Normative Data

Descriptive and inferential statistical analyses were carried out in R. ROC curves were generated using the pROC package.

There were no significant differences in age (p-value: 0.4223), or sex distribution (p-value: 0.09413) between the three diagnostic categories in the sample (Table 1). Educational attainment was nominally significant (p-value: 0.015), approximately 15% fewer of the cognitively impaired subjects reported continuing education past high school compared with the two other categories.

To estimate the sensitivity and specificity of the P-Cog in classifying cognitive impairment compared with the MoCA, we calculated three receiver operating characteristic (ROC) curves for each test by including A) diagnoses of cognitive impairment versus no cognitive impairment. B) diagnoses of cognitive impairment and mild cognitive impairment versus no cognitive impairment. C) mild cognitive impairment versus no cognitive impairment. For category A, the area under the ROC curve (AUC) for the P-Cog was 0.9929 [95% CI: 0.9854-1] and MoCA was 0.9845 [95% CI: 0.9714-0.9976]. Difference between the AUCs was not significant, p-value = 0.1929. For category B, the P-Cog AUC was 0.9169 [95% CI: 0.8846-0.9491] and the MoCA 0.9185 [95% CI: 0.8865-0.9505]. The difference between the AUCs was not significant, p-value = 0.9138. Considering the MCI alone, the P-Cog AUC was 0.7985 [95% CI: 0.7282-0.8689] and the MoCA AUC 0.8159 95% [CI: 0.7487-0.883]. The difference in AUC was not significant, p-value = 0.6226. (Figures 1 and 2).

The validity of the P-Cog scores was evident from the high correlation with the MoCA (r = 0.93 [95 CI 0.91 - 0.94], p < 2.2e-16). Time to completion was also correlated between the P-Cog and MoCA (r = 0.51 [95 CI 0.41 - 0.61] p = 4.441e-16).P-Cog time was associated with diagnostic category