Design
This was a longitudinal study assessing participants’ pain type and severity at baseline assessment and over a follow up period of 3 years.

Participants and Procedure
The sample population for this study consisted of 188 men and women with a diagnosis of MS since 1995 based on McDonald Criteria, registered at the three major MS clinics in the Greater Montreal, including Montreal Neurological Hospital (MNH), Centre Hospitalier de l’Universite´ de Montreal (CHUM), and Clinique Neuro Rive-Sud (CNRS). The reason for including patients diagnosed since 1995, was based on MS natural history, severity of symptoms and progression of disability. Since 1995, advances in neuro-imaging techniques and disease modifying therapies (DMTs) have facilitated earlier diagnosis of disease and reduce the speed of disease progression (16).  Therefore, the recognition of this era is an important methodological requirement in designing studies in this field as those people with MS diagnosed after 1995 likely not follow the same course as people diagnosed with MS prior to 1995 (16).
From those 188 individuals, 78 who reported pain were included as study participants. If participants experienced a relapse in the previous month of study and if they were younger than 18 years old, they were excluded from the study. In addition, participants with severe cognitive impairments, and those with health conditions that could impact function, such as mental illness, heart disease, rheumatoid arthritis, malignancy, renal failure, HIV/Aids, or liver failure were excluded from participating in the study as well. In order to follow-up the pain experience of this sample, after three years of first assessment, participants reporting pain at the first assessment (n=78) were re-contacted to answer and complete a questionnaire on pain severity and type; 56 persons agreed to be re-interviewed.

Measurement
Socio-demographic and disease-related variables (e.g. MS type, years since MS diagnosis and symptoms onset) were collected. The severity of neurological impairment was assessed by the treating neurologists and reported according to the Expanded Disability Status Scale (EDSS) (17).

Pain type
At baseline, participants were asked to choose as many as of the words that applied to them from a list containing the most common pain sensation descriptors taken from the McGill Pain Questionnaire (MPQ) (18). Sensations such as tingling, numbness, shooting, stabbing, electric shock-like, and burning are considered as NP descriptors, while other particular sensory adjectives such as throbbing, gnawing, or grating are significantly considered as N-NP descriptors (19, 20). The ID-Pain Questionnaire (ID-Pain), which is designed for accurately detecting NP as differentiated from N-NP pain, was used to assess type of pain at follow up (21). The psychometric properties of ID-Pain have been reported (21).

Pain severity
Zero to 10 Numeric Rating Scale (NRS)  was applied to assess  the lowest, and worst pain severity over the last week  and  pain at the time of  assessment (22) The reliability and validity of NRS have been reported (22, 23, 24). NRS was completed at two points in time (baseline and follow up) to estimate change in pain severity.

Statistical  methods
Descriptive statistics were used to characterize the sample. To determine the stability of the proportion of NP pain type over time, we compared the overall agreement on the classification of NP based on the results of MPQ and ID-Pain using the McNemar’s test (25). To measure the agreement between within-person recorded changes in pain type over time, Cohen’s un-weighted Kappa Coefficient was calculated (26). Paired Student t- tests was used to examine if serial ratings of NRS were stable or changed over time. Longitudinal analysis using generalized estimating equations (GEE) was performed to test change in lowest and worst pain severity between the different pain type groups at first assessment and follow up (27). GEE models are extensions of generalized linear models, where outcomes for each record are not assumed to be independent (27). Participants were categorized into five groups according to the presence of NP including the reference group, the group with participants who had NP at any time point, group of participants who lost their NP at follow up, group of participants who developed NP at follow up, and group of participants who had NP at both time points. The reference group was study participants who did not have NP at either time point. Statistical analyses were performed using the Statistical Analysis Systems (SAS) Version 9.2.

 

Of the 78 participants who reported pain in the original study, 56 persons (72%) agreed to participate in the second phase of the study. Socio-demographic and clinical characteristics of participants and non-participants are presented in Table 1.
Table 2 summarizes pain characteristics of the participants. Overall, the result of McNemar’s test was not significant (p value= 0.44), indicating that there was no difference between type of pain over time (Table 2). Results of Kappa test also indicated that there was an agreement between MPQ and ID-Pain (Kappa = 0.50).
The participants’ descriptions of their abnormal sensations using the words from the MPQ and ID-Pain are detailed in Table 2. The words most frequently used were numbness, tingling, burning and shooting, indicating that NP was the most reported type of pain. As it is presented in Table 2, there was no significant difference in the proportions of specific NP quality descriptors from MPQ and ID-Pain at first assessment and follow up.
The mean values of participants’ responses to the global pain severity scales are also summarized in Table 2. Results of paired t-test on serial measure of NRS showed that on average all ratings have been increased, though only for lowest pain this difference was statistically significant (p=0.0006).
Table 3 presents the comparison of pain severity measured by NRS across different patterns of pain type over time. Interestingly, a descriptive analysis of the NRS scores revealed discrepancies between change in pain type and the numerical change in pain severity scores. In comparison to reference group (participants who did not have NP at any time point), participants who had NP at any time point showed the highest increases in the current, lowest, and worst pain severity scores.   
Results of a GEE analysis suggested that pain type was a significant predictor of lowest pain severity scores over the follow-up period (P<0.05). Participants with NP at either time point had 1.5 points higher amount of lowest pain intensity score than those without NP (CI 95%: 0.68-2.3). In contrast, when modeled, type of pain did not emerge as a significant predictor of worst pain rating (Table 4).
 

