Skip to main content

Establishing clinically meaningful within-individual improvement thresholds for eight patient-reported outcome measures in people with relapsing-remitting multiple sclerosis



As disease-modifying therapies do not reverse the course of multiple sclerosis (MS), assessment of therapeutic success involves documenting patient-reported outcomes (PROs) concerning health-related quality of life, disease and treatment-related symptoms, and the impact of symptoms on function. Interpreting PRO data involves going beyond statistical significance to calculate within-patient meaningful change scores. These thresholds are needed for each PRO in order to fully interpret the PRO data. This analysis of PRO data from the PROMiS AUBAGIO study, which utilized 8 PRO instruments in teriflunomide-treated relapsing-remitting MS (RRMS) patients, was designed to estimate clinically meaningful within-individual improvement thresholds in the same manner, for 8 PRO instruments.


The analytical approach followed a triangulation exercise that considered results from anchor- and distribution-based methods and graphical representations of empirical cumulative distribution functions in PRO scores in groups defined by anchor variables. Data from 8 PRO instruments (MSIS-29 v2, FSMC, MSPS, MSNQ, TSQM v1.4, PDDS, HRPQ-MS v2, and HADS) were assessed from 434 RRMS patients. For MSIS-29 v2, FSMC, MSPS, and MSNQ total scores, available anchor variables enabled both anchor- and distribution-based methods to be applied. For instruments with no appropriate anchor available, distribution-based methods were applied. A recommended value for meaningful within-individual improvement was defined by comparing mean change in PRO scores between participants showing improvement of one or two categories in the anchor variable or those showing no change. A “lower bound” estimate was calculated using distribution-based methods. An improvement greater than the lower-bound estimate was considered “clinically meaningful”.


This analysis produced estimates for assessing meaningful within-individual improvements for 8 PRO instruments used in MS studies. These estimates should be useful for interpreting scores and communicating study results and should facilitate decision-making by regulatory and healthcare authorities where these 8 PROs are commonly employed.


Multiple sclerosis (MS) is a chronic immune-mediated disease affecting the central nervous system [1, 2]. It has been estimated that 2.8 million people had MS in 2020, equivalent to ~ 35.9 per 100,000 population, which has risen by 30%, globally, since 2013 [3], with females and young adults disproportionately affected [3, 4]. As such, MS is the leading cause of nontraumatic neurological disability in young adults, with a mean age at diagnosis (globally) of 32 years [3, 5]. Being a life-long disease, effective disease monitoring and management are critical ideally from symptom onset. The cornerstone of MS management is with disease-modifying therapies (DMTs) [6], which decrease frequency, duration, and severity of relapses in people with relapsing forms of MS (relapsing-remitting MS [RRMS]). Certain DMTs can also reduce disease progression [7, 8]; however, there is no treatment that can reverse the disease course of MS [5, 9]. Therefore, treatment also focuses on improving or maintaining health-related quality of life (HRQOL), minimizing the impact of disability and side-effects of treatment, and maximizing wellness [10]. Consequently, patient-reported outcomes (PROs), where patients self-report and describe their perceived health status, function, and/or experiences, are especially relevant in MS, as they provide individualized perspectives about health experience and treatment outcomes [10] during periods of worsening or stable disease [11]. Indeed, PROs are recognized as having an increasingly important role in MS clinical research and within real-world clinical practice [12, 13].

Both MS-specific and generic PRO instruments are beneficial in evaluating HRQOL in people with MS and in assessing the impact of MS on function; hence, a combined approach using both generic and MS-specific PROs is often recommended [10]. Given their importance, interpretation of PRO scores is critical to enable communication about PRO results to regulators, patients, physicians, and healthcare providers alike and to inform decision-making by regulatory boards and healthcare authorities [14]. However, there is currently little standardization in how PRO measures are scored or presented, and there is some confusion reported regarding the thresholds used to assess clinical significance [10]. The number of PRO measures used in MS trials makes comparison difficult [12]. Interpretation and communication of PRO results is challenging given that many MS PROs are routinely used in clinical trials but they lack established meaningful within-patient change thresholds, which are needed to interpret responder and time-to-event analyses [11]. Without commonly derived thresholds in place across a range of frequently used instruments, application of PROs for the decision-making process for regulatory or reimbursement agency reviews for MS medications could limit their utility, just when their importance is increasingly recognized [15, 16]. Consequently, further information is needed to help validate the use of multiple PROs within the MS healthcare arena and to summarize clinical meaningfulness from these different assessments.

The Prospective, Non-Interventional Trial to Analyze Patient-Reported Outcomes in MS Patients Treated with Teriflunomide (AUBAGIO®) in Routine Clinical Practice (PROMiS AUBAGIO study [TERIFL07766]) was a prospective, noninterventional study in US participants with RRMS treated with oral teriflunomide in routine clinical practice. Eight PROs were utilized in the PROMiS AUBAGIO study to evaluate the impact of RRMS on HRQOL, fatigue, functional performance, work capacity, daily activities, cognitive impairment, anxiety and depression, and satisfaction with treatment. This analysis of PROMiS AUBAGIO data was designed to estimate meaningful within-individual improvement thresholds for eight PRO instruments used in this study. It was hoped that this would enable better interpretation of improvement scores for these eight PRO instruments in past, current, and future MS studies.



