Objectives: To assess the effectiveness of a training intervention on doctors’ diagnoses of oral manifestations of HIV.
Design: Prospective uncontrolled study
Setting: Outpatient medical clinic in rural Uganda
Intervention: Didactic teaching, seminars and calibration with patients by dental students.
Methods: Doctors’ diagnoses after the intervention were compared against the trainers’ diagnoses and patients’ HIV status in the short term, and patients’ HIV status only in the long term.
Main outcomes: Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV) and likelihood ratios (LR) using HIV status as gold standard.
Results: Outcomes for the doctors’ diagnoses immediately after the intervention compared with trainers’ diagnoses for the same sample (n=79, of which 44 were HIV positive):
|
Group |
Sensitivity |
Specificity |
PPV |
NPV |
LR |
|
Dental student |
0.23* |
0.97 |
0.83 |
0.51 |
7.67 |
|
Doctor |
0.14* |
1 |
1 |
0.52 |
∞ |
* p<0.001 (McNemar Test)
Over the next three months, the doctors did not diagnose any further lesions in a sample of 259 participants, of which 111 were HIV positive (sensitivity, PPV and LR all fell to 0). This reduction in sensitivity was significant (p<0.001, Fisher’s exact test)
Conclusions: Doctors can be trained in diagnosis of oral manifestations of HIV but diagnostic validity decreases with time from training. Further and sustained increases in diagnostic validity may be possible with increased or more effective training and if oral diagnosis formed an integral part of patients’ health care, rather than part of a research project.