Chandelier
Senior Member (Voting Rights)
Provisional scoring framework and interpretation
The VPEM Questionnaire yields a multidimensional PEM profile and, in addition, a provisional aggregate score intended for subsequent psychometric evaluation. All scored components are coded so that higher values indicate greater PEM-related burden. Questions 3 and 4 are scored as frequency-weighted mean category scores: the reported crash counts serve only to weight the category values and are not added to the score themselves, so these components remain within a fixed range irrespective of the absolute number of crashes reported.
Item-level scoring
Frequency-weighted scoring of Questions 3 and 4
For each item, the number of crashes reported in a category is multiplied by the category value, summed across categories, and divided by the total number of crashes reported in that item:
Q3 or Q4 = Σ (number of crashes in category × category value) / total number of crashes
Because “no full recovery” denotes failure to return to approximately the level immediately before a crash rather than a finite recovery duration, Question 4 is interpreted as a recovery-burden score rather than as a mean recovery duration. The “no full recovery” category retains its assigned value of 7 in the frequency-weighted Q4 calculation. For example, if three crashes are reported in this category, they contribute 3 × 7 to the numerator. In addition, a separate binary non-recovery indicator is reported (1 if at least one crash without full recovery is reported, 0 otherwise). This binary indicator does not replace or alter the crash count used in the Q4 calculation. No additional weighting beyond the assigned category value of 7 is applied.
Aggregate score
VPEM total score = Q1 + Q2a + Q2b + Q2c + Q2d + Q2e + Q3 + Q4 + Q5
For responses permitting calculation of all components, the nominal score range is 0–54, with higher values indicating greater PEM-related burden. Questions 3 and 4 may take non-integer values; their unrounded values are used to calculate the total score, which may subsequently be rounded to two decimal places for reporting. Pacing (Questions 6 and 7) is excluded, because a low crash frequency may reflect either mild disease or successful exertion avoidance.
Worked example
A respondent reports crashes about once a week (Q1 = 2) and indicates that crashes are highly likely at greater exertion in all five domains (Q2a–Q2e = 2 each; 10 points). In Question 3, three crashes are reported with moderate limitations (value 3) and four with severe limitations (value 4): Q3 = [(3 × 3) + (4 × 4)] / 7 = 25 / 7 = 3.57. In Question 4, the same seven crashes are distributed as two lasting 1–3 days (value 4), four lasting 4–7 days (value 5), and one involving no full recovery (value 7): Q4 = [(2 × 4) + (4 × 5) + (1 × 7)] / 7 = 35 / 7 = 5.00. Because at least one crash involved no full recovery, the binary non-recovery indicator is coded as 1. New symptoms occur during approximately 10–25% of crashes (Q5 = 2). The total score is 2 + 10 + 3.57 + 5.00 + 2 = 22.57 out of 54.
Consistency requirements and non-evaluable responses
Before calculation, entries in Questions 3 and 4 must be non-negative integers, and the total number of crashes reported in Question 3 must equal the total reported in Question 4. Both totals should also be broadly compatible with the crash-frequency category selected in Question 1. Because Question 1 partly refers to crash frequency over time, whereas Questions 3 and 4 count distinguishable crash episodes, clear discrepancies are reviewed as indicators of potential non-evaluability rather than assessed using rigid numerical cut-offs. Raw crash counts are used only to weight the category distributions and therefore cannot increase Questions 3 or 4 beyond their defined ranges. Incomplete or internally inconsistent responses are classified as non-evaluable; no aggregate score is calculated, and no maximum value is imputed.
Special rules
If no crashes are reported and Question 1 is answered “never”, Questions 3 and 4 are assigned a value of 0. The value 0 therefore denotes the absence of crashes and is a structural category rather than a severity level; where crashes are present, category values begin at 1. If Question 1 is answered “continuously” and no distinguishable episodes are reported, Questions 3 and 4 cannot be calculated. Such responses are classified as a continuous, non-episodic course and reported separately; an aggregate score is not computed and a maximum value is not imputed, as this would not be methodologically justified. Because Question 1 contributes its maximum value only when “continuously” is selected, the upper end of the nominal range is reached only where distinguishable episodes are nevertheless reported.
Measurement assumptions
This scoring framework is provisional and documents the intended scoring approach at the instrument-development stage. Questions 3 and 4 are calculated as frequency-weighted mean category scores, whereas the aggregate VPEM score is the unweighted sum of the component scores. Because the components have different score ranges and Question 2 contributes five separate domain scores, their potential contributions to the aggregate score are not equal. The framework uses consecutive integer coding for ordered response categories. Severity and recovery duration are summarised separately because the questionnaire does not link the severity of an individual crash to its recovery duration. The dimensional structure of the questionnaire, the suitability of combining the component scores into an aggregate score, their relative contributions, and score interpretation require psychometric evaluation.
Last edited:
