Up to 25% of all patients with diabetes mellitus experience significant neuropathic pain.Reference Abbott, Malik, van Ross, Kulkarni and Boulton 1 - Reference Schmader 3 In those with severe diabetic neuropathy, the prevalence of painful symptoms is as high as 60%.Reference Abbott, Malik, van Ross, Kulkarni and Boulton 1 Painful diabetic neuropathy (PDN), described as a burning, stabbing, pricking, or aching sensation, primarily affects the toes, feet, and legs in a distal, symmetric pattern,Reference Dyck, Albers and Andersen 4 , Reference Tesfaye, Boulton and Dyck 5 and often interferes with mobility, sleep, mood, and other domains of quality of lifeReference Schmader 3 and is therefore a cause of considerable morbidity. Although some patients may experience spontaneous improvement or even symptom resolution, PDN is a chronic disease for most patients.Reference Veves, Backonja and Malik 6 , Reference Daousi, Benbow, Woodward and MacFarlane 7
In a meta-analysis of randomized controlled trials, pharmacological management with analgesic antidepressants, including serotonin-norepinephrine reuptake inhibitors and tricyclic antidepressants, anticonvulsants (notably, pregabalin), and opioid analgesics were better than placebo for pain control in those suffering from PDNReference Griebeler and Morey-Vargas 8 —similar to the outcome in other chronic neuropathic pain conditions.Reference Finnerup, Sindrup and Jensen 9 , Reference Attal, Cruccu and Baron 10 Published guidelines, based on systematic reviews, are available from the American Academy of Neurology (AAN) and the European Federation of Neurological Societies to guide clinical management of PDN. The AAN recommends the use of pregabalin as first-line treatment for PDN, noting its established effectiveness with Level A evidence.Reference Bril, England and Franklin 11 In addition to pregabalin, the European Federation of Neurological Societies adds tricyclic antidepressants, gabapentin, and serotonin-norepinephrine reuptake inhibitors (duloxetine, venlafaxine) as potential first-line treatments.Reference Attal, Cruccu and Baron 10 However, the randomized controlled trials that form the basis for these clinical practice guidelines are limited by evidence of efficacy in only short-term trials, generally of ≤3 months’ duration.Reference Griebeler and Morey-Vargas 8 , Reference Bril, England and Franklin 11 Thus, evidence is lacking as to the long-term effectiveness of these medication classes for this chronic disease. Although high-quality randomized controlled studies used to support clinical recommendations have good internal validity, their external validity (real-world outcome) is often left to be ascertained.Reference Rothwell 12 In the case of PDN, adherence to a specific pharmacologic regimen over the long term in this patient population may alter the generalizability of the results; thus, the aim of this study is to determine the long-term clinical effectiveness of the real-world management of PDN at tertiary pain centres.
Methods
The Canadian Neuropathic Pain Database was established in 2008 to provide a registry for neuropathic pain patients seen in academic tertiary care pain clinics in Canada.Reference Moulin, Clark and Gordon 13 From this registry, we identified a cohort of patients with type 1 or type 2 diabetes mellitus to carry out a long-term observational prospective study of the management of PDN. The diagnosis of PDN was established by clinical criteria 14 and supported by the Douleur Neuropathique en 4 Questions questionnaire, which is a reliable discriminator of neuropathic pain.Reference Bouhassira, Attal and Alchaar 15 Patients with comorbid pain of other etiologies were included if they reported that their neuropathic pain was on average more intense and more disabling than their other pains. Informed consent was obtained for every participant before enrolment. Independent review boards representing each participating institution (University of Calgary, Alberta; Western University, McMaster University, University of Toronto and University of Ottawa, Ontario; McGill University, Quebec; Capital District Health Authority Research Ethics Board, Nova Scotia) approved the study.
Standard outcome measures for chronic pain according to Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials guidelinesReference Dworkin, Turk and Farrar 16 were obtained at baseline and at 12 months. The primary outcome measure was the composite of a reduction of ≥30% in average pain intensity and 1-point drop in the Pain Interference Scale of the Brief Pain Inventory (BPI, 0-10) relative to baseline at 12 months. Secondary outcome measures were impact on function (Pain Disability Index), mood (Profile of Mood States), quality of life (12-item short form health survey [SF-12]), catastrophizing (Pain Catastrophizing Scale), and patient satisfaction (Patient Global Satisfaction Scale). Use of prescription medications was recorded both at baseline and then again at 12-month follow-up.
