MATERIALS AND METHODS

A total of 1325 women aged 15–54 participated in a large community based reproductive heath survey in 20 villages in the Farafenni area of the Gambia, which is described elsewhere.⁴ Syphilis testing was included as part of the survey. For this testing a field laboratory was set up in each village. All necessary equipment and consumables were transported daily to the site being surveyed. A portable generator provided electricity for the centrifuge and shaker. All reagents, RPR kits, rapid syphilis test strips, and samples collected were kept in a cool box that was replenished with ice packs daily.

For RPR testing a 10 ml venous blood sample was collected into a plain vacutainer, allowed to clot for about 15 minutes, and centrifuged for 10 minutes at 2000 g. A standard RPR 18 mm circle card test (Quorum Diagnostics) was carried out, mixing one drop of serum with one drop of RPR reagent, mixing on a shaker for 8 minutes, and read in the best available light. Positive and negative control sera were included in each day’s testing.

For RST testing, 100 μl of serum was aliquoted into a fresh serum tube. A RST strip was removed from the foil pouch and added to the tube. This was left for 15 minutes and the results read, according to the manufacturer’s guidelines. Where the sera reacts with the Treponema pallidum recombinant antigens in the strip a double pink line results and these were read as positive. Sera were regarded as negative if only the single control line was visible. If no pink line occurred the test was discarded as invalid.

At the end of each day samples were returned to the well equipped laboratory at the MRC Laboratories field station at Farafenni, which has a constant power and water supply. Samples were received, catalogued, and stored frozen. Within 1 week RPR and TPHA testing was carried out using standard techniques. The same laboratory assistant performed the tests in the laboratory as in the field but did not have access to field results.

The TPHA was a standard Fujeriebo test (Mast Laboratories, UK). Sera were diluted to 1/160 and mixed with sensitised and unsensitised red blood cells. This was read after 1 hour at room temperature. Sera were considered positive if agglutination occurred with sensitised cells only. Samples were considered void if agglutination occurred with unsensitised cells.

RESULTS

From the 1325 serum samples obtained, 1295 samples were RPR tested in the field, and all 1325 samples were tested in the laboratory; the 30 women not screened in the field were revisited and offered treatment if positive results were subsequently found. Field screening in these 30 women was not carried out owing to logistical difficulties, either generator or equipment failure or lack of consumables in the field laboratory. In the field 76 samples were read as RPR positive (5.9%). In the laboratory, 47 samples were read as RPR positive (3.5%), 16 as weakly positive (1.2%), and 40 (3.1%) were RPR/TPHA positive. Using the rapid syphilis strips 92 samples were positive; of these 33 were RPR positive, 51 TPHA positive, and 30 positive by both tests in the laboratory.

The performance of the field RPR and RST tests against serologically active syphilis (defined by laboratory RPR positive and TPHA positive) is shown in table 1. Calculations of sensitivity and specificity against this standard and of positive and negative predictive values for this population (where the prevalence of active syphilis is about 3%) are also presented.

Table 2 directly compares RPR results on samples tested both in the field and in the laboratory. There was agreement in 1207 negative samples and 35 positive samples (95.9%); 41 (3.2%) samples were positive in the field but negative in the laboratory. A further 12 (0.9%) samples were read as negative in the field and positive in the laboratory.

A comparison of the rapid syphilis test with the TPHA test, which is a test also based on T pallidum antigens, is shown in table 3. All 29 TPHA positive/RST negative samples remained RST negative on repeat testing.

DISCUSSION

The last guidelines for serological diagnosis for syphilis, produced by the World Health Organization,⁵ recommended the use of a cardiolipin test such as the RPR and the TPHA for screening purposes. These guidelines are still in place in many countries but with the development of sensitive and specific treponemal antigen based enzyme immunoassays (EIA) they have been extended in some countries, including the United Kingdom.⁶ Results using these assays now suggest a sensitive EIA, as a single screening test would give similar results to RPR and TPHA in combination. In areas where ELISA technology is readily available screening can be automated and more standard reliable results obtained.⁷ In addition, the FTA-abs, previously considered as the “gold standard” confirmatory test, has been shown to have a poor specificity and is being superseded by newer, easier antitreponemal IgM ELISAs.⁸ In this study we use the traditional standard of laboratory RPR and TPHA positive to indicate active syphilis and we compared this standard both with RPR carried out in field conditions that are typical of many developing country health centres and with the newly developed RST strip.

There is little information available on the performance of syphilis tests under field conditions, although decentralised syphilis prevention programmes in antenatal clinics using RPR testing has been recommended.⁹ A study in an antenatal clinic in South Africa¹⁰ using RPR testing showed that clinic testing had a sensitivity of 92.8% and a specificity of 96.3% when compared to reference laboratory results, which led to its recommendation for use. In contrast, Van Dyck et al¹¹ using the RPR teardrop test in field clinics found it to be 69.7% sensitive and 96.5% specific compared to standard RPR/TPHA tests and concluded it was not reliable in these circumstances. The intermediate sensitivity of 77.5% we found here for field RPR testing is closer to that found in the latter study.

