RESEARCH

Comparison of the diagnostic accuracy of a rapid immunochromatographic test and the rapid plasma reagin test for antenatal syphilis screening in Mozambique

Comparaison de la précision diagnostique d'un test immunochromatographique rapide et du test rapide à la réagine pour le dépistage prénatal de la syphilis au Mozambique

Comparación de la precisión diagnóstica de una prueba inmunocromatográfica rápida y de la prueba de reagina rápida en plasma para el cribado de la sífilis prenatal en Mozambique

Pablo J Montoya^I,1; Sheila A Lukehart^II; Paula E Brentlinger^III; Ana J Blanco^I; Florencia Floriano^I; Josefa Sairosse^IV; Stephen Gloyd^III

^IHealth Alliance International, PO Box 23, Maputo, Mozambique
^IIDepartments of Medicine/Infectious Diseases, Pathobiology, Microbiology, Periodontics. University of Washington, Seattle, WA, USA
^IIIInternational Health Program, Department of Health Services. University of Washington, Seattle, WA, USA
^IVMinistry of Health of Mozambique, Predio de governo 4o andar, Repartição de Asistencia Médica, Direcção provincial de Saúde de Sofala. Beira, Mozambique

ABSTRACT

OBJECTIVE: Programmes to control syphilis in developing countries are hampered by a lack of laboratory services, delayed diagnosis, and doubts about current screening methods. We aimed to compare the diagnostic accuracy of an immunochromatographic strip (ICS) test and the rapid plasma reagin (RPR) test with the combined gold standard (RPR, Treponema pallidum haemagglutination assay and direct immunofluorescence stain done at a reference laboratory) for the detection of syphilis in pregnancy.
METHODS: We included test results from 4789 women attending their first antenatal visit at one of six health facilities in Sofala Province, central Mozambique. We compared diagnostic accuracy (sensitivity, specificity, and positive and negative predictive values) of ICS and RPR done at the health facilities and ICS performed at the reference laboratory. We also made subgroup comparisons by human immunodeficiency virus (HIV) and malaria status.
FINDINGS: For active syphilis, the sensitivity of the ICS was 95.3% at the reference laboratory, and 84.1% at the health facility. The sensitivity of the RPR at the health facility was 70.7%. Specificity and positive and negative predictive values showed a similar pattern. The ICS outperformed RPR in all comparisons (P<0.001).
CONCLUSION: The diagnostic accuracy of the ICS compared favourably with that of the gold standard. The use of the ICS in Mozambique and similar settings may improve the diagnosis of syphilis in health facilities, both with and without laboratories.

Keywords: Syphilis serodiagnosis; Prenatal diagnosis; Mozambique (source: MeSH, NLM).

RÉSUMÉ

OBJECTIF: Les programmes de lutte contre la syphilis dans les pays en développement se heurtent à l'insuffisance des services de type analytique, aux retards de diagnostic et aux doutes que suscitent les méthodes actuelles de dépistage. L'étude visait à comparer la précision diagnostique d'un test immunochromatographique rapide en bandelette (ICS) et du test rapide à la réagine (RPR) avec la méthode combinée de référence (RPR, test d'hémagglutination du tréponème pâle combiné au test d'immunofluorescence directe dans un laboratoire de référence) pour le dépistage de la syphilis au cours de la grossesse.
MÉTHODES: L'étude a pris en compte les résultats des tests de 4789 femmes ayant subi leur première visite prénatale dans l'un des six centres de santé de la Province de Sofala au centre du Mozambique. Elle a comparé la précision diagnostique (sensibilité, spécificité et valeurs prédictives positives et négatives) des tests de type ICS et RPR pratiqués dans les centres de santé et de l'ICS effectué au laboratoire de référence. Des comparaisons ont également été réalisées dans des sous-groupes définis en fonction du statut à l'égard du VIH et du paludisme.
RÉSULTATS: Pour la syphilis active, la sensibilité du test ICS était de 95,3 % au laboratoire de référence et de 84,1 % dans le cadre d'un centre de santé. La sensibilité du test RPR pratiqué dans un centre de santé était de 70,7 %. Pour la spécificité et les valeurs prédictives positives et négatives, on a observé des différences analogues. L'ICS a donné de meilleurs résultats que le RPR dans toutes les comparaisons (p<0,001).
CONCLUSION: La précision diagnostique de l'ICS n'est pas très éloignée de celle de la méthode de référence. L'utilisation de ce test au Mozambique et dans des contextes similaires pourrait améliorer le diagnostic de la syphilis dans les centres de santé, qu'ils soient dotés ou non d'un laboratoire.