Statistical Analysis
Descriptive statistics, including means and standard deviations for continuous characteristics and frequencies and percentages for categorical characteristics, were calculated for baseline and for 12-month follow-ups. In addition, for the composite outcome, 95% confidence intervals (CIs) are presented. McNemar’s chi-square test for dichotomous values was used to assess the change in the proportion of patients using major classes of analgesics (analgesic antidepressants, anticonvulsants, opioid analgesics) from baseline to the 12-month follow-up. The change in the opioid dose was evaluated using a Wilcoxon signed-rank test. Fisher’s exact test was used to detect any difference in responder rate (achievement of primary outcome) between those being treated with two analgesic classes and those being treated with all three classes. For secondary outcome measures, baseline and 12-month values were compared using paired t tests. Univariable logistic regression was used to evaluate the association between baseline characteristics and the primary 12-month outcome. The p value for the association between opioid treatment and the proportion of subjects achieving and not achieving the primary outcome at 12 months was based on the chi-square test. The p value for the difference in primary outcome based on opioid doses was derived from the Wilcoxon two-sample test.
Results
Of the 789 patients recruited to participate in the Canadian Neuropathic Pain Database, 60 were identified as having PDN. Thirteen (21.7%) were lost to follow-up, leaving 47 patients with evaluable data and 43 patients with complete data at 12-month follow-up. Baseline characteristics of these patients with PDN are presented in Table 1 and patient analgesic history is presented in Table 2. The most common baseline analgesics were anticonvulsants (53.3%), followed by opioids (48.3%).
Patient characteristics at baseline (N=60)
Data are mean±standard deviation, median (Q1,Q3), number; DN4=Douleur Neuropathique en 4 Questions: score ≥4 indicates probable neuropathic pain.
Patient analgesic history (N=60)
MED=morphine equivalent dose (mg/day); NSAIDs, nonsteroidal anti-inflammatory drugs. Data are mean±standard deviation, median (Q1,Q3), number.
At 12-month follow-up, the proportion of patients using analgesic antidepressants, anticonvulsants, or opioid analgesics increased across all categories but did not reach statistical significance (Table 3).
Major analgesic class use at baseline and 12-month follow-up (N=47)
The proportion of subjects achieving a ≥30% reduction in pain at 12 months relative to baseline was 16/43 or 37.2% (95% CI, 23.0-53.3) and the proportion of subjects experiencing a reduction of at least 1 point on the Pain Interference Scale (BPI) was 22/43 or 51.2% (95% CI, 35.5-66.7). The primary outcome measure—that is, the proportion of subjects achieving both at least a 30% reduction in pain and a 1-point reduction on the Pain Interference Scale—was achieved in 13/43 patients or 30.2% (95% CI, 17.2-46.1).
Opioid analgesics were included in the pain management strategy of 27/47 (57.5%) patients at 12-month follow-up. This was an increase from 22/47(46.8%) at baseline, although this increase was not significant (Table 3). Of the 47 patients accounted at 12-month follow-up, 16 were using opioids at both baseline and 12-month follow-up; opioid doses had significantly increased (p=0.002) from baseline to 12-month follow-up in this subgroup, although this was not associated with a significant improvement in the primary outcome measure. Table 4 details the opioid dose of the patients who remained on this treatment modality.
Opioid dose in patients with opioid use at baseline and 12-month follow-up (N=16)
MED=morphine equivalent dose.
Data were complete in 25 patients whose pain management included opioid analgesia at 12-month follow-up. Of the 13 responders achieving the primary outcome, seven (53.9%) had a treatment strategy that included opioid analgesia. In the nonresponder group, 18/30 (60%) were using opioids. Thus, by comparing the proportion of subjects on opioids between responders and nonresponder groups, there was no association between opioid treatment at 12 months and the achievement of the primary outcome (p>0.05). Furthermore, as illustrated in Table 5, no significant difference in opioid dose was found between the responder and nonresponder groups in those using opioid analgesics at 12-month follow-up.
