Objectives: To estimate the positive and negative predictive values (PPVT and NPVT), sensitivity and specificity of a full and abridged screening questionnaire of physical and psychological health, using primary care doctors' (medical officers [MOs]) assessments as to whether the servicemen needed medical help as a gold standard.
Methods: From a tri-service random sample of those who completed a questionnaire, all 'screen-positive' and an equal random sample of 'screen-negative' were selected to attend their medical centre. MOs were aware that the screening was aimed at detecting psychological illness, but were blind as to the 'screen-positivity' of any serviceman. The MO completed a questionnaire that asked whether the patient needed medical help and whether s/he was previously aware of this need.
Results: 314 subjects were available for analysis. The PPVT was 47% (95% confidence interval [CI] 36–59%) for the full questionnaire and 48% (95% CI 36–60%) for the abridged questionnaire. Of those 'screen-positive' subjects whom the MO rated as needing help, one third had problems already known to the MO, regardless of the length of the questionnaire. The sensitivity and specificity of the full and abridge questionnaires were 43% and 74%, and 36% and 83% respectively. The PPVT did not vary greatly between health dimensions nor did selection of servicemen with very high scores.
Conclusions: The use of MOs as a gold standard is important because of their central role in initiating the management of any condition uncovered by a screening programme. Using MOs as a gold standard, the validity of the screening questionnaires for physical and psychological health in the military was mediocre.
