Use of CAM and conventional treatment
Directly asking primary care consulters with chronic musculoskeletal pain about how they treated their pain, without indicating in the question that we were interested in any particular class or group or type of treatment and not asking the interviewees to distinguish between conventional or CAM therapy, we found the prevalence of CAM use high. More than four of every five patients interviewed had used at least one CAM treatment, and more than two thirds of CAM users were current users, representing more than half of the study sample.
Using an open question alone, even though clearly explaining that we were interested in every type of treatment used, would still have provided us with incomplete information, had we not used the list which appeared to lead participants to mention more treatments. This could be a simple issue of recall, but could also represent their reluctance to admit to using certain treatments, especially knowing that the interviewer was a doctor. Previous surveys have found a similar effect [
23,
24,
31]. The prevalence of CAM use, in one survey, increased from 56% to 85% when introducing a treatment list [
33].
It has been shown that the larger the number of treatments included on the list, the larger will be the estimates of prevalence.[
6,
34]. It is possible that using such lists could trigger incorrect memory or confuse participants into thinking that they have used treatments that they have not, because of similarities either in names or in actual treatment technique (e.g. acupuncture and acupressure). Such a possibility could inflate the prevalence with inaccurate estimates. This has led some researchers to confine their exploratory work to a small number of treatments in order to obtain precise information [
6]. We were aware of this issue and the trade-off between obtaining accurate information on a small number of treatments and exploring all treatments that were being used for pain. Because we used face-to-face interviews, we used the opportunity to deal with any possible confusion and clarify that the participant had actually used the treatment. Such opportunity would not have been available if other methods had been used e.g. postal survey. Secondly, we asked a number of detailed questions about each treatment reported, and this would have reduced the chance that the participant might have mentioned the treatment by mistake.
The extent of the increase in reporting treatment use in conjunction with the list was twice as high with CAM as with conventional treatments (164/321, 51%
vs 60/235, 25.5%). A possible explanation is that participants might have been less likely to volunteer, to a doctor, information related to CAM compared with conventional treatments, an observation made in previous surveys [
9,
24,
35]. This would correlate with the comments of many participants that they did not realise that we were also interested in these (CAM) treatments. We were not able to find a similar comparison between the effect of using a list on reporting CAM and conventional treatments among the published surveys in the literature.
We were not aware of published surveys that specifically targeted primary care patients in the UK to explore their CAM use for musculoskeletal pain. It is difficult therefore to compare our findings with those from previous surveys in the UK, many of which either targeted patients attending hospital clinics and who had known diagnoses e.g. rheumatoid arthritis [
14,
16,
17,
35,
36], fibromyalgia [
15,
37], multiple sclerosis [
38] or post-spinal cord injury [
39] or targeted individuals in the community [
10,
11] suffering from chronic pain in general and not specifically musculoskeletal pain. Surveys that looked at CAM use in primary care in the UK mainly explored access to CAM and GPs' use and attitude towards it. It was found in a number of these surveys that between 39% and 83% of participating GPs were 'active' with respect to CAM, i.e. practising it, referring for it or endorsing it to their patients [
4,
5,
19,
20,
22,
23]. This could be one possible reason for the high rate of use of these treatments among their patients. We did not ask our participants whether some CAM treatments were practised by conventional health professionals. We know, however, that a third of CAM users in our study said that they came to use CAM because they were referred to it, or it was recommended, by a health professional. For a quarter of CAM users in our study, that was their main reason for using CAM treatments. This obviously refers only to CAM use for chronic musculoskeletal pain and does not include its use for other reasons. This could indicate high 'activity' in relation to CAM in the general practices in the area where we conducted our study, which could explain high CAM use among individuals like our sample of primary care patients. It would be interesting to explore any direct association between GPs activity with regard to CAM and its use among their patients.
In the USA, surveys have explored CAM use among primary care patients [
40‐
42]. However, it was general use of CAM that was explored rather than use linked to a specific condition or symptom. In one survey [
40], it was found that 21% of patients, interviewed while visiting their primary care doctor, had used CAM treatment for the medical problem linked with that GP visit. It is difficult to apply findings related to family medical practice in the USA with primary care in the UK because of the variations in structure, profile and activity.
Surveys among pain sufferers found the prevalence of CAM use ranging from 16% up to 100% in the UK [
11,
12,
14‐
17] and the USA [
10,
35‐
39,
43]. The variation in the prevalence figures is likely to reflect variation in survey methodologies. The majority of these surveys, however, showed the prevalence of CAM use to be consistently higher among pain sufferers compared with other patients.
The two main characteristics that our participants have, namely that they are actively using primary healthcare and that they suffer from chronic musculoskeletal pain, would make them, according to these previous surveys, the more likely users of CAM. This puts the high prevalence figure observed in our study into perspective.