Mots clés: Séro-diagnostic syphilis; Diagnostic prénatal; Mozambique (source: MeSH, INSERM).

RESUMEN

OBJETIVO: En los países en desarrollo los programas de control de la sífilis tropiezan con la falta de servicios de laboratorio, las demoras del diagnóstico y las dudas existentes respecto a los actuales métodos de cribado. Decidimos comparar la precisión diagnóstica de una prueba con tira inmunocromatográfica (IC) y de la prueba de reagina rápida en plasma (RRP) con la prueba de referencia combinada (RRP, hemaglutinación de Treponema pallidum e inmunofluorescencia directa en un laboratorio de referencia) para la detección de la sífilis en el embarazo.
MÉTODOS: Consideramos los resultados analíticos correspondientes a 4789 mujeres que realizaron su primera visita prenatal en alguno de los seis centros de salud elegidos en la provincia de Sofala (Mozambique central). Comparamos la precisión diagnóstica (sensibilidad, especificidad, y valores predictivos positivo y negativo) de la IC y la RRP realizadas en los centros de salud y la IC realizada en el laboratorio de referencia. Hicimos también comparaciones de subgrupos en función del estado serológico respecto al VIH y la malaria.
RESULTADOS: Para la sífilis activa, la sensibilidad de la IC fue de un 95,3% en el laboratorio de referencia, y de un 84,1% en el centros de salud. La sensibilidad de la RRP en este último fue del 70,7%. La especificidad y los valores predictivos positivo y negativo mostraron diferencias similares. La IC superó a la RRP en todas las comparaciones (P < 0,001).
CONCLUSIÓN: La precisión diagnóstica de la IC fue buena en comparación con la prueba de referencia. En Mozambique y en otros entornos semejantes, la IC permitiría mejorar el diagnóstico de la sífilis en los centros de salud, tengan o no laboratorio.

Palabras clave: Serodiagnóstico de la sífilis; Diagnóstico prenatal; Mozambique (fuente: DeCS, BIREME).

Introduction

Syphilis is an important cause of perinatal morbidity and mortality in resource poor settings. Adverse infant or fetal outcomes arise in 50–80% of pregnancies that survive beyond 12 weeks of gestation,^1–3 especially if pregnancy coincides with the early stages of infection.^4–6 Syphilis is also a substantial cause of adult morbidity and might increase the risk of human immunodeficiency virus (HIV) transmission.^{7, 8}

Syphilis control is facilitated by the availability of inexpensive and sensitive diagnostic tests and effective and affordable treatment.^{7, 9–12} Antenatal screening and treatment for the disease is highly cost-effective as a means to reduce fetal and infant morbidity and mortality,¹³ and furthermore, such measures could contribute to reduced HIV transmission.^{8, 14}

Every year, about 1.6 million pregnant women with syphilis remain undiagnosed in sub-Saharan Africa, including more than one million attending antenatal care.¹⁵ Syphilis — as diagnosed by a positive rapid plasma reagin (RPR) test — at the first antenatal visit in Mozambique has a prevalence of about 10% for the country as a whole and 15% for Sofala Province, where our study was conducted.¹⁶

Scarcity of laboratory services, staff, and training as well as late diagnosis and treatment have hampered efforts to prevent congenital syphilis in Mozambique.^17–19 Furthermore, doubts have also been raised about the accuracy of the currently used syphilis screening tests, such as the RPR, especially in populations with a high prevalence of HIV²⁰ and malaria.²¹

The recent introduction of rapid immunochromatographic strip (ICS) to screen for treponemal infection would allow syphilis to be both diagnosed and treated in a single visit.^{22, 23} Antenatal clinic nurses can do the ICS test in health facilities without a laboratory. Unlike RPR reagents, the ICS can be stored at room temperature and does not require special procedures. A price reduction to less than US$ 0.50 per test makes ICS a feasible option for use in settings with scarce resources.

