Development and Psychometric Testing of a New Short-Form of the Premenstrual Symptoms Questionnaire (PSQ-S)

Abstract

Purpose: The Premenstrual Symptoms Questionnaire (PSQ) is a patient-reported outcome measurement tool for premenstrual symptoms. Although the PSQ is a very useful tool with 14 items to screen for a wide variety of premenstrual symptoms, not everyone will respond to all the questions. Fewer questions would be less burdensome on the respondents. We aimed to develop and analyze the psychometric properties of a short-form of the PSQ (PSQ-S).

Patients and methods: The study participants were from an earlier study with a sample of 1388 female students. We reanalyzed data collected from 922 students with regular menstrual cycles who completed the PSQ, premenstrual dysphoric disorder (PMDD) scale, Somatic Symptom Scale-8 (SSS-8), and numerical rating scale (NRS) for menstrual pain. First, we selected nine items for the PSQ-S based on the results of the corrected item-total correlation analysis. The PSQ-S was then analyzed for reliability and validity using the classical test theory. Moreover, item response theory was applied to test the psychometric properties of the PSQ-S.

Results: Cronbach's α for the PSQ-S score was 0.93. Principal component factor analysis revealed a one-factor model. Confirmatory factor analysis showed that the one-factor model was a good fit. The PSQ-S total score was strongly correlated with the PSQ total score, PMDD scale score, and SSS-8 score (r=0.978, 0.854, and 0.648, respectively) and moderately correlated with the NRS (r=0.437). Item response theory analyses showed that the constructs and items of the PSQ-S had satisfactory discrimination, difficulty parameters, item information curves, and test information curves. Receiver operating characteristic curve analysis revealed a cut-off score of ≥22 for suspected premenstrual disorders based on the total PSQ-S score.

Conclusion: The PSQ-S, consisting of nine items from the PSQ, had sufficient reliability and validity and could be a convenient assessment tool for premenstrual symptoms in routine clinical practice.

Keywords: item response theory; menstruation; premenstrual disorders; validity testing.

Figure 1

Flow chart of the study.

Figure 2

Scree plot for the PSQ-S.

Figure 3

One-factor model of the PSQ-S (A) and the PSQ (B) and standardized parameter estimates.

Figure 4

Item characteristic curves for the PSQ-S. Theta at the horizontal axis indicates the trait level (severity of premenstrual symptoms). The vertical axis shows the probability for an answer in each of the PSQ-S questionnaire items. Four curves (from 1 to 4) indicate each possible response: curve 1, not at all; curve 2, mild; curve 3, moderate; curve 4, severe. Curve 1, the leftmost curve, shows the probability that each individual would choose 1 out of 4 ratings at different trait levels. The curve shows that the lower the trait level, the higher the probability of choosing 1. Curve 4, the rightmost curve, shows the probability that each individual would choose 4 out of 4 ratings at different trait levels. The curve shows that the higher the trait level, the higher the probability of choosing 4. Curves 2 and 3 between them are unimodal, indicating that at a certain range of trait levels, 2 or 3 is chosen.

Figure 5

Item information curves for the PSQ-S. Theta at the horizontal axis indicates the trait level (severity of premenstrual symptoms). The vertical axis shows the amount of information for an answer in each of the PSQ-S questionnaire items. The higher the curve, the greater the amount of information in a narrower range of trait levels and, consequently, the more highly discriminating the item is. On the other hand, items with low discrimination power show a flat curve across a wide range of trait levels, which means that they are less informative.

Figure 6

Test information curve and standard error of measurement for the PSQ and PSQ-S. Theta at the horizontal axis indicates the trait level (severity of premenstrual symptoms). The vertical axis shows the amount of summative test information for the PSQ and the PSQ-S.