The PROMiS AUBAGIO study included participants aged ≥ 18 years with RRMS treated with oral teriflunomide in routine clinical practice in the US. The primary objective was to describe changes in HRQOL in RRMS participants who initiated treatment with teriflunomide, using scores from eight MS-specific and generic PRO instruments. All participants provided written, informed consent prior to entry into the study. It was planned that 740 patients would be recruited from the MS ONE-TO-ONE program such that, accounting for attrition, ~ 500 participants would complete PRO assessments at Month 12. This analysis was based on data from 434 RRMS participants who signed the electronic informed consent, had ≥ 1-day study medication exposure, and baseline and ≥ 1 post-baseline PRO assessment. This cohort is defined as the full-analysis set (FAS).

PRO assessments

A total of eight unique PRO instruments were administered to participants during the PROMiS AUBAGIO study; these instruments cover a variety of MS-specific symptoms, including HRQOL, fatigue, functional performance, work capacity, daily activities, cognitive impairment, anxiety and depression, and satisfaction with treatment. A summary of PRO scores and schedule of assessments is given in Table 1. All data were collected electronically through the online portal at baseline and subsequent months depending on the instrument (Table 1). All data were directly entered by study participants. All participant health information was encrypted. All PROs were selected based on their context of use and ability to capture the concepts most relevant to patients/caregivers in the RRMS population. Each PRO had evidence in the literature for its reliability, as outlined below.

Table 1 Patient-reported outcome instruments administered during the PROMiS AUBAGIO study

Multiple Sclerosis Impact Scale 29 items version 2 (MSIS-29 v2)

The MSIS-29 v2 is a PRO instrument developed to evaluate specific physical and psychological impact of MS from the patient’s perspective; the instrument comprises 29 items grouped into physical impact (20 items) and psychological impact (9 items) scores [17, 18]. Participants respond to each item regarding the condition’s impact on their daily life during the last two weeks and select an answer on a five-point Likert scale that strongly represented their status. The patient’s scores on the two subscales can be summed and converted to a measure between 0 and 100, where higher scores indicate a greater physical and psychological disease impact (worse health status [17]). The current study used each scale separately. Test–retest reliability has been shown to be high (r = 0.65–0.90) [17]. The administration schedule for the scale is given in Table 1.

Fatigue Scale for Motor and Cognitive Functions (FSMC)

The FSMC was developed to evaluate MS-related motor and cognitive fatigue, comprising 20 items that are grouped into a physical fatigue score (10 items), cognitive fatigue score (10 items), and combined total score (20 items) [19]. Each dimension score ranges from 10 to 50, and the total score from 20 to 100, with a higher total score indicating more severe fatigue. Internal consistency (Cronbach’s alpha > 0.91) as well as test–retest reliability (r > 0.80) has been shown to be high [19]. The administration schedule for the scale is given in Table 1.

Multiple Sclerosis Performance Scale (MSPS)

The MSPS was developed to evaluate MS-associated disability overall and in different dimensions and comprises 8 performance scales (mobility, hand function, vision, fatigue, cognition, bowel/bladder, sensory, spasticity), and a total score [20]. Each individual scale ranges from 0 (normal) to 5 (total disability), and the total score ranges from 0 (normal) to 40 (total disability), with higher individual and total scores indicating more disability. Spearman’s rank correlations assessing convergent construct validity range between 0.55 and 0.78 [20]. The administration schedule for the scale is given in Table 1.

Multiple Sclerosis Neuropsychological Screening Questionnaire (MSNQ)

The MSNQ was developed to identify neuropsychological impairment in MS and includes a patient and an informant/caregiver version [21, 22]. The patient version of the MSNQ, which comprises 15 items, was used in this study. Patients rate themselves from 0 (never; does not occur) to 4 (very often; very disrupted) on specific cognitive and behavioral problems that may arise in daily life. These 15 items are grouped into a total score ranging from 0 to 60, where higher scores indicate increased neuropsychological impairment or depressive disorder. Cronbach's alpha coefficient for the patient MSNQ is 0.93, and the PRO has been strongly correlated with a more general cognitive complaints questionnaire [22]. The administration schedule for the scale is given in Table 1.

Treatment Satisfaction Questionnaire with Medication version 1.4 (TSQM v1.4)

The TSQM (version-1.4) was developed as a generic instrument to evaluate patients’ satisfaction with medication and comprises 14 items grouped into side-effects score (4 items), effectiveness score (3 items), convenience score (3 items), and global satisfaction score (3 items) [23, 24]. The remaining question is a filter item [24]. Each score ranges from 0 to 100, where higher scores indicate greater satisfaction with the different aspects of medication. Internal consistency reliability in MS patients is reportedly high (Cronbach’s α > 0.90) [25]. The administration schedule for the scale is given in Table 1.

Patient-Determined Disease Steps (PDDS)

The PDDS scale was developed to evaluate MS-associated disability overall and in different dimensions, and is a single ordinal item ranging from 0 (normal) to 8 (bedridden), with a higher score indicating more disability [26, 27]. PDDS had a strong [27], albeit not perfect [26], correlation with the Expanded Disability Status Scale (EDSS). In translation, PDDS has shown excellent test–retest reliability [28,29,30]. The administration schedule for the scale is given in Table 1.

Health-Related Productivity Questionnaire in Multiple Sclerosis version 2 (HRPQ-MS v2)

The HRPQ-MS v2 was developed from the original HRPQ instrument [31] to evaluate health-related productivity in people being treated for MS [32, 33]. This tool comprises 9 open and multiple-choice questions that quantify hours of employment-related lost productivity and household-related lost productivity. HRPQ has good construct and criterion validity with the Work Productivity and Activity Impairment Questionnaire (WPAI) (Pearson’s r ≥ 0.6, P < 0.05) [32]. The administration schedule for the scale is given in Table 1.

