PROMIS-10 global health.

The PROMIS-10 is a validated self-reported ten item questionnaire. Each item is scored on a 5-point ordinal scale, with 1 indicating poorer health. The scale has two domains of mental health and physical health, as well as a total composite score. The raw domain scores are converted to a T-score, normed to the general population in the U.S. (mean = 50, standard deviation = 10), such that a person with a score of 60 is one standard deviation healthier than the average [25].

RiKsStoke and ICHOM stroke set questionnaires.

The RiksStroke Questionnaire consists of three questions (one three-point scale, and two binary items) related to walking, toileting, and being able to dress, and the International Consortium for Health Outcome Measures (ICHOM) Stroke Specific Question Set consists of two binary questions: having a feeding tube; and or communication problems.

It is these 15 items that comprise the PRO and are to be validated with three samples.

European Brain Injury Questionnaire.

The European Brain Injury Questionnaire (EBIQ) is a validated self-reported 63-item questionnaire, measuring a range of difficulties experienced by brain-injured patients. Each item is scored on a three-point scale, indicating how often the patient has experienced a particular difficulty in the past month. The scale consists of eight subscales (somatic, cognitive, motivation, impulse, depression, isolation, physical, and communication), and a composite score. Missing item data were not scored within the subscales.

Multiple sclerosis impact scale v2.

The Multiple Sclerosis Impact Scale v2 (MSIS-29) is a validated self-reported 29-item questionnaire measuring the physical and psychological impact of MS on patients’ daily lives. Each item is measured on a four-point ordinal scale, indicating how much the patient has been bothered by each symptom in the past 2 weeks. Scores for both domains are standardised to the common range, 0 to 100, with higher scores indicating higher levels of impact. Missing item data resulted in subscales being reported as missing.

Parkinson’s disease questionnaire.

The Parkinson’s disease questionnaire (PDQ-39) is a validated, self-reported 39-item questionnaire measuring the difficulties in daily living experienced by patients with Parkinson’s disease. Items are measured on a 5-point ordinal scale, indicating how often the patient has experienced difficulties in the last month, across eight dimensions: Mobility; Activities of daily living; Emotional wellbeing; Stigma; Social Support; Cognitive impairment; Communications; and Bodily discomfort. Missing item data resulted in subscales being reported as missing.

The EuroQol-5D.

The EuroQol-5D (EQ-5D) is a validated generic self-reported QoL questionnaire, it has two parts. Part 1 consists of five domains (mobility, self-care, usual activities, pain/discomfort, anxiety/depression), which are rated on four-point scale. The digits for the 5 dimensions are combined to provide a 5-digit description of current health state. The composite EQ-5D health state scores were converted to a crosswalk index score using the EQ-5D index value calculator (version 1.1; [13]). The crosswalk index can be interpreted such that 1 reflects a ‘perfect’ health score. The second part asks the respondent to rate their health on a visual analogue scale (VAS) from 0 to 100.

Exploratory Factor Analysis (EFA).

One item from the PRO (tube feeding) was omitted from the factor analysis due to having a poor discrimination since only one person answered ‘yes’ to having a feeding tube per cohort.

The PD cohort was used for EFA of the PRO, fitting up to 3 factors. The sample correlation matrix exhibited two eigenvalues above greater than one (8.103, 1.468; see S1 Fig), and thus up to two factors may be extracted [40]. Adequate fit was achieved at both the two- and three-factor solutions, with an improved fit achieved at the three-factor solution, according to the goodness of fit indices (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Goodness of fit indices for exploratory factor analysis and confirmatory factor analysis models across all cohorts.

https://doi.org/10.1371/journal.pone.0251484.t001

Confirmatory factor analysis.

The ABI and MS cohorts were used to confirm the EFA factor structure, the cohorts fell below the minimum sample and present as first evidence. The model fit for both cohorts was improved as the number of factors was increased from one to three. The one-factor solution did not provide adequate fit for either cohort. The 3-factor structure, with factor loadings shown in Table 2, provided adequate fit for the MS cohort, but the fit was more marginal for the ABI cohort.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Standardised factor loadings for exploratory factor analysis of the PD sample and confirmatory factor analysis (within brackets ABI / MS) for the 3-factor solution.

https://doi.org/10.1371/journal.pone.0251484.t002

Interpretation of the factor analysis.