Our study participants were, by selection, users of conventional healthcare, most having made more than two visits to their general practitioner in the previous year. We found that most of this study population were actually using conventional treatment in combination with CAM. This suggests that patients whom GPs saw most frequently for musculoskeletal pain were more likely than not to be using CAM treatments as well. This is consistent with the extensive use of CAM and conventional healthcare services by patients with chronic pain observed in previous surveys [
11,
44].
The high rate of combined use of CAM and conventional treatments could reflect high unmet needs. Surveys have shown that regardless of whether a chronic illness was reported, CAM users tend to report poorer health compared with non-users [
40,
44‐
46]. However, in one survey [
46] it was found that use of CAM was more than twice as common among high users as among low users of medical services in general suggesting that high combined use represents a characteristic of the individuals and is unrelated to their health status or needs.
The high grade of severity of reported pain among our participants would probably be expected with our participants being active users of the health service because of their pain. The positive association between the frequency of GP visits and pain severity has been identified in previous surveys [
11,
47]. 60% of our participants had visited their GP at least twice during the previous year for musculoskeletal pain and 80% were using some form of conventional treatment, mainly prescribed medications. It appears, therefore, that the majority of our patients with chronic musculoskeletal pain who use both CAM and conventional treatments and are visiting their GP (for whatever reason) are still symptomatic and it is possible that this is one reason for their use of both types of treatment. This is important because it increases the relevance of the high use of CAM treatments by chronic musculoskeletal pain patients. It seems to tally with the common perception among GPs of the lack of effective treatments for such symptoms. This would lead some patients at least to try anything that might help with their pain.
It is important to note that by targeting chronic musculoskeletal pain sufferers who are using primary care services, a group of patients with the same pain and who are not using primary care services were not reached by this study. This is a potentially important group of patients, some of whom might be exclusively using CAM for pain. We cannot comment in this study on the use of CAM among such group and our findings remain only applicable to primary care consulters.
Socio-demographic characteristics of treatment users
The typical socio-demographic characteristics of the majority of CAM users in our study did not echo those from the majority of studies in which CAM users were found more likely to be women [
11,
48] from higher social class groupings. [
9,
11,
44,
46,
48] There are, however, studies which did not find a link between higher CAM use and higher income [
40] or any significant difference by gender among CAM users [
5,
35,
48].
The reason for the contrast between our results and other studies' might lie in regional variations in CAM use [
46,
49] or might represent patterns specific to sufferers of chronic musculoskeletal pain.
It has been suggested that the observed regional variation in CAM use more likely to reflect variation in access and availability than regional differences in public attitude and interest [
50]. Access to these treatments can be severely restricted, with 90% of CAM provided in the private sector [
44], leading to the suggestion that its use is related to the affluence of the area [
4]. Surveys have shown that CAM use in the south west of England, for example, was higher than the national average (16%
vs 10%) [
22].
Geographical variation in the availability and provision of CAM has been suggested as another possible explanation for variation in use [
4,
11,
50]. One factor that was shown to influence CAM availability is the nature of local conventional healthcare services and primary care in particular (i.e. practices' attitude towards CAM and its provision; GP's special interest in CAM or antipathy towards it) [
22]. GP endorsement of these treatments varied between areas (38% in Liverpool area
vs 54% in the south west of England) [
22,
23] as well as their active involvement i.e. practicing CAM [
4,
22]. Variation in demand could also influence availability of CAM. It has been shown that the prevalence of chronic pain, one of the most common health problems for which CAM is used, varied widely across geographical areas. [
51]
It is interesting to attempt to explain our finding of the higher use of CAM and conventional treatments combined among women compared with men. It has been shown in one survey at least, that women were more likely than men to report chronic pain with no difference between genders in the reported severity of pain.[
47,
52] which could arguably offer an explanation. Women were also more likely to report high expressed needs than men [
47]. In another study, where use of healthcare services was explored, women were found more likely to have used prescription and non-prescription medications, alternative therapist and alternative medication [
11]. This could suggest that the high use among women, compared with men, of both conventional and unconventional medicine for pain, is related to their higher expressed needs and not to the severity of the reported pain.
Perceived helpfulness from using CAM
Attempts to assess this have been made in past surveys from information mainly based on doctors' reports of their patients' benefit from using CAM [
23,
48,
19,
53]. However doctors' knowledge of their patients' use of CAM is often very limited [
9,
24] and the views of doctors and patients on the usefulness of CAM may differ [
40].
There has been much recent debate about the lack of available evidence regarding the efficacy of CAM treatments. CAM treatments in our study were generally found to be helpful by participants, echoing previous findings from one systematic review [
24]. This might represent what is called as the effectiveness gap [
56], although in a reversed way. The effectiveness gap is said to exist when a treatment is shown to have an effect based on its pharmacological action but shows a smaller effectiveness in clinical practice. Here, the gap seems to exist when treatments (such as some CAM treatments) are perceived to be helpful by users when no evidence for their effect exists.