Hospital Anxiety and Depression Scale (HADS)

The HADS is a generic self-administered questionnaire to evaluate states of anxiety and depression and comprises 14 items that are grouped into an anxiety (7 items) and depression score (7 items) [34,35,36]. Each score ranges from 0 (normal) to 21 (abnormal), with higher scores indicating increased anxiety or depression. HADS Anxiety and Depression scales have acceptable internal consistency reliability as measured by Cronbach's alpha (0.86, 0.82, respectively) in people with MS [37]. In addition, test–retest reliability was 0.83 for both HADS Anxiety and HADS Depression scales [37]. The administration schedule for the scale is given in Table 1.

Statistical analysis

Data analysis

Analysis was performed using SAS software version 9.4 (SAS Institute Inc, Cary, NC, USA). Continuous variables are described by frequency, mean, standard deviation (SD), median, [minimum, maximum], and number of missing values. Categorical variables are described by frequency and percentages, with missing data included in calculation of percentages. Missing item responses within a PRO questionnaire were managed in the creation of PRO scores as specified by the developers of the instruments. Analyses were conducted on the FAS.

Anchor-based analyses

Anchor-based methods are the primary approach for determining a meaningful within-individual change threshold, as recommended by the US Food and Drug Administration (FDA) [15], and were used for the MSIS-29, FSMC, MSPS, and MSNQ scales. Available variables used as potential anchors were PDDS for MSIS-29 (physical and psychological impact scores); MSPS (total score); MSPS fatigue item for FSMC (physical fatigue, mental fatigue, total scores); and MSPS cognition item for MSNQ (total score). Time period assessed was change from baseline to Month 12, with the exception of FSMC, for which baseline to Month 11 was used.

Correlation between each anchor variable and the corresponding PRO score and between change in each anchor variable and change in the corresponding PRO score was estimated using Spearman’s rank-order correlation coefficients. Change in PRO scores was described in the groups defined according to the change in anchor variables over the same time period.

Receiver-operating characteristic (ROC) curves were plotted to characterize the separation of groups defined according to various dichotomous categorizations of the anchor variables by the change in corresponding PRO score; from these ROC curves, change in PRO score that maximized the separation between dichotomous categorizations of the anchor variable was estimated by the smallest sum of squares of 1-sensitivity and 1-specificity [38].

Empirical cumulative distribution function (eCDF) and probability density function (PDF)

eCDF and PDF of change in MSIS-29, FSMC, MSPS, and MSNQ scores were plotted over the specified time periods according to groups defined by the potential anchors. Graphical representations of eCDFs of change in PRO scores in groups were defined by the anchor variables (Additional file 1: Fig. S1). For the MSIS-29 v2 and FSMC scores, as well as MSPS and MSNQ total scores, anchor variables (ie, variables assessing similar concepts) were available for both anchor-based and distribution-based methods and, therefore, could be applied.

Distribution-based analyses

Distribution-based methods were used to generate additional supportive data to define meaningful within-individual change in all PRO scores, including PROs covered by anchor-based methods. For each score, values corresponding to previously defined effect size (ES) thresholds [39] were calculated using the score SD at baseline: 0.2SD, 0.5SD, and 0.8SD. The standard error of measurement (SEM) of each score was then calculated; SEM of a score was defined from the score SD and its reliability coefficient (r) \(\left( {SD \times \sqrt {\left( {1 - r} \right)} } \right)\), which was estimated using Cronbach’s coefficient alpha calculated at baseline.

Triangulation and PRO responder analysis

The results of the analyses were triangulated to define a reference value and a conservative value for the clinically meaningful within-individual change in MSIS-29, FSMC, MSPS, MSNQ, and a reference value for the clinically meaningful within-individual change for TSQM, PDDS, HRPQ-MS, and HADS scores. Results from the anchor-based methods represent meaningful within-individual improvement only, while results from the distribution-based methods are based on the score distribution hence have no direction and represent meaningful change. The reference value was established as change in PRO score of participants who showed an improvement of one or two categories in the corresponding anchor variable for PRO scores for which anchor-based methods were applied. For PRO scores in which only distribution-based methods were applied, the reference value was defined by 0.5SD at baseline.

For establishing a range of meaningful within-patient improvement, a “conservative” value was defined based on anchor-based and distribution-based methods. This value was defined to acknowledge the uncertainty of the estimate and provide a value that can be considered a “worst-case scenario”. The values obtained by the analyses were compared with the minimum detectable change \(\left( {1.96 \times \sqrt 2 \times SEM} \right)\), which is the smallest change that is above measurement error [40]. The estimated meaningful within-individual improvement value was then evaluated to determine whether it can be observed in practice; if not, the closest greater observable value was used as the meaningful within-individual improvement value. We considered any improvement greater than the lower-bound estimate using the “worst case scenario” approach to be “clinically meaningful.”

A responder was defined as a participant with a change in score from baseline greater than the clinically meaningful within-individual improvement value. The percentage responder analysis was calculated at Month 10 for TSQM v1.4 scores, Month 11 for FSMC and HRPQ-MS v2 scores, and Month 12 for all other scores.


Patient characteristics

Baseline characteristics and PRO scores in the FAS (n = 434) are summarized in Table 2. Participants were heterogeneous in terms of both demographic and clinical characteristics. The majority of participants were female (77%) and white (86%), and the mean ± SD age was 50 ± 11 years. Time since first diagnosis of MS was 11 ± 10 years, with 25% of participants diagnosed for less than 3 years, and 25% for more than 17 years. The majority of patients (62%) had experienced ≥ 1 relapse in the past year.

