Consideration Of Chronic Pain In Trials To Promote Physical Activity For Diabetes: A Systematic Review Of Randomized Controlled Trials

Background Chronic pain has been estimated to affect 60% of patients with diabetes and is strongly associated with reduced activity tolerance. We systematically reviewed randomized controlled trials (RCTs) that explored interventions to improve physical activity among patients with diabetes to establish whether co-morbid chronic pain was captured at baseline or explored as an effect modifier and if trials reported a component designed to target chronic pain. Methodology/principal Findings We searched CINAHL, Cochrane Central Registry of Controlled Trials, EMBASE, ERIC, MEDLINE, SPORTDiscus and PsycInfo from inception of each database to March 2012 for RCTs that enrolled patients with diabetes and randomly assigned them to an intervention designed to promote physical activity. Two reviewers independently selected trials and abstracted data. We identified 136 trials meeting our inclusion criteria, only one of which that reported capturing chronic pain measures at baseline. No trial reported on specific interventions to address chronic pain as a competing demand, or as an effect modifier. Conclusion/significance Only 1 trial identified that aimed to promote physical activity among patients with diabetes reported that co-morbid chronic pain was captured at baseline. No trials reported exploring chronic pain as an effect modifier or targeting it as part of its intervention.


Introduction
Physical activity is considered a cornerstone of diabetes management, along with diet and medication [1]. The benefits of physical activity are well established for patients with diabetes, especially when combined with diet [2], and include increased cardiorespiratory fitness, increased vigour, improved glycemic control, improved lipid profile and maintenance of weight loss [3]. However, chronic pain has been identified as a barrier to physical activity.
Krein et al. conducted a cross-sectional survey of 993 adult patients with diabetes (94% response rate) and found that 60% of respondents reported experiencing pain not due to cancer that was present most of the time for 6 or more months during the past year (henceforth called chronic pain). The most commonly reported pain locations involved the back (60%) and hip or knee (60%). Patients reporting chronic pain were more likely to be using insulin and to present with a higher body mass index. The presence of chronic pain was strongly associated with difficulty exercising even after adjusting for the presence of depressive symptoms, general health status, other co-morbid conditions, and priority given to diabetes care (adjusted odds ratio [OR] = 3.0; 95% confidence interval [CI] = 2.1 to 4.1) [4].
Chronic pain affecting exercise participation for type II diabetes was also described qualitatively, in patient interviews, by Casey et al., ''I have difficulty… you know my knees… I have pain 24/7 in my arms and legs, in my lower back'' [5]. In another study of diabetic patients, Lawton and colleagues reported numerous accounts of painful knees, joints and swollen feet. For example: ''They tell you to exercise and I exercise a little, but I can't move around a lot because I have problem with my leg [arthritis]. If I walk a little then it swells up'' [6]. These forms physical pain and arthritis are frequently reported as barriers to programs of physical activity for type II diabetes [7,8]. Lastly, two recent reviews reported both an association between neuropathic pain and impaired physical functioning [9], and that 2548 diabetic neuropathy patients across 7 identified studies have a significantly lower health-related quality of life compared to the general population (mean pooled health utility score = 0.61; 95% CI = 0.56-0.66) [10].
These findings emphasize the importance of considering chronic pain when exploring interventions targeted at increasing physical activity for patients with chronic diseases such as diabetes. We conducted a systematic review of randomized controlled trials (RCTs) to establish whether interventions aimed at increasing physical activity for patients with diabetes capture co-morbid chronic pain at baseline, if chronic pain was explored as an effect modifier, and if study interventions specifically targeted co-morbid chronic pain as a competing demand.

Methods
Two reviewers (JJR, JWB) formulated a search strategy to identify relevant randomized controlled trials (RCTs), in English, by a systematic search of CINAHL, Cochrane Central Registry of Controlled Trials, EMBASE, ERIC, MEDLINE, SPORTDiscus and PsycInfo from inception of each database up to March 2012 (see Search Strategy S1). The search strategy combined terms for RCT, diabetes, and physical activity. The terms included free text words and subject headings specific to each database. Reviewers scanned the bibliographies of all retrieved trials and other relevant publications, including reviews and meta-analyses, for additional eligible articles.
Two teams of reviewers screened (JJR, DJB, MA, MI), independently and in duplicate, titles and abstracts of identified citations and retrieved the full text publication of articles that both reviewers judged potentially eligible. Three teams of reviewers (JJW, AHYC, RAL, MA, MA, MI) independently applied eligibility criteria to the methods section of potentially eligible trials. Eligible trials met the following criteria: (1) random allocation of patients to an intervention, including pilot studies, designed to increase physical activity or a control, and (2) inclusion of patients with type I or II diabetes. Since exercise is an effective treatment for chronic pain, any trials that incorporated primarily supervised physical activity in the domains of strength, flexibility or aerobic capacity as part of the intervention were excluded. Any disagreements were resolved by discussion to achieve consensus.
Two teams of reviewers (JJR, JJW, MA, MI) extracted data independently and in duplicate from each eligible study. Data abstracted included demographic information, methodology, intervention details (including if interventions contained components specifically targeting chronic pain), and whether co-morbid chronic pain was captured at baseline or explored as an effect modifier -either by conducting an adjusted analysis (adjusted for the presence of chronic pain, treatment group and an interaction term [chronic pain 6treatment group]) or a subgroup analysis. To ensure consistency across reviewers we carried out calibration exercises before starting the review.
We assumed that consideration of chronic pain in trials to increase physical activity among patients with diabetes would be associated with: (1) starting enrolment of patients after the survey by Krein et al. was published (January 2005) [4], and (2) publication in higher impact journals, as we have recently shown that subgroup analyses are more common in high impact journals (adjusted OR = 2.64, 95% CI = 1.62 to 4.33) [11]. We reviewed the methods section of each trial, and trial registries when reported, to acquire the date that investigators began enrolling patients. We used the Institute of Scientific Information's Journal Citation Reports to obtain the 2011 impact factor for each journal in which an eligible RCT appeared. We did not register the protocol for our review.

Results
We identified 13,925 potentially eligible studies, and retrieved 419 studies in full text; 136 proved eligible for our review (see Figure 1). The chance-adjusted between-reviewer agreement (estimated kappa) on title and abstract screening was 0.74, and 0.90 for full text eligibility. Most trials (116 of 136) enrolled patients with type II diabetes, with 9 trials enrolling patients with type I diabetes, and 11 trials enrolling mixed populations.
The majority of eligible trials (101 of 136) were published in 2006 or later; however, the date that patient enrolment began was only available for 53 of 136 trials (39%). The 2011 impact factor for listed journals ranged from 0.427 (Australian Journal of Advanced Nursing) to 11.462 (Archives of Internal Medicine).
Only 1 of the 136 eligible trials (see References Included in Review S1) recorded co-morbid chronic pain at baseline; Amoako et al. recorded that 15 of 68 participants (22%) reported chronic pain at enrolment [12]. The investigators did not conduct a statistical analysis to formally explore the effect of chronic pain on therapy, but did report that pain and stiffness were the most common reasons for patients not participating in physical activity. No trial included a component of their intervention that was explicitly directed at addressing co-morbid chronic pain. However, 28 trials did report pain-related measurements in general rather than ones specific to chronic pain, either through pain-specific measurement or as part of a general health-related quality of life instrument. (see Table 1).  Enhanced usual care from primary care physician addressing depression, diabetes and/or coronary heart disease, n = 108 Overall quality of life score at baseline and postintervention   Table 2). As only 1 eligible trial specifically considered chronic pain, we did not conduct our planned analyses.

Discussion
Our systematic review of RCTs aimed at improving physical activity among patients with diabetes found that trials did not specifically capture co-morbid chronic pain at baseline, nor explore the impact of chronic pain as an effect modifier. Also, interventions directed at improving physical activity among patients with diabetes did not contain components specifically targeting co-morbid chronic pain as a competing demand.
Our study does have some limitations. We only considered English-language trials; however only 2 non-English trials were identified in our screening of abstracts, which suggests their exclusion, had little if any impact on our findings. As well, trials may have indeed considered aspects of chronic pain, but not reported this information. Our findings are strengthened by our comprehensive search and broad eligibility criteria, use of standardised screening and data extraction forms, and calibration exercises to enhance the consistency between reviewers.
Following the survey by Krein et al. that found 60% of patients with diabetes experience co-morbid chronic pain and suggested chronic pain should be considered as a competing demand in selfcare regimes [4], Butchhart and colleagues surveyed 624 primary care patients, including 221 with diabetes (77% response rate) and found that 74% of patients with diabetes reported taking pain medications and 34% had attended a pain clinic in the last year for treatment [13]. Common locations of pain in the sample were almost identical to the findings of Krein et al.: upper back or lower back (60%), hips or knees (52%), and feet (43%). The majority of respondents (57%) managed their chronic pain with rest and reducing their level of activity and the authors noted: ''It is possible that patients are not aware of the role exercise can play in pain management''.
Effective pharmacological and non-pharmacological treatments for non-specific chronic pain exist [14,15]. Back pain was the most commonly reported chronic pain complaint in the survey by Krein et al. and the American Pain Society has recently reviewed the literature and found some evidence for tricyclic antidepressants, cognitive-behavioural therapy, exercise, spinal manipulation and interdisciplinary rehabilitation for chronic low back pain [16,17]. Specific to diabetic neuropathic pain, a recent review identified that pregabalin and duloexitine medications are effective treatment options [18]. However, our review found that these strategies do not appear to be considered in trials exploring strategies to increase physical activity among patients with diabetes. Chronic pain appears to be a common complaint among patients with diabetes and is associated with reduced tolerance for physical activity; however, current trials of interventions to improve physical activity in this population do not appear to consider the effect of chronic pain. When exploring interventions to improve physical activity among patients with diabetes, trialists may wish to consider baseline chronic pain and explore the impact of chronic pain as an effect modifier. Consideration could also be given to incorporating strategies within trials promoting physical activity to address complaints of co-morbid chronic pain.

Supporting Information
Checklist S1 PRISMA Checklist.

(DOC)
References Included in Review S1.