The three-factor solution contained three latent domains within PRO, physical, emotional, and functional capacity (Table 2). The first factor related to self-perceived levels of physical health (PH), containing four items from the PROMIS-10: quality of life, physical health, general health, and fatigue. The second factor grouped three RiksStroke items and two PROMIS-10 items, relating to daily functional capacity (FN; going to the toilet, dressing, walking, and general physical activities) and pain rating. These items make up a ‘functional capacity’ factor. The third factor explained mental health (MH) status, grouping psychosocial aspects of health (mental health, social life, social activities, emotional problems and communication). Item 15 (communication problems) may be problematic due to a low loading score of 0.31 [41]. Factor scores are the sum of each item within factor, and lower scores were indicative of a poorer level of health/functioning.

Reliability and validity.

Using the latent domains identified from the factor analysis we estimated the reliability and validity of the latent factors (physical health, functional capacity and mental health) and composite score of the PRO. The internal consistency of the physical health (PH) factor was good for the ABI and MS cohorts (∝>0.80) and excellent for the PD cohort (∝ = 0.89). Cronbach’s alpha for the functional capacity (FN) factor was acceptable for the MS and PD cohorts (∝>0.70) but was less established in the ABI cohort (∝ = 0.61). The reliability of the mental health (MH) factor was good across all three cohorts (∝>0.79). Cronbach’s alpha coefficient was excellent for PRO total score across all three cohorts (∝≥0.89). Within factors PH and MH, alpha could not be improved by deletion of any item and the ITC varied between 0.34 and 0.89, and no items were identified as problematic. Within the functionality factor, the question relating to pain was identified as a potential problematic item in cohorts ABI and MS, with the ITC less than 0.3 and AID suggesting that fit may be improved with removal of the item, but this was not a concern within the PD cohort. McDonald’s omega (ω) found similar findings for the PH factor, ω>0.8 across all cohorts. For the ABI and MS cohorts, omega for the FN factor was acceptable (ω≥0.75), and omega for FN was good in the PD cohort. The MH factor was found to have good internal reliability across the three cohorts, according to omega (ω>0.8).

Table 3 shows the correlations between PRO scores and EQ-5D, EQ VAS and the three cohort specific scales. For correlation between PRO scores and the EQ-5D, the PH and MH factors were moderately to highly correlated with EQ-5D for all cohorts (rho≥0.58), and the FN factor was highly correlated with EQ-5D across all three cohorts (rho>0.8). The EQ5D-VAS was at least moderately correlated with the PH factor across the 3 cohorts (rho = 0.87, 0.76, 0.64 for ABI, MS, and PD, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Construct validity (Spearman’s correlation) between the PRO and RiksStroke, ICHOM, EQ-5D, EQ5D-VAS, and disease specific scales for each cohort.

https://doi.org/10.1371/journal.pone.0251484.t003

Across the cohorts, all three factors correlated with the equivalent disease specific subdomains. Physical health scores of the PRO were moderately correlated with the somatic and physical scores of the EBIQ, physical MSIS-29 score and PDQ-39 mobility score. Functionality was moderately correlated with physical EBIQ and MSIS-29 scores, and PDQ-39 scores of mobility and activities of daily living. Mental health score was most correlated with the isolation, cognition, motivation and depression subscales of the EBIQ, MSIS-29 psychological score, and PDQ-39 emotional wellbeing score, from moderate to high correlations. Factors PH and MH were found to be moderately correlated with the core domain score of the EBIQ. All factors were highly correlated with PDQ-39 total score.

Concurrency with other literature.

Researchers have previously assessed the psychometric utility of disease-specific instruments to address the needs of patients from the three neurological conditions included here. However, it has often been reported that the instruments do not report patients concerns adequately [5]. We have shown that our PRO has good psychometric properties may be used for three neurological conditions.