In this study, we aimed to: 1) compare the diagnostic accuracy²⁴ of ICS and the RPR with a composite gold standard for the detection of syphilis in pregnancy (Treponema pallidum haemagglutination assay (TPHA), RPR and direct immunofluorescence stain done at a reference laboratory); 2) compare diagnostic accuracy of ICS and RPR in women with and without HIV or malaria; 3) compare results from the reference laboratory with those from field tests; and 4) describe operational issues that affect the diagnostic accuracy of the tests done in health facilities.

Materials and methods

We recruited participants from a population of pregnant women attending their first antenatal visit at one of six typical health facilities in Sofala Province, Mozambique. We chose health facilities that had a laboratory, that were in a region with high prevalence of syphilis, HIV and malaria, and that had a high number of antenatal patients. Infrastructure at the health facilities was basic and unlike the reference laboratory, buildings at the health facilities did not have air conditioning. Mozambique Ministry of Health staff at the six facilities received 3 days' training on the study and laboratory procedures.

After providing written informed consent, participants answered a questionnaire that included questions about their obstetric and syphilis history. They had an expanded physical examination, which included a search for mucocutaneous lesions suggestive of syphilis, and they provided capillary and venous blood samples.

Staff at the antenatal clinic collected 20 µl of capillary blood with EDTA or non-EDTA capillary tubes from a finger puncture for the SD BioLine Syphilis 3.0 ICS test (Standard Diagnostics Inc., Republic of Korea). An additional blood drop from the same finger prick was taken for a thick blood smear to detect malaria.

Serum from a 5 ml venous blood sample was used for the RPR at the health facilities and the ICS, RPR and TPHA tests (RPR and TPHA from Neomedic Ltd, Healthease, Sea Cow Lake, South Africa) at the Beira Central Hospital reference laboratory (Fig. 1).

We used a solid phase treponemal immunochromatographic assay that provides qualitative detection of antibodies directed towards three T. pallidum recombinant antigens; results are obtained after 5–20 minutes. We chose the RPR as our comparison test because it is widely used in syphilis screening.^{6, 9, 25–29}

Samples of genital mucocutaneous lesions were smeared on glass slides and sent to the University of Washington, Seattle, WA, USA to be tested with use of direct immunofluorescence stain for T. pallidum (ViroStat, Portland, ME, USA). All tests were done in accordance with manufacturers' recommendations. Frequent supervision and assessment took place to ensure valid and reliable results.

Nurses at the antenatal clinics did the initial ICS (hereinafter ICS_HF) during the patient's visit. At the health facilities, a laboratory technician who was unaware of the results of the ICS_HF did a first qualitative RPR test (hereinafter RPR_HF).

At the reference laboratory, technicians who were unaware of the results of previous tests repeated the ICS and RPR tests using serum and the same kit lots as were used at the health facility (hereinafter ICS_Ref and RPR_Ref). The TPHA was performed as a confirmatory test, and a quantitative RPR was done on serum that had any degree of reactivity in the treponemal (ICS and TPHA) and non-treponemal (RPR) tests.

At the Beira reference laboratory, thick blood smears were stained and read quantitatively for malaria (parasite count per 500 leukocytes adjusted for a presumed leukocyte count of 8000 per µl).

The syphilis status of each participant was defined by a gold standard composed of the TPHA and RPR_Ref, and a direct immunofluorescence stain. Possible combinations were: "no syphilis" for TPHA-negative/RPR-negative results; "active syphilis" for TPHA-positive/RPR-positive results; "old or treated syphilis" for TPHA-positive/RPR-negative results, and biological false-positive for TPHA-negative/RPR-positive results. Patients with positive immunofluorescence stain for T. pallidum specimens were diagnosed with primary syphilis, irrespective of other laboratory results. All participants with a positive TPHA test result were included in the "all syphilis" serological status group, irrespective of the RPR result.

Data about HIV status were obtained from patients who underwent voluntary counselling and testing for HIV infection as part of routine antenatal care. HIV-positive patients were referred to specialized HIV clinics, as is the normal procedure in Mozambique.

Syphilis was treated with benzathine penicillin and malaria with sulfadoxine-pyrimethamine or chloroquine, dependant on gestational age and in accordance with the clinical protocols of the Mozambique Ministry of Health.^{30, 31}

Data were analysed with Stata 7.0. We assumed that with a sample size of 4000 women, we would obtain a precision standard error (SE) of ±3% for the diagnosis of active syphilis. We compared the performance of the ICS_Ref, ICS_HF and RPR_HF tests with the gold standard for all women, stratifying by syphilis serological group, HIV status and malaria status. Differences between subgroups were calculated using c² or Fisher's exact test as appropriate. We calculated crude and adjusted odds ratios (OR) for factors associated with syphilis serological groups (i.e.,outcome) using logistic or multinomial logistic regression analysis.