Although the number of participants who reported experiencing harm in the form of worsening pain symptoms following the use of some CAM treatments was small and although these data do not represent an objective measure of effectiveness, one conclusion is that, although beneficial effects on pain from each CAM treatment are commonly reported, many users do not perceive CAM to be automatically beneficial, and a number of them (substantial if extrapolated nationally) considered themselves to have experienced harmful effects.
The range of scores for perceived helpfulness from the commonly used conventional treatments was wider than for CAM treatments, and there were higher harm scores. The latter might be balanced or off-set by evidence of effectiveness the likes of which is lacking for many CAM treatments. Interestingly, some of the favourable CAM treatments, such as chondroitin sulphate and osteopathy, had higher average ratings for perceived helpfulness than paracetamol, ibuprofen and co-codamol. The differences in the numbers of users, however, make accurate comparison difficult beyond mathematical extrapolation. This issue merits further investigation.
An important finding in our study was the instances where participants reported harm attributed to the use of treatments. The eight instances of harm attributed to the use of seven CAM treatments represent a small percentage of the total number of instances on which CAM treatments were used. These seven CAM treatments had been used 52 times in this study's population.
Some observations could be made on these harm reports. Firstly, these harm scores were reported for some treatments that also received high perceived helpfulness scores from other users. Electrotherapy received nine positive scores (+2 to +8), massage received 14 positive scores (+1 to +10) and acupuncture received 5 positive scores (+3 to +9). Secondly, although the question was about perceived helpfulness in relation to pain, we believe that reported negative scores might not have always meant "worsening of pain following using the treatment" but might also meant other adverse effects which may not be related to pain. We did not expand on the nature of the harmful effect that was reported and this information was collected as a score on the negative arm of (-10) to (+10) VAS. Thirdly we do not know whether these effects were reliably caused by these treatments. The answers were purely subjective.
It is interesting to compare the number of these reported harmful incidents with the number of participants who said that they have stopped CAM because it caused them problems or side effects. These reasons for stopping were given on fewer occasions of CAM use (six) than reported harmful events (eight) and not all these cases are the same. This could either reflect the unreliability of the assessment made by the participants, or that some of them did report harm which was not of a type or severity that had made them stop the treatment. Finally, it seems that reporting harm was more likely to be related to practitioner dependent treatments, although the number of instances was too small to validate this conclusion.
Although the number of reported perceived harm instances is small, they are nevertheless important. They highlight the fact that these treatments are not universally experienced or perceived as harmless. They are also important in the debate about the safety of CAM and its integration within the mainstream health services.
Harmful events attributed to CAM use have been reported previously. 38% of GPs in one survey reported adverse effects related to CAM use by their patients [
22]. In another survey, 21% of responding GPs reported similar harmful effects [
23]. A survey in Australia found that 25% of users of naturopathy reported effects [
55]. The adverse events reported in the surveys studied in a systematic review of the use of CAM in rheumatology were low [
26]. In addition to users' views and perceptions on harm, doctors, on the other hand, have a different view of the harm they perceive and attribute to using CAM. In one survey 62% of the participating physicians suggested that CAM use prevents patients from getting proper treatment [
21].
The issue of perceived helpfulness is important, with implications for safety, integration and future research. With the increasing use of CAM and the increasing amount of anecdotal evidence for its helpfulness, or otherwise, by users, some are suggesting that there should be room for debate as to who decides what is and what is not effective and on what basis, at least in the NHS [
12].
Response and generalisability
A limitation to the study was the higher than anticipated reluctance to be interviewed, which meant that we had a final study population of 138 as opposed to the 180 pre-specified in the sample size pre-requisite. We had also underestimated the amount of CAM use. Revisiting the power calculation post hoc and taking a 20% difference in CAM use based on a greater base value of 80% CAM use in the study population, meant that given a sample size of 138 we had 79% power of detecting this difference if it existed i.e. there was little loss in power compared to the prior calculation.
We took advantage of the fact that our patients had completed earlier postal surveys to compare responders and non-responders with respect to gender, age, pain and health status scores and health care use in general, as well as use of specific 'home remedies' (e.g. cod liver oil) which had been enquired about in the postal questionnaires. Differences were small and it is unlikely that those interviewed represent an unusual sample of our target population with respect to their general experience of pain and willingness to use a variety of treatments. Furthermore CAM was not mentioned or referred to during our study, and so responders are unlikely to represent a group specifically interested in this topic.
Generalising our study findings to the wider population of all patients with chronic musculoskeletal pain who are using primary healthcare services in the UK would require caution. CAM use varies between different parts of the country [
23,
19], and this may influence use among consulters also. This variation might explain why the use of some individual CAM treatments, such as Homeopathy, was lower among our participants compared with other surveys' [
6]. However it seems unlikely that the broad patterns identified here would differ substantially in other primary care settings.