In the original validation of the PROMIS-10 by Hays et al. [42] the authors suggested (and CFA confirmed) a two-factor structure for the 10 items. A similar structure was identified from the item factor analysis of the 14-item PRO. Here, we compare the two solutions. The new PH factor consists of the previously physical loaded items Global01 (general health), Global03 (physical health) and Global08 (fatigue). In addition to these, the item Global02 (quality of life) was loaded to PH. In Hays et al. [42] this item cross loaded to the PH and MH factors with almost identical loadings (0.45). In our sample the item clearly loads on PH, suggesting patients with these specific diseases interpret quality of life in terms of physical health. Within the PROMIS-10, Global06 (physical activities) and Global07r (pain) loaded to the factor relating physical health. However, with the addition of the three RiksStroke questions, these items load now to a more specific Functionality factor. Finally, with respect to the new MH factor of the PRO, it consists of the items Global04 (mental health), Global05 (social life), and Global10 (emotional problems), again, similar to the results shown by Hays et al. [42]. In addition, the item Global09 (social activities) cross loaded to both factors [42], whereas in our data it loads more closely to the MH factor. We found the ICHOM item relating to communication problems also loads to the new MH factor, prompts to the fact that the factor is developed away from simply measuring mental health, to also encompassing aspects of psychosocial functioning. This is consistent with current view that both physical, and mental, and social functionality are the driving needs for older people [17, 43].

The PRO expands on the previous two factors from the PROMIS-10, capturing an important patient identified, aspect of daily wellbeing that can be impacted by all of the three neurological conditions. The addition of the Functionality factor, alongside both the Physical Health and Mental Health factors, can provide a more complete view of patient reported outcomes. The Functional factor resembles a patient’s ability to conduct day to day functional tasks exhibiting independence whilst pain free. We found that patients’ independence, and specifically their functionality, is important, which is consistent with other research [17].

We found that the PRO had validity with the EQ-5D, and previous authors had also found that the EQ-5D was reliable compared to other disease-specific questionnaires [44]. However, it has also be argued that the EQ-5D had inadequate detail to be used as a routine clinical follow up instrument [9, 45]. In contrast to the EQ-5D the PRO used in this study provides far more detail regarding specific symptoms that affect these three conditions. For example, details such as communication difficulties and it provides three separate domains of patient reported health: mental, physical and functional health.

Consistent with Hunter et al. [4], we found that a single instrument has the potential to represent a wider variety of long-term conditions to reduce the burden on patients, and offers efficiency to the healthcare system, as a harmonised measure consistent for studies of three neurological conditions. Patient’s responses to the PRO were consistent for each of the three conditions, which mirrored other findings from Esbjörnsson, Skoglund and Sunnerhagen [46].

Strengths and weaknesses.

This is the first evaluation of the psychometric properties of a patient relevant outcome measure for multiple neurological diseases in the same study. The study consented patients exhibiting a typical background of patients of the three conditions. A limitation of the current study is that sample sizes were relatively small, of the three cohorts, the ABI and MS cohorts did not reach the minimum suggested sample size of 100 and further evidence of support should be identified [29, 47].

Implications on future clinical practice and research.

This study’s findings suggest that the PRO is likely to reduce the burden for patients completing their follow-up in routine clinical practice, as well being straightforward for clinicians to interpret. Using a single, transferable assessment across three different conditions also offers health care providers advantages. For example, it is simple to use, and efficient to train staff in the use of a single assessment, and it may be useful in assessing changes in healthcare outcomes following changes in service delivery and design.

Measuring PRO across time periods is a key part of their function and future research should include serial measurements of the PRO to ensure that it has a sensitivity to detect changes in disease progression as well as assess the test-retest reliability of the PRO. Future research should also include a larger number of patients.

While basic interpretation of PRO domain scores would suggest that lower scores indicate an increased severity of impact, it is unclear how domain scores relate to severity of the neurological conditions, or how they compare to the general population. Further research is needed to determine a standardised score to the general population, and clinically meaningful thresholds to aid the interpretation.