Association between opioid treatment at 12 months and average pain intensity (BPI) reduction of ≥30% and Interference Scale Score reduced by ≥1.0 (n=25)
SD=standard deviation.
Polypharmacy occurred in the majority of patients with PDN by 12-month follow-up. Of the 47 participants, 26 (55.3%) were using at least two medications from the major analgesic classes, and 12 (25.5%) were using all three major analgesic classes (analgesic antidepressants, anticonvulsants, and opioids). There was no difference observed between responder rate (achievement of primary outcome) between those being treated with two analgesic classes and those being treated with all three (p=0.423, Fisher’s exact test).
Cannabinoid use at 12 months was reported in four patients (8.5%). Only a small number of patients sought nonpharmacological treatment modalities. Of the nonpharmacological treatment modalities, the most commonly used by 12 months were physiotherapy and acupuncture, both reported by 10.6% of patients. Other methods included psychotherapy (8.5%), local anesthetic or steroid injections (2.1%), surgery (2.1%), and transcutaneous electrical nerve stimulation (2.1%). All patients were encouraged to increase their level of function despite ongoing pain.
All secondary outcome measures at 12-month follow-up relative to baseline were significantly improved except for Profile of Moods State, SF-12 Mental and Patient Global Satisfaction (Table 6). In particular, there was a significant improvement in mean pain intensity and the mean Interference Scale Score of the BPI at 12-month follow-up relative to baseline. Univariable analysis of associations between baseline characteristics and improvement in pain and function at 12 months were nonsignificant. None of these parameters, including age, pain duration, mean pain intensity, measures of disability (Interference Scale Score of the BPI, Pain Disability Index), mood (Profile of Mood States), quality of life (SF-12), catastrophizing (Pain Catastrophizing Scale), or global satisfaction, predicted outcome. In addition, being on an opioid analgesic at baseline and being on a higher median dose at baseline did not predict outcome (p>0.1 and p>0.4, respectively).
Changes from month 0 to month 12 for secondary outcome measures
BPI (0-10); PDI (0-70), higher score indicates greater disability; POMS–SF (0-120), higher score indicates greater impairment; SF-12 (0-100), score<50 indicates below average health status; Pain Catastrophizing Scale (0-52), higher score indicates greater distress; Patient Global Satisfaction (0-10), higher score indicates greater satisfaction.
Discussion
The chosen primary outcome measure, that is, the proportion of patients who achieved the composite of at least a 30% reduction in average pain intensity on the BPI and a 1-point reduction in the Interference Scale Score (0-10) of the BPI at 12 months, recognizes clinically significant improvement in both pain and function. Almost one-third of patients with PDN treated in tertiary pain centres achieved this clinically significant measure at 12-month follow-up. Although the overall proportion of patients using analgesic antidepressants, anticonvulsants, or opioid analgesics after 12 months was not significantly different from the proportion using these classes at baseline, this improvement may be accounted for by care provided in the tertiary care clinic where patients are exposed to interdisciplinary approaches to pain management including physiotherapy and psychological techniques. In addition, clinicians routinely customize doses of adjuvant analgesics in some patients and switch analgesics in others to optimize analgesia and minimize side effects. More patients had achieved a functional improvement (51.2%) than a diminished pain rating (37.3%). The population studied had a median pain duration of 4 years; thus, these subjects represent those who have progressed to refractory chronic pain and are unlikely to experience spontaneous symptom resolution as has been reported for patients with acute PDN.Reference Daousi, Benbow, Woodward and MacFarlane 7 , Reference Archer, Watkins, Thomas, Sharma and Payan 17 Furthermore, a higher likelihood of unfavourable outcomes was expected secondary to the referral bias introduced by conducting the study at a tertiary care setting, where the patient population consists of those with difficult to treat or refractory pain. A study of PDN management in the primary care setting may represent a different population and more positive results would be anticipated.
Secondary outcome measures showed statistically significant improvement relative to baseline (p<0.05) for pain and function parameters as well as catastrophizing. Catastrophizing is known to be a particularly important predictor of pain expression and activity intolerance in patients with chronic pain.Reference Thibault, Loisel, Durand, Catchlove and Sullivan 18 The lack of association between baseline characteristics and improvement in pain and function at 12 months may be related to the relatively small sample size of the patient population.
