Logistic regression models (formulas) for sialometrical and sialochemical variables combined as tests for SS. The probability of a subject having SS is represented in logistic regression formula as “p”. A p value <0.5 is considered positive for SS. For comparison with cut off point approach see tables 6 and 7
| Formulas for classifying SS test (logistic regression approach) | Specificity | Sensitivity | LR | No | PPV | NPV | |
|---|---|---|---|---|---|---|---|
| LR, likelihood ratio; No, number of cases included in the analysis out of total (observation group n=100). Cases were rejected when data were missing in any of the variables in the formula; PPV, positive predictive value; NPV, negative predictive value; LNR, loge; SM/SL, submandibular/sublingual. | |||||||
| Parotid sialochemical variables | |||||||
| 10. | Parotid sodium (X1) and parotid chloride (X2) | ||||||
| (parotid phosphate removed by analysis) | 0.93 | 0.75 | 11 | 57 | 0.91 | 0.79 | |
| Formula: LNR (p)=2.8603−0.2044X1−0.0547X2 | |||||||
| SM/SL sialochemical variables | |||||||
| 11. | SM/SL chloride (X1) and SM/SL phosphate (X2) | ||||||
| (SM/SL sodium removed by analysis) | 0.90 | 0.72 | 7 | 49 | 0.81 | 0.85 | |
| Formula: LNR (p)=0.9882−0.1411X1+0.6928X2 | |||||||
| Parotid and SM/SL sialochemical variables | |||||||
| 12. | Parotid sodium (X1) and parotid chloride (X2) and SM/SL phosphate (X3) (SM/SL sodium, SM/SL chloride, and parotid phosphate removed by analysis) | 0.96 | 0.80 | 20 | 43 | 0.92 | 0.90 |
| Formula: LNR (p)=1.6479−0.2274X1−0.1268X2+1.3265X3 | |||||||
| Sialometrical variables | |||||||
| 13. | Stimulated SM/SL (X1) flow and parotid lag phase (X2) | ||||||
| Formula: LNR (p)=−0.1546+1.1286X1−0.0050X2 | 0.60 | 0.67 | 2 | 100 | 0.68 | 0.58 | |
| Sialometrical and sialochemical variables | |||||||
| 14. | Parotid sodium (X1) and stimulated SM/SL flow (X2) | ||||||
| (parotid chloride and lag phase removed by analysis) | 0.90 | 0.76 | 8 | 90 | 0.89 | 0.78 | |
| Formula: LNR (p)=0.6765−0.2353X1+3.3929X2 | |||||||