they may also re¯ect the in¯uence on those of semantic memory. Thus the congruence we have found between the core descriptions of dysmenorrhoea during the pain episode, at recall and when described by subjects who have never experienced it could re¯ect the bi-directional effects of episodic and semantic memory. However the concept of episodic memory has recently changed from the memory for temporally dated events to the conscious re-experiencing of events from one's past (Wheeler et al., 1997). This re-experiencing is dependent upon autonoetic consciousness which allows us to mentally represent past subjective experiences in the present. Thus the memory of a pain episode differs from the memory of pain facts not in terms of the type of information but in terms of the subjective experience at retrieval. If autonoetic consciousness was used by subjects to re-experience dysmenorrhoea then it may be expected that the recall of the pain quality be better than `fair' and the ratings of pain intensity should not have differed signi®cantly. Further evidence for the absence of autonoetic consciousness from pain recall has also been reported by Morley (1993) who found that not one subject out of 136 could re-experience a vivid everyday pain. However the re-experiencing of pain does occur for the pain prior to the amputation of a limb and can be triggered by thoughts and feelings (Katz, 1992) and for the pain of labour if triggered by viewing a video of the second stage of labour (McKay and Lund Barrows, 1992). We would argue that for often repeated pain events the semantic memory system is relied upon to recall a pain quality as it deals with facts about the pain a person has experienced and allows a person to know that a MPQ descriptor applies to a pain without consciously recollecting or re-experiencing the pain episode. In this study an attempt was made to address the issue of accuracy of recall by comparing the Pain subjects' recall of a pain experience as assessed on the MPQ with the No-pain
93
subjects' recall of their previous MPQ responses. In both groups, the recall was similar, with kappas falling in the `fair' range. This suggests that the task of recognising the MPQ descriptors selected 2 weeks previously, is not particularly simple. Although the Pain subjects had higher kappa values, even with the small number of No-pain subjects post-hoc analysis demonstrated that the difference was non-signi®cant. Although the recall of pain intensity was not the focus of this study, it is interesting to note that both No-pain and Pain groups assessed dysmenorrhoea as moderately severe, placing their VAS response at approximately the midpoint of the VAS scale. The use of the MPQ responses to calculate pain intensity, similarly resulted in assessments which did not differ signi®cantly between those who recorded their pain and recalled it and those who drew on semantic knowledge to assess dysmenorrhoea. The congruence between these data is notable re¯ecting that information about the quality and quantity of a pain is freely available in society. Further evidence to support this view has come from research carried out in non medicalised societies where it was found that individuals could readily rank order the intensity of pains of diverse origin (labour pain, the pain of a broken leg, toothache etc.), even when they have no personal experience of the pains (Morse, 1989). However the ®nding that women who have never experienced dysmenorrhoea or labour pain (Niven and Brodie, 1996) are as good at describing those pains as women who have experienced them may not be generalisable to male pain experiences. This is because recent research reported that women are more aware of the pain in others than men (Koutantji et al., 1998). Thus further research is required to disentangle the in¯uence of semantic and episodic memory systems in the recall of pain which takes account of the fact that some pains are gender speci®c and women may be more knowledgeable about the pain experience of others than men.
References Beese A, Morley S. Memory for acute pain experience is speci®cally inaccurate but generally reliable. Pain 1993;53:183±189. Dubuisson D, Melzack R. Classi®cation of clinical pain descriptors by multiple group discriminant analysis. Exp Neurol 1976;51(2):480±487. Erskine A, Morley S, Pearce S. Memory for pain: a review. Pain 1990;41:255±265. Fernandez E, Towery S. A parsimonious set of verbal descriptors of pain sensation derived from the McGill Pain Questionnaire. Pain 1996;66:31±37. Fleiss J. Statistical methods for rates and proportions, New York: Wiley, 1981. Katz J. Psychophysiological contributions to phantom limbs. Can. J. Psychiatry 1992;37(5):282±298. Koutantji M, Pearce SA, Oakley DA. The relationship between gender and family history of pain with current pain experience and awareness of pain in others. Pain 1998;77:25±31. McKay S, Lund Barrows T. Reliving birth: maternal responses to viewing
94
E.E. Brodie, C.A. Niven / Pain 84 (2000) 89±94
video tapes of their second stage labors. Image J Nursing Scholarship 1992;24:27±33. Melzack R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975;1:277±299. Melzack R. Labor pain as a model of acute pain. Pain 1993;53:117±121. Morley S. Vivid memory for everyday pains. Pain 1993;55:55±62. Morse JM. Cultural variation in behavioural responses to parturition: childbirth in Fiji. Med Anthropol 1989;12:35±54. Niven CA, Brodie EE. Memory for labor pain: Context and quality. Pain 1996;64:387±392. Niven CA, Gijsbers K. Obstetric & non obstetric factors related to labor pain. J Reproduct Infant Psychol 1984;2:61±78. Reading A. The internal structure of the McGill Pain questionnaire in dysmenorrhoea patients. Pain 1979;7:353±358. Redelmeier DA, Kahneman D. Patients' memories of painful medical treat-
ments: real time and retrospective evaluations of two minimally invasive procedures. Pain 1996;66:3±9. Roche P, Gijsbers K. A comparison for memory for induced ischemic pain and chronic rheumatoid pain. Pain 1986;24:337±343. Scott J, Huskisson EC. Graphic representation of pain. Pain 1976;2:175± 184. Tulving E. Episodic and semantic memory. In: Tulving E, Donaldson W, editors. Organization of memory, New York: Academic Press, 1977. Tulving E. How many memory systems are there? Am. Psychol. 1985;40:385±398. Walker A. Menstruation and the premenstrual syndrome. In: Niven CA, Walker A, editors. Reproductive potential and fertility control, Oxford: Butterworth Heineman, 1996. pp. 39±58. Wheeler MA, Stuss DT, Tulving E. Towards a theory of episodic memory. Psychol. Bull. 1997;121(3):331±354.