Although there is some evidence that chronic neuropathic pain and, in particular, PDN, responds to opioid therapy, this evidence is conflicting.Reference Bril, England and Franklin 11 , Reference Eisenberg, McNicol and Carr 19 There was no significant impact on achievement of the primary outcome measure in the 16 patients who were treated with opioid analgesics throughout the study despite the increased opioid dose. Sizable increases were observed in these 16 patients, with the median morphine equivalent dose increasing from 90 mg/day at baseline to 173.5 mg/day at 12 months. Furthermore, there was no association between opioid treatment at 12 months and the achievement of the primary outcome (p>0.05); that is, patients treated with opioids achieved the primary outcome just as frequently as those without opioid treatment. This is in keeping with evolving basic science mechanisms that suggest that opioid analgesics can actually increase pain in some patients because of paradoxical hyperalgesia.Reference Rivat and Ballantyne 20 A recent AAN position paper strongly cautions against escalating opioid doses above 80 to 120 mg/day morphine equivalent dose in the treatment of chronic non-cancer pain unless the patient has benefitted with regards to pain and function.Reference Franklin 21 Keeping pace with acquired pharmacodynamic opioid tolerance can prompt prescribing higher opioid doses with time. Ensuing pharmacological tolerance may not be overcome with dose escalation.Reference Ballantyne and Shin 22 Furthermore, the significant risks of opioid-related morbidity and mortalityReference Rivat and Ballantyne 20 should prompt the routine use of predetermined endpoints of pain relief and physical function that, if not achieved, would lead to a trial opioid taper to truncate opioid risks when patients are not attaining meaningful benefit. To make an objective determination of meaningful benefit, both pain and function, can be tracked at every visit using brief, validated instruments. The BPI used for this study is one such example.Reference Dworkin, Turk and Farrar 16 , Reference Cleeland and Ryan 23 , Reference Keller, Bann, Dodd, Schein, Mendoza and Cleeland 24
Polypharmacy occurred in the majority of patients with PDN by 12-month follow-up. This requirement for polypharmacy may reflect the limited effectiveness of solitary medications in achieving symptom reduction satisfactory to the patient, particularly in those with chronic PDN who merit referral to a tertiary care setting. However, the disadvantage of this strategy is that of increasing adverse medication effects, such as drowsiness and dizziness. There is some support for polypharmacy in that combinations of morphine and gabapentin,Reference Gilron, Bailey, Tu, Holden, Weaver and Houlden 25 nortriptyline and gabapentin,Reference Gilron, Bailey, Tu, Holden, Jackson and Houlden 26 and morphine and nortriptylineReference Gilron, Tu, Holden, Jackson and DuMerton-Shore 27 provided additive analgesia relative to each agent alone in patients with PDN and postherpetic neuralgia. In this study, although responders were more likely to be using more than one pharmaceutical agent, no difference in response rate was seen between those using two analgesic classes versus three analgesic classes. Thus, it may be possible to achieve similar benefit in the realms of pain reduction and function by limiting the number of pharmaceutical classes used and, in so doing, also limit the hazards of polypharmacy such as compounding adverse effects and medication interactions.
Finally, the authors acknowledge two principal limitations of this observational study: the small number of patients recruited for participation and the 22% dropout rate at 12-month follow-up. Nevertheless, this study helps to clarify the long-term outcomes that are specific to PDN. Overall, our data support the referral of chronic PDN patients to a tertiary care centre to achieve meaningful improvements in pain and function.
Acknowledgements and Funding
This study was funded by Canadian Foundation for Innovation (grant no. 7878) and by Pfizer Canada.
Disclosures
AJC reports grants from Pfizer Canada/Canadian Foundation for Innovation during the conduct of the study and personal fees from Pfizer Canada outside the submitted work. ASG reports grants from Pfizer and CIHR during the conduct of the study and grants from Purdue Pharma and Allergan outside the submitted work. CS reports grants from Pfizer Canada during the conduct of the study. CT reports grants from Pfizer Canada during the conduct of the study. DEM received grant support from Pfizer Canada and speaker’s honoraria and/or consulting fees from Merck-Frosst, Lilly, Johnson and Johnson, and Amgen Canada. The remaining authors do not have anything to disclose.