For subgroup analysis, age was treated as a categorical variable with seven groups: <15 years, 15–19, 20–24, 25–29, 30–34, 35–39, and 40 years or older. Likewise, we treated number of pregnancies as a categorical variable with six groups: 1, 2, 3, 4, 5, and 6 or more pregnancies.

The study was approved by the Instituto Nacional de Bioética para a Saúde in Mozambique and the Human Subjects Division of the University of Washington, WA, USA.

Results

Between August 2003 and January 2004, we approached 4769 women who had not had a syphilis test during their current pregnancy about participation in our study. 276 (5.8%) declined to participate. Non-systematic interviews with these women showed that the main reasons for choosing not to participate were: an unwillingness to know their HIV status (even though HIV tests were not performed as part of the study); a lack of time to go through the data collection process; and need for partner's approval. We excluded results from 6 (0.1%) women because their serum samples did not arrive at the Beira reference laboratory. Thus, data from 4487 women were included in analysis.

Seven (0.2%) ICS_HF test results and 65 (1.4%) RPR_HF test results were lost because results were not recorded after the test had been performed, but we did not exclude these participants. If the participant had a result from either ICS_HF or RPR_HF we compared that result to the gold standard.

Of the 106 samples from genital ulcers, six specimens did not arrive at the laboratory; the remaining 100 were tested with direct immunofluorescence stain for T. pallidum. Two samples were positive, two had inadequate volume of specimen, and 96 were negative for T. pallidum.

Table 1 summarizes participants' general characteristics. After controlling for health facility, we did not note any differences in socioeconomic status or syphilis prevalence between the HIV-tested women and those who did not enrol in voluntary testing. Nearly all malaria cases were asymptomatic and caused by infection with P. falciparum. Mean asexual parasite density was 3488 parasites/µl (±9759) in parasitaemic women.

Results of the gold standard test showed 381 (8.5%) participants with active syphilis, 150 (3.3%) old or treated cases, 46 (1.0%) biological false-positives, and 2 (0.04%) primary syphilis cases. Of the active syphilis cases, 282 (74.2%) had RPR titres < 1:8 and 29 (8%) had clinical presentations compatible with primary (n = 20) and/or secondary (n = 11) syphilis. Physical manifestations consistent with primary and secondary syphilis were poorly correlated with positive direct immunofluorescence stain for T. pallidum. One TPHA-positive/RPR-positive and one TPHA-negative/RPR-negative case were confirmed as primary syphilis by direct immunofluorescence stain for T. pallidum. Of the 318 women who reported a positive syphilis history, 212 (67%) had negative TPHA and RPR results.

Table 2 shows the agreement between TPHA (done at the Beira reference laboratory), ICS and RPR results from tests performed at the reference laboratory and the health facilities. Agreement between ICS (ICS_HF and ICS_Ref) and TPHA results and between ICS_Ref and ICS_HF results was significantly higher than for the comparison of the RPR results from different sites (P<0.001).

The coefficient of agreement between ICS_Ref and TPHA was significantly higher than that for the comparison of ICS_HF with TPHA (P = 0.002). More than two-thirds of the inter-laboratory ICS discordances were negative ICS_HF and positive ICS_Ref results.

There were 66 weak positive ICS_Ref results (13% of the positive ICS_Ref results), which accounted for 40% of the ICS_Ref-to-ICS_HF discordances. We did not note any difference in the diagnostic accuracy of the ICS with the use of EDTA or non-EDTA capillary tubes (P = 0.1). Two-thirds (n = 50) of the ICS_Ref-to-TPHA discordances occurred in samples from women whose ICS_HF-to-ICS_Ref results were concordant.

Using multinomial logistic regression, we identified factors associated with discordant negative ICS_Ref and positive TPHA results: HIV infection (OR 3.5, (95% CI: 1.5–11)), and condylomatous lesions (4.6 [1.3–16.6]) for the all syphilis group, and malaria (3.6 (95% CI: 1.2–11.3) for the active syphilis group only. Positive ICS_Ref to negative