Table 2 showed that there was no significant difference in subjects’ demographics between the real ship motion and the vertical oscillation trials. High reliability scores of MSSQ were obtained in both trials (test-retest r = 0.92 and 0.90). The ship motion trial showed significantly higher Graybiel scores (Wilcoxon test: Z = –3.381, P < 0.01), but lower percentage of INSUS subjects (χ2 = 4.89, P < 0.05) than the vertical oscillation trial. The percentage distribution of Graybiel score-defined and nausea latency-defined seasickness susceptibility (SUS and INSUS subjects) showed no difference in the vertical oscillation trial. Incidence of nausea was significantly higher than incidence of vomiting in both trials (χ2 = 9.44 and 5.78, P < 0.01 and 0.05), while no difference was observed in nausea or vomiting incidence between two trials.
The levels of the MSSQ and its subscale scores showed no difference between two trials (Table 3). Childhood item total scores (MSV child + MSN child) was significantly higher than the adulthood ones (MSV adult + MSN adult) in both trials (ship motion: 10.43 ± 12.51 vs. 7.59 ± 9.22, Wilcoxon test: Z = –3.475, P < 0.01; vertical oscillation: 9.91 ± 9.95 vs. 7.27 ± 9.30, Wilcoxon test: Z = –2.769, P < 0.01). Significant positive correlations of the MSSQ and its subscales with the Graybiel score were observed in both trials. In the vertical oscillation trial, there was a significant negative correlation between Graybiel score and nausea latency (r = –0.840, P < 0.001). The MSSQ and the MSV subtotal (and its subitems), but not the MSN subtotal (or its subitems), showed negative correlations with the nausea latency (Table 3).
In the real ship motion trial, all the MSSQ, and its subscales and subitems, except for the MSN child and the MSN subtotal, had significant and similar AUC values of the ROC curve established for predicting the SUS and the INSUS subjects defined by the Graybiel score (ROC analysis: P < 0.05 or 0.01; Table 4). The MSN child and the MSN subtotal showed no predictive ability for the INSUS subjects (ROC analysis, P > 0.05), while the MSN child also had lower AUC for SUS and INSUS prediction than the MSV adult (0.620 vs. 0.699 Z = 1.975, P < 0.05; Figure 1A) and the MSV child (0.651 vs. 0.558, Z = 2.66, P < 0.01, Figure 1C), respectively. Although the correlation of the MSN subtotal with the MSSQ total was high (r = 0.943), the MSN subtotal had significantly lower AUC for INSUS discrimination than the MSSQ total and the MSV subtotal (AUC: 0.587 vs. 0.647 and 0.644, Z = 2.91 and 2.59, P < 0.01; Figure 1D). No significant difference was observed in AUC values among the MSV subtotal, the MSN subtotal and the MSSQ total for SUS prediction (Figure 1B).
In the laboratory vertical oscillation trial, there was no significant difference in AUC values for prediction of SUS among the child and adult subitems or among the MSV subtotal, the MSN subtotal and the MSSQ total. (Figure 2A and B for Graybiel score-defined SUS and Figure 3A and B for nausea latency-defined SUS). The MSN child showed no discriminating ability for the INSUS subjects defined by either Graybiel score or nausea latency in contrast to other MSSQ subscales and subitems which had significant AUC for prediction of both the SUS and the INSUS subjects (ROC analysis: P < 0.05, 0.01 or 0.001, Table 4). The AUC values of the MSN child for discriminating INSUS subjects were significantly lower than those of the MSV child (Graybiel score-defined: 0.592 vs. 0.747, Z = 2.57, P < 0.05; Figure 2C; Nausea latency-defined: 0.637 vs. 0.803, Z = 2.22, P < 0.05; Figure 3C). Similar to the ship motion trial, the vertical oscillation trial also showed lower AUC in the MSN subtotal than that of the MSSQ total and the MSV subtotal in predicting INSUS subjects defined by the Graybiel score (AUC: 0.696 vs. 0.761 and 0.788, Z = 2.784 and 2.10, P < 0.01 and 0.05; Figure 2D), despite a high correlation of the MSN subtotal with the MSSQ total (r = 0.95). Lower AUC in the MSN subtotal than the MSSQ total was also observed when INSUS subjects were defined by nausea latency (AUC: 0.691 vs. 0.758, Z = 2.145, P < 0.05 and 0.01; Figure 3D). In contrast, the MSV subtotal did not differ from the MSSQ total in prediction of both SUS and INSUS subjects defined by the two criteria. Relative to the real ship motion trial, the laboratory vertical oscillation trial showed significantly higher predictive efficacy of the MSSQ total for the Graybiel score-defined SUS subjects (0.840 vs. 0.686, Z = 2.126, P < 0.05) as a result of higher AUC in both the MSV subtotal (0.840 vs. 0.690, Z = 2.519, P < 0.05) and the MSN subtotal (0.806 vs. 0.651, Z = 2.582, P < 0.01).