Validity Testing of a Flowchart: Sensitivity and Specificity

The sensitivity of a flowchart is the proportion of infections detected and treated when applying the flowchart to patient management. A high sensitivity implies a large proportion of the infections will be detected and treated, so the cure rate will be higher, and only a small proportion of infections will be missed. Missed cases place the infected person at a continued risk of more serious complications and result in further disease dissemination.

The specificity of a flowchart is the proportion of patients without infection that are not treated. A high specificity implies that patients without infection are not being told they have an infection and are not being treated. A low specificity means too many patients without infection are being treated. This has social, biological and financial implications. Socially, people without STDs are being told to have their partners treated; biologically, over use of antibiotics may lead to resistance strains; and financially expensive drugs are being wasted.

Work is ongoing on the identification, development, and use of simple tests that could improve the sensitivity and specificity of syndromic diagnosis, including simple, rapid tests to identify oxidase-producing organisms (e.g. gonococci) in urethral discharge and tests to detect leukocytes in cervical and urethral specimens which would be suggestive of chlamydial and/or gonococcal infections.

The study population for validity testing has to be as representative as possible of the population in which a flowchart will be used. As long as enough STD patients are consulted to reach the necessary sample size within a reasonable time and a reference laboratory is available for the gold standard tests, a primary health care setting is suitable. In general, sophisticated laboratory tests are needed for gold standard diagnosis. If some patients consulting at the primary health care level have already taken drugs, it is important not to exclude them from the validity study.

When testing the validity of flowcharts, it is preferable to have two examination rooms. In one room, a health care worker can follow the directives of the flowchart and make a therapeutic decision. Then the patient can be asked to enter a second room, where a more systematic examination can be done and where specimens will be taken for the gold standard laboratory tests. Since a speculum examination is sometimes not included in STD flowcharts, using such a setting will ensure that the therapeutic decision is not based on the findings of the speculum examination.

The likely outcomes of diagnoses made using a flowchart can be simulated retrospectively using data from history taking, physical examination and laboratory tests. This desk exercise can be very useful for evaluating modifications to flowcharts without repeating studies. For example, it could be used to assess what would happen to the specificity of a flowchart if the gram stain examination were omitted.