The new RST was easier to use and easier to interpret than the RPR test especially where field conditions were difficult. The RPR and RST performed similarly as field screening tests for diagnosing active syphilis, although the field RPR was a little more sensitive and slightly more specific than the RST. Neither test predicted well the presence of active syphilis. Table 1 shows that the chance of someone with a positive test being confirmed as having serological active syphilis in the laboratory was less than 50% for both tests. These predictive values are influenced by the low prevalence of active syphilis (3%) in this population. To extrapolate to settings of differing prevalence assuming that sensitivity and specificity remain the same we present calculations based on the likelihood ratio for a positive test (LR(+) = sensitivity/(1 − specificity)) and for a negative test (LR(−) = (1 − sensitivity)/specificity). These can be used to estimate the predictive values of the tests in settings of differing prevalence of active syphilis.¹² From table 1 the LR(+) for the RST is 15.6 and that for the field RPR is 21.6. The LR(−) is 0.26 for the RST and 0.23 for the field RPR. Table 4 shows the calculated predictive values of the tests at different active syphilis prevalences. Thus, our data suggest that where syphilis is a major public health problem (10–15% prevalence), and laboratory facilities are limited, a positive test for either field RPR testing or RST would imply a 65–80% probability of active syphilis, and a decision to treat these patients for active syphilis is straightforward. Of more concern is the unreliability of the negative test: 3–4% of subjects with a negative test would escape detection and treatment, reflecting the requirement for screening tests to be as sensitive as possible.

Discrepant samples, for RPR testing, were more likely to be positive in the field and negative in the laboratory (table 2). The field conditions were often poor, the temperature was between 35–42°C, the atmosphere was very dusty, and the light available poor. RPR reagent has been found to be temperature sensitive.^13,14 Reactivity decreases below 23°C and increases above 29°C, and even though the reagent was kept in a cool box between samples the ambient temperature was such that the testing was at temperatures above 29°C. This, we think, explains why more positives were found with the field RPR, though other factors will have contributed to the discordant results. The tests had a tendency to dry within the mixing time and although they were performed under a cover the amount of dust in the atmosphere must also have affected results. Poor light for reading did not help, particularly with samples that were difficult to determine. The laboratory assistants who undertook the tests were well trained and conscientious, used to performing RPR tests, and were provided with controls; this would not always be the case in rural health centres. In contrast, the conditions in the laboratory were good, air conditioned, and had minimal dust and good light. Reproducibility was satisfactory when a subset of samples was retested by RPR in the laboratory for quality control purposes.

A previous evaluation of the RST¹⁵ in a UK laboratory showed a much better performance, with an overall sensitivity of 94%. However, both conditions for testing and the population studied were different from those evaluated in this study. Like the TPHA the RST is based on specific treponemal reactivity, in this case to a recombinant 47 kDa protein antigen. Though this recombinant antigen is specific to T pallidum, post-translational processing of the protein in vivo or HLA haplotype restricted responsiveness could lead to lack of reactivity during infection. Conversely, the large number of treponemal antigens exposed during the TPHA test could increase the chance that some cross reaction with other organisms may occur. This may explain some of the discrepancies we observed between the two tests (table 3). Perhaps a mixture of several recombinant test antigens might overcome some of these problems. It is traditional to call for appropriate cheap and better syphilis testing, but it is far from clear how we should decide whether one test for syphilis is better than another, given these kinds of difficulties with defining a sensitive and specific gold standard. One approach may be to compare the utility of different tests by evaluating different testing strategies directly against public heath outcome.¹⁶

Similar questions arise in decision making over screening and testing strategies. If the traditional gold standard, RPR/TPHA positive, means active syphilis in this setting then the field RPR and RST have positive predictive values of only 30–40%. This could be acceptable in higher prevalence settings but some overdiagnosis and overtreatment of patients is bound to occur. There are some hazards to the patients here: suggestions have been made that administering STD treatment and partner notification can cause domestic disruption and even violence in rural societies in similar settings to ours.¹⁷ Social as well as laboratory and treatment issues would need to be taken into account when looking at the benefits of screening policies. If screening tests are used where conditions for testing are poor, some means of communicating the diagnostic doubt is necessary in patient counselling and partner notification policies.

Our data also suggest that negative field testing does not reliably exclude active syphilis in high prevalence situations; hence even in an ideal situation where all mothers are screened adverse birth outcomes could still be anticipated. This situation is unsatisfactory but until better alternatives can be found or a better gold standard for syphilis testing can be achieved these problems must be taken into account when using syphilis screening in public health circumstances.