Table 1. Characteristics of the sample at the first assessment



Table 2. summarizes pain characteristics of the participants.


Table 3.Comparison of pain severity measured by NRS across different groups of pain type (N=56)  


Table 4. Light field microscopy of the inhibitory effect of fluconazole combined with amphotericin B on fluconazole-resistant C. albicans (ATCC 10231) biofilm formation at different concentrations. Magnification × 40, Bar = 50 µm. Flu: fluconazole, AmB: amphotericin B.

Discussion

This study assessed the change in pain type and severity over a follow up period of 3 years among a sample of individuals with MS. Consistent with previous studies, (3, 4, 8) results of serial measures of NRS in this study showed that, on average, all ratings of pain severity increased, however, only lowest pain showed the significant increase. Considering the contribution of pain to problems with function in people with MS, these results emphasis priority of early treatment and assessment of pain as it may get worse over time. The group-based analysis further showed that pain type was stable in the majority of study participants. This is in agreement with the results of previous study (11) suggesting that type of pain is associated with MS duration or MS progression.
Consistent with previous studies, (10, 11) the results of our work confirm that NP is more common than N-NP in MS population. This emphasizes the need for adequate pain investigation, something that is still a challenging task to researchers and clinicians as there are no universally accepted diagnostic criteria for NP (23). Although, using the questionnaires have the advantage as they are fast and easy to administer, we acknowledge that a more accurate diagnosis needs to be based on patient history, and physical examination (10).
Using GEE analysis, we found that everyone who had NP at any time point had their lowest pain severity significantly higher than those who never had NP, while worst pain severity was not affected by type of pain. Having high score of lowest pain intensity is likely to be debilitating as the person is always in a state of pain.
This study had several strengths. It assessed change over time in pain type and severity which is often not assessed in MS pain literature. The present study was also the first to evaluate the predictive influence of neuropathic type of pain on pain rating over a period of three years in people with MS. Moreover, the participants in  this study were randomly recruited from three MS clinics in different areas of the greater Montreal area, thus they were culturally diverse in MS severity, type, and gender  and so a good representative of the Canadian MS population. As with any research, this study has limitations that should be considered when interpreting the results. First, although results of previous studies support the idea that using pain verbal descriptors can be the easiest and most reliable way to discriminate NP from N-NP, (28-30) it should be acknowledged that bias may have been introduced due to overlap in descriptors of NP and N-NP. Second as only patients diagnosed with MS since 1995 were included in the study, this may underrepresent symptoms in study sample. Third, because of small sample size of study there would have been little power to find strong conclusion. Nevertheless, the combination of a robust statistical analysis (i.e. GEE) and the innovative approach to dealing with small sample sizes (i.e. person-specific analysis) we were able to answer the study’s questions.
Findings from the current study provide useful information to help health care professionals, clinicians, and researchers to a have a better understanding about NP in MS. In the clinical setting, distinguishing type of pain is essential for its treatment, since each type of pain needs specific treatment approaches according to its underlying mechanism (31). In addition, as NP is often a symptom over which clinicians cannot have significant control, maintaining an awareness of a patient's pain experience, and observing changes in serial pain measurements, increase our insight about the nature of pain type in MS and can give both patients and clinicians a greater sense of control. Considering that NP is distressing and impacts on functioning, (10, 11) this would give significant physical and psychological benefits, and so improving HRQL (12).
As found in other studies (21, 32), results of the present study confirm that a single measure of pain severity such as NRS alone, is an inadequate assessment tool as it may not sufficiently represent the construct of pain. Two individuals, who report same amount of pain severity on an NRS, may indicate different pain sensations. This finding has clinical implications. For treatment to be successful, measuring other aspects of pain using valid and reliable measures is necessary.
By tracking MS pain fluctuations over time the dynamic nature of pain can be examined, a feature that has been relatively ignored in the MS pain literature mostly due to the unpredictability of the disease course, as well as the interaction between pain and other symptoms and contextual factors. A biopsychosocial model of pain should be considered in any assessment of pain treatment efficacy (32). This would subsequently improve the development of more effective, comprehensive treatment efforts directed toward enabling individuals with MS to maintain their active life.

 

Conclusion

The present study examined stability in pain type and severity among individuals with MS. While pain severity increased with time, pain type was stable. The findings of this study have practical applications for chronic pain management programs. Observing changes in serial pain measurements increases our understanding about the nature of pain in MS.

 

Acknowledgements

Authors acknowledge the contribution of Susan C. Scott who helped for statistical analyses and Élaine Roger, Ayse Kuspinar, and Shang Yuan Teng, who recruited patients from CHUM and MNH, respectively.

 

Conflicts of Interest

Authors declared no conflict of interests.