Table 2 Baseline characteristics and PRO scores

Baseline PRO scores suggested that participants typically had moderate physical and psychological impact and symptoms, low anxiety and depression, and low satisfaction with prestudy treatment (Table 2). Participants reported moderate impact on their ability to perform daily activities and to remain in the workplace (Table 2).

Evaluation of the association of the PRO scores with the possible anchor variables

Correlations between each anchor variable and the corresponding PRO score at baseline are shown in Table 3. Overall, high correlations (> 0.7) between the anchor variable and corresponding PRO score at baseline were observed, except for PDDS and MSIS-29 v2 psychological impact score, for which correlation coefficient was moderate (0.38).

Table 3 Correlation coefficients between the anchor variables and the corresponding PRO scores at baseline

Lower correlations were observed between the change in anchors and change in corresponding PRO scores from baseline to Month 12. Overall, correlations were below 0.3 (Additional file 1: Table S1), indicating low correlations between change captured by anchors and change in PRO scores. Only change in MSPS total score and change in PDDS were moderately correlated (0.31).

Meaningful within-individual improvements for PRO scores

Results for the instruments assessed by anchor-based and distribution-based with triangulation of results for meaningful improvement in PRO scores are summarized in Table 4. The supportive data on meaningful within-individual change for those PRO instruments where only distribution-based methods were possible, since no anchor variable was available (multi-item and single-item scores), are shown in Table 5; the recommended value was defined by 0.5SD at baseline, and the conservative value (multi-item scores only) was based on minimum detectable change. The responder analysis, based on the number of participants with an improvement in score from baseline greater than the clinically meaningful within-individual improvement, was calculated for all PRO scores and is summarized in Additional file 1: Fig. S2. The proportion of responders ranged from 9.1% (HADS Depression) to 34.2% (MSIS-29 v2 psychological impact). The proportion of nonresponders ranged from 52.3% (TSQM effectiveness) to 76.5% (FSMC physical fatigue). Between 5.4% and 23.8% of responses were missing, depending on PRO.

Table 4 Within-individual estimates from anchor-based and distribution-based methods for MSIS-29 v2, FSMC, MSPS, and MSNQ scores
Table 5 Within-individual change estimates from distribution-based methods for TSQM v1.4, HADS, PDDS and HRPQ-MS v2 scores


This analysis of PRO data from the PROMiS AUBAGIO study has provided candidate values for meaningful within-individual improvement in scores for eight PRO instruments commonly used in people with MS, namely the MSIS-29 v2, FSMC, MSPS, MSNQ, TSQM v1.4, PDDS, HRPQ-MS v2, and HADS (anxiety and depression) instruments. Although not all of these instruments are specific to MS, determination of clinically meaningful within-individual improvement thresholds followed the principles currently recommended by the FDA [15] and resulted from triangulation of results from anchor-based methods, when available, supported by graphical representations of eCDFs of the change in PRO scores groups defined by the change in anchor variables, and distribution-based methods. A systematic approach was applied in this triangulation, with the results from anchor-based methods considered primary, and the full set of results providing a sense of robustness of these results. It is our hope that these values, particularly for the MSIS-29 v2, FSMC, MSPS, and MSNQ scores applying anchor-based methods to provide recommended values, will be helpful in terms of interpreting responses to interventions in the future but also to give better context for previously reported MS studies where PRO data are currently presented.

PRO data are considered of central importance in the assessment of people with MS, and the use of PRO data is likely to become more widespread and clinically important with the development of on-line assessments and smartphone-based technologies that facilitate the exchange of such data between MS patients and their care providers [41]. Furthermore, PROs are often used as secondary or exploratory endpoints in clinical trials, but without a standardized approach, it is difficult to provide a comparative assessment of effectiveness or to enable PRO-based decision-making by regulators or other decision makers. Indeed, a recent literature review has identified inappropriate reporting of PRO data to be a common weakness in MS trial publications and, as a result, an area in need of improvement [42]. To our knowledge, such information was previously only available for certain of the PROs evaluated. For example, the recommended meaningful within-individual improvement for the MSIS-29 v2 physical impact score estimated in our study (− 6.68) is within the range of published values from other studies for worsening and improvement (range of − 4.84 to − 8 based on anchor-based methods, and range of − 2.22 to − 10.4 based on distribution-based methods) [43,44,45]. We used the PDDS in our anchor-based methods, and previous work using the EDSS as the anchor has reported a meaningful change in MSIS-29 v2 physical impact score of approximately 7.5 for worsening [44, 45] and 5 for improvement [43]. In addition to the MSIS-29 v2 physical impact score, meaningful within-individual improvement value on the MSIS-29 v2 psychological impact score (recommended value of − 7.40) was also estimated, which has not been as widely documented. However, given the wide-ranging impact that MS has on patients' well-being [46], providing meaningful change thresholds on the different subscales broadens the utility of the PRO instruments and allows meaningful improvements in these symptoms to be captured from the patient’s perspective.

Application of minimally important change represents an important tool for enhancing the interpretability of PROs; however to realize the full benefit of the value, an improved understanding is needed particularly around reporting the fundamental properties of the change [47]. To address some of the shortcomings, the FDA are developing a series of methodological patient-focused drug development documents to address how stakeholders can collect and submit patient and caregiver experience data in a stepwise manner for medical product development and regulatory decision-making [16]. Currently available in draft form, the documents aim to provide industry with information to support integration of the patient experience into drug development programs and to guide next steps, such as with external stakeholders who may want to undertake the development of tools within a given disease area [16]. With a more standardized and comparable approach, including providing updated guidance on reliability, validity, and ability to detect minimally important change [48], we hope PROs will contribute more reliably to regulatory decision-making in the future. Providing clinically meaningful thresholds for eight commonly used PRO instruments for MS, albeit requiring some further validation, can only assist with these endeavors in this healthcare space.

Of the eight commonly used PRO instruments selected, some are validated in MS patients [19, 20, 26, 32, 35] and others are studied to various degrees in MS [18, 25, 49,50,51,52]. These scales included measures of HRQOL, fatigue, functional performance, work capacity/daily activities, cognitive impairment, anxiety/depression, and satisfaction with treatment, all pertinent to MS symptoms. Performance scales selected had good internal consistency, and most had published data on test–retest reliability. However, PROs used in the PROMiS AUBAGIO study were not exhaustive, and other instruments have been developed, validated, and/or are commonly used in MS, eg, the FSIQ-RMS [53], Neuro-QoL™ [54], SymptoMScreen [55], and FACIT-TS-G [56], among others. Hence, there remains the need for establishing meaningful within-individual change thresholds in a similar manner for these measures, to further support our work. The current study focused on traditionally used instruments that are often applied in the clinical trial or real-world setting, where documenting a meaningful within-individual change in symptoms would be a benefit for people with MS, healthcare providers, and decision makers. The PROMiS AUBAGIO study required enrollees to have fluent English language skills, so we do not know if our observations would be similar for patients completing translated PROs in non-English languages, for which many have been validated. The methods and assumptions used in the current study to determine our observations were based on recommendations from the FDA [15], using triangulation of both anchor-based (primary) and distribution-based (supportive) methods, alongside other considerations highlighted in the literature when calculating clinically meaningful within-individual changes [38, 40, 57]. Although we followed FDA-recommended anchor-based methods, there are no standardized triangulation procedures, to our knowledge. Our triangulation approach involved critically reviewing all estimates from the various methods used, identifying a reference value, then defining a range of values that could be considered “clinically meaningful”. We took this approach to allow estimates for meaningful change for each PRO from the various methods to be considered qualitatively and allow reflection on the variability of the estimates. We did not address questions around different meaningful change values for different subpopulations, as this is an area of debate beyond the scope of this analysis; for example, is meaningful change linked to score only or to context of use? Instead, we took a considered approach, and it is our hope these observations will ultimately facilitate decision-making by regulatory and healthcare authorities moving forwards.

Our study should be considered in light of certain limitations, as the determination of clinically meaningful within-individual change in PRO scores was not a prespecified objective of the PROMiS AUBAGIO study. There is, therefore, some uncertainty associated with the estimated values for meaningful within-individual improvement in all PRO scores based on both anchor- and distribution-based methods. For several PRO scores, no anchor variable assessing the same concept was available; therefore, only distribution-based methods could be applied. For example, values for meaningful change in TSQM v1.4 scores were estimated using distribution-based methods, with estimates of − 10.81 and − 9.99 for the convenience and effectiveness scales, respectively. However, given the high variability in the TSQM these values should be used with caution. The TSQM is commonly used cross-sectionally not longitudinally, although this is not unique to the current study. Although not a valid measure for assessing treatment satisfaction in people with RRMS, scale-to-sample targeting implied that treatment satisfaction may be underestimated by the TSQM; hence, further research is required to overcome this limitation [25]. We therefore recommend the estimated clinically meaningful within-individual improvement values from scales where an acceptable anchor was available (ie, MSIS-29 v2 physical and psychological impact scores; FSMC physical and cognitive fatigue scores, and total score; MSPS total score; MSNQ total score), where both a recommended value and conservative value are provided, and may be useful for sensitivity analyses to allow better interpretation in future studies.

Also of note, the anchor-based estimates were obtained using anchor variables that were not optimal, as indicated by the low correlations of changes in PRO score and anchor variables (Additional file 1: Table S1) and by the overlap observed in eCDFs of change in PRO scores in the categories defined by the anchor variable. This limited discrimination probably led to underestimation of meaningful improvement values. These analyses, especially the anchor-based methods, focused on determination of meaningful within-individual improvement; as such there is no certainty that these values could also be used to characterize clinically meaningful within-individual worsening. Given MS is generally a disease of declining function, particularly over the long-term, further study is needed to clarify conclusions of clinically meaningful within-individual worsening. We also did not look at how values correlate with physician’s perspectives of clinical importance.

Despite these limitations, our study also had notable strengths. Our study took a broad approach to examine eight commonly used PRO instruments. The size of the study population meant that a reasonably large PRO database was generated, and, given the methods of analysis used, as discussed above, we feel our approach can be considered robust. The study population of people with RRMS was also heterogenous with regard to baseline characteristics, demographics, and baseline PRO scores. Although this introduces inherent variability into our analyses and probably led to higher SDs around the PRO scores (and, thus, overestimated values from distribution-based methods), the heterogenicity of the study population does help to ensure the calculations may be applicable to a broader range of people with RRMS. Our estimates should, therefore, be relevant to a range of future clinical studies in RRMS populations.


This study has produced recommended estimates for assessing meaningful within-individual improvements based on results from PRO instruments used in MS clinical studies. These estimates will be useful for interpreting improvement scores and communicating the results of future studies evaluating the impact of RRMS on patients and will facilitate decision-making by regulatory and healthcare authorities. Additional work to estimate and confirm meaningful within-individual improvement thresholds in MS target populations, eg, primary-progressive MS (PPMS) and/or nonrelapsing secondary progressive MS (nrSPMS) patients is needed. In addition, work to estimate and confirm meaningful within-individual thresholds for worsening scores for all MS patient populations is an area for future research.

Availability of data and materials

Anonymized data will be shared by reasonable request from any qualified investigator.


  1. Lublin FD, Coetzee T, Cohen JA, Marrie RA, Thompson AJ (2020) The 2013 clinical course descriptors for multiple sclerosis: a clarification. Neurology 94(24):1088–1092

    Article  PubMed  PubMed Central  Google Scholar 

  2. Thompson AJ, Banwell BL, Barkhof F et al (2018) Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. Lancet Neurol 17(2):162–173

    Article  PubMed  Google Scholar 

  3. Walton C, King R, Rechtman L et al (2020) Rising prevalence of multiple sclerosis worldwide: Insights from the Atlas of MS, third edition. Mult Scler 26(14):1816–1821

    Article  PubMed  PubMed Central  Google Scholar 

  4. Multiple Sclerosis International Federation (MSIF). Atlas of MS (3rd Edition): Mapping multiple sclerosis around the world key epidemiology findings. Available at: Accessed 12 Jul 2022

  5. Hauser SL, Cree BAC (2020) Treatment of multiple sclerosis: a review. Am J Med 133(12):1380–1390.e2

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Rae-Grant A, Day GS, Marrie RA et al (2018) Practice guideline recommendations summary: disease-modifying therapies for adults with multiple sclerosis: report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology 90(17):777–788

    Article  PubMed  Google Scholar 

  7. Zhang J, Shi S, Zhang Y et al (2017) Alemtuzumab versus interferon beta 1a for relapsing-remitting multiple sclerosis. Cochrane Database Syst Rev 11(11):CD010968

    PubMed  Google Scholar 

  8. Coles AJ, Twyman CL, Arnold DL et al (2012) Alemtuzumab for patients with relapsing multiple sclerosis after disease-modifying therapy: a randomised controlled phase 3 trial. Lancet 380(9856):1829–1839

    Article  CAS  PubMed  Google Scholar 

  9. Straus Farber R, Harel A, Lublin F (2016) Novel agents for relapsing forms of multiple sclerosis. Annu Rev Med 67:309–321

    Article  CAS  PubMed  Google Scholar 

  10. Manohar N, Perkins T, Laurion R, Oliver B (2019) Patient-reported outcomes in multiple sclerosis: an overview. J Sci Commun 26(6):277–288

    Google Scholar 

  11. Klineova S, Lublin FD (2018) Clinical course of multiple sclerosis. Cold Spring Harb Perspect Med 8(9):a028928

    Article  PubMed  PubMed Central  Google Scholar 

  12. Nowinski CJ, Miller DM, Cella D (2017) Evolution of patient-reported outcomes and their role in multiple sclerosis clinical trials. Neurotherapeutics 14(4):934–944

    Article  PubMed  PubMed Central  Google Scholar 

  13. Brichetto G, Zaratin P (2020) Measuring outcomes that matter most to people with multiple sclerosis: the role of patient-reported outcomes. Curr Opin Neurol 33(3):295

    Article  PubMed  PubMed Central  Google Scholar 

  14. Wyrwich KW, Norquist JM, Lenderking WR, Acaster S (2013) Industry Advisory Committee of International Society for Quality of Life Research (ISOQOL). Methods for interpreting change over time in patient-reported outcome measures. Qual Life Res 22(3):475–483

    Article  CAS  PubMed  Google Scholar 

  15. Food and Drug Administration (FDA). Incorporating clinical outcome assessments into endpoints for regulatory decision-making. Available at: Accessed 15 May 2022

  16. Food and Drug Administration (FDA). Plan for issuance of patient‐focused drug development guidance. Available at: Accessed 13 Jan 2023

  17. Hobart J, Lamping D, Fitzpatrick R, Riazi A, Thompson A (2001) The multiple sclerosis impact scale (MSIS-29): a new patient-based outcome measure. Brain 124(Pt 5):962–973

    Article  CAS  PubMed  Google Scholar 

  18. McGuigan C, Hutchinson M (2004) The Multiple Sclerosis Impact Scale (MSIS-29) is a reliable and sensitive measure. J Neurol Neurosurg Psychiatr 75(2):266–269

    CAS  Google Scholar 

  19. Penner IK, Raselli C, Stocklin M, Opwis K, Kappos L, Calabrese P (2009) The Fatigue Scale for Motor and Cognitive Functions (FSMC): validation of a new instrument to assess multiple sclerosis-related fatigue. Mult Scler 15(12):1509–1517

    Article  CAS  PubMed  Google Scholar 

  20. Marrie RA, Goldman M (2007) Validity of performance scales for disability assessment in multiple sclerosis. Mult Scler 13(9):1176–1182

    Article  CAS  PubMed  Google Scholar 

  21. Benedict RH, Cox D, Thompson LL, Foley F, Weinstock-Guttman B, Munschauer F (2004) Reliable screening for neuropsychological impairment in multiple sclerosis. Mult Scler 10(6):675–678

    Article  PubMed  Google Scholar 

  22. Benedict RH, Munschauer F, Linn R et al (2003) Screening for multiple sclerosis cognitive impairment using a self-administered 15-item questionnaire. Mult Scler 9(1):95–101

    Article  CAS  PubMed  Google Scholar 

  23. IQVIA Treatment Satisfaction Questionnaire for Medication (TSQM)© Available at:

  24. Atkinson MJ, Sinha A, Hass SL et al (2004) Validation of a general measure of treatment satisfaction, the Treatment Satisfaction Questionnaire for Medication (TSQM), using a national panel study of chronic disease. Health Qual Life Outcomes 2:12

    Article  PubMed  PubMed Central  Google Scholar 

  25. Vermersch P, Hobart J, Dive-Pouletty C, Bozzi S, Hass S, Coyle PK (2017) Measuring treatment satisfaction in MS: Is the Treatment Satisfaction Questionnaire for Medication fit for purpose? Mult Scler 23(4):604–613

    Article  PubMed  Google Scholar 

  26. Learmonth YC, Motl RW, Sandroff BM, Pula JH, Cadavid D (2013) Validation of patient determined disease steps (PDDS) scale scores in persons with multiple sclerosis. BMC Neurol 13(1):37

    Article  PubMed  PubMed Central  Google Scholar 

  27. Hohol MJ, Orav EJ, Weiner HL (1999) Disease steps in multiple sclerosis: a longitudinal study comparing disease steps and EDSS to evaluate disease progression. Mult Scler 5(5):349–354

    Article  CAS  PubMed  Google Scholar 

  28. Aldughmi M, Al-Shorman A, Khalil H, El-Salem K, Alghwiri A (2022) Translation and validation of the Arabic version of the patient determined disease steps in people with multiple sclerosis. Physiother Theory Pract 38(9):1281–1288

    Article  Google Scholar 

  29. Kahraman T, Özdoğar AT, Özakbaş S (2021) Cross-cultural adaptation, validity and reliability of the Turkish version of the patient determined disease steps scale in persons with multiple sclerosis. Physiother Theory Pract 37(4):527–534

    Article  PubMed  Google Scholar 

  30. Solà-Valls N, Vicente-Pascual M, Blanco Y et al (2019) Spanish validation of the telephone assessed Expanded Disability Status Scale and Patient Determined Disease Steps in people with multiple sclerosis. Mult Scler Relat Disord 27:333–339

    Article  PubMed  Google Scholar 

  31. Kumar RN, Hass SL, Li JZ, Nickens DJ, Daenzer CL, Wathen LK (2003) Validation of the Health-Related Productivity Questionnaire Diary (HRPQ-D) on a sample of patients with infectious mononucleosis: results from a phase 1 multicenter clinical trial. J Occup Environ Med 45(8):899–907

    Article  PubMed  Google Scholar 

  32. Tundia N, Hass S, Fuldeore M et al (2015) Validation and U.S. population norms of health-related productivity questionnaire. Value Health. 18(3):A24

    Article  Google Scholar 

  33. Hardy TA, Parratt J, Beadnall H et al (2022) Treatment satisfaction in patients with relapsing-remitting multiple sclerosis initiated on teriflunomide in routine clinical practice: Australian observational data. BMJ Neurol Open 4(2):e000315

    Article  PubMed  PubMed Central  Google Scholar 

  34. Moore P, Hirst C, Harding KE, Clarkson H, Pickersgill TP, Robertson NP (2012) Multiple sclerosis relapses and depression. J Psychosom Res 73(4):272–276

    Article  PubMed  Google Scholar 

  35. Watson TM, Ford E, Worthington E, Lincoln NB (2014) Validation of mood measures for people with multiple sclerosis. Int J MS Care 16(2):105–109

    Article  PubMed  PubMed Central  Google Scholar 

  36. Zigmond AS, Snaith RP (1983) The hospital anxiety and depression scale. Acta Psychiatr Scand 67(6):361–370

    Article  CAS  PubMed  Google Scholar 

  37. Marrie RA, Zhang L, Lix LM et al (2018) The validity and reliability of screening measures for depression and anxiety disorders in multiple sclerosis. Mult Scler Relat Disord 20:9–15

    Article  PubMed  Google Scholar 

  38. Froud R, Abel G (2014) Using ROC curves to choose minimally important change thresholds when sensitivity and specificity are valued equally: the forgotten lesson of Pythagoras. Theoretical considerations and an example application of change in health status. PLoS ONE 9(12):e114468

    Article  PubMed  PubMed Central  Google Scholar 

  39. Cohen J. Statistical power analysis for the behavioral sciences. Academic Press, 2013

  40. de Vet HC, Terwee CB, Ostelo RW, Beckerman H, Knol DL, Bouter LM (2006) Minimal changes in health status questionnaires: distinction between minimally detectable change and minimally important change. Health Qual Life Outcomes 4:54

    Article  PubMed  PubMed Central  Google Scholar 

  41. Altmann P, Hinterberger W, Leutmezer F et al (2021) The smartphone app haMSter for tracking patient-reported outcomes in people with multiple sclerosis: protocol for a pilot study. JMIR Res Protoc 10(5):e25011

    Article  PubMed  PubMed Central  Google Scholar 

  42. Khan T, Khalid M, Dunford B et al (2022) Incomplete reporting of patient-reported outcomes in multiple sclerosis: a meta-epidemiological study of randomized controlled trials. Mult Scler Relat Disord 63:103819

    Article  PubMed  Google Scholar 

  43. Baert I, Smedal T, Kalron A et al (2018) Responsiveness and meaningful improvement of mobility measures following MS rehabilitation. Neurology 91(20):e1880–e1892

    Article  PubMed  Google Scholar 

  44. Costelloe L, O’Rourke K, Kearney H et al (2007) The patient knows best: significant change in the physical component of the Multiple Sclerosis Impact Scale (MSIS-29 physical). J Neurol Neurosurg Psychiatr 78(8):841–844

    Article  Google Scholar 

  45. Phillips GA, Wyrwich KW, Guo S et al (2014) Responder definition of the Multiple Sclerosis Impact Scale physical impact subscale for patients with physical worsening. Mult Scler 20(13):1753–1760

    Article  PubMed  PubMed Central  Google Scholar 

  46. Hanna M, Strober LB (2020) Anxiety and depression in multiple sclerosis (MS): antecedents, consequences, and differential impact on well-being and quality of life. Mult Scler Relat Disord 44:102261

    Article  PubMed  PubMed Central  Google Scholar 

  47. Devji T, Carrasco-Labra A, Guyatt G (2021) Mind the methods of determining minimal important differences: three critical issues to consider. Evid Based Ment Health 24(2):77–81

    Article  PubMed  Google Scholar 

  48. Food and Drug Administration (FDA). Guidance for industry patient-reported outcome measures: use in medical product development to support labeling claims. Available at:

  49. Benedict RH, Zivadinov R (2006) Predicting neuropsychological abnormalities in multiple sclerosis. J Neurol Sci 245(1–2):67–72

    Article  PubMed  Google Scholar 

  50. Skoog B, Link J, Tedeholm H et al (2019) Short-term prediction of secondary progression in a sliding window: a test of a predicting algorithm in a validation cohort. Mult Scler J Exp Transl Clin 5(3):2055217319875466

    CAS  PubMed  PubMed Central  Google Scholar 

  51. Cleanthous S, Cano S, Kinter E et al (2017) Measuring the impact of multiple sclerosis: enhancing the measurement performance of the Multiple Sclerosis Impact Scale (MSIS-29) using Rasch measurement theory (RMT). Mult Scler J Exp Transl Clin 3(3):2055217317725917

    PubMed  Google Scholar 

  52. Bacci ED, Wyrwich KW, Phillips GA, Vollmer T, Guo S (2016) Analysis of the psychometric properties of the Multiple Sclerosis Impact Scale-29 (MSIS-29) in relapsing-remitting multiple sclerosis using classical and modern test theory. Mult Scler J Exp Transl Clin 2:2055217316673235

    CAS  PubMed  PubMed Central  Google Scholar 

  53. Hudgens S, Schuler R, Stokes J, Eremenco S, Hunsche E, Leist TP (2019) Development and validation of the FSIQ-RMS: a new patient-reported questionnaire to assess symptoms and impacts of fatigue in relapsing multiple sclerosis. Value Health 22(4):453–466

    Article  PubMed  Google Scholar 

  54. Medina LD, Torres S, Alvarez E, Valdez B, Nair KV (2019) Patient-reported outcomes in multiple sclerosis: validation of the Quality of Life in Neurological Disorders (Neuro-QoL) short forms. Mult Scler J Exp Transl Clin 5(4):2055217319885986

    PubMed  PubMed Central  Google Scholar 

  55. Fitzgerald KC, Salter A, Tyry T et al (2019) Validation of the SymptoMScreen with performance-based or clinician-assessed outcomes. Mult Scler Relat Disord 29:86–93

    Article  PubMed  Google Scholar 

  56. Engebretson E, Seale RA, Valdez B, Vollmer TL, Medina LD (2020) Validation of the functional assessment of chronic illness therapy-general treatment satisfaction (FACIT-TS-G) in multiple sclerosis. Mult Scler Relat Disord 45:102413

    Article  PubMed  Google Scholar 

  57. Schwartz CE, Ayandeh A, Motl RW (2014) Investigating the minimal important difference in ambulation in multiple sclerosis: A disconnect between performance-based and patient-reported outcomes? J Neurol Sci 347(1–2):268–274

    Article  PubMed  Google Scholar 

Download references


Medical writing support was provided by Karen Burrows, MPhil, CMPP, of Curo (a division of Envision Pharma Group) and was funded by Sanofi. Authors would also like to thank Jérôme Msihid of Sanofi for his thoughtful biostatistical reviews and insights on this work.


This study was funded by Sanofi.

Author information

Authors and Affiliations



All authors were involved in either the concept or design of the study, in data acquisition, analysis, or in interpretation of the data. All authors were involved in drafting the manuscript. All authors were involved in making critical revisions to the manuscript during development. All authors agree to be accountable for the integrity of the final content. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Nupur Greene.

Ethics declarations

Ethics approval and consent to participate

Participants signed an electronic informed consent form prior to participation in the study. Collected personal data for this observational study were processed in compliance with laws governing personal data protection in force at the time in the United States.

Consent for publication

All authors consent to publication and approved the final draft of the manuscript.

Competing interests

NG and KH are full-time employees of Sanofi and may own stock or stock options. SQ, FM, MM’H, AL, and AR are employees of Modus Outcomes and were paid consultants to Sanofi in relation to this analysis. DPB was an employee of Sanofi at the time of the study and is now a full-time employee of Novartis.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Denise P. Bury: Sanofi, Cambridge, MA, USA (at the time of the study).

Supplementary Information

Additional file 1: Fig. S1.

Empirical cumulative distribution functions (eCDFs) for change in PRO scores where anchor variables were available. A MSIS-29 v2 physical impact; B MSIS-29 v2 psychological impact C FSMC Cognitive Score; D FSMC Motor Score E FSMC Total Score; F MSPS Total Score G MSNQ. Fig. S2. Responders for A multi-item and B single-item PROs using recommended meaningful within-individual improvement thresholds. Table S1. Correlation coefficients between change in anchor variables and the change in the corresponding PRO scores.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Greene, N., Quéré, S., Bury, D.P. et al. Establishing clinically meaningful within-individual improvement thresholds for eight patient-reported outcome measures in people with relapsing-remitting multiple sclerosis. J Patient Rep Outcomes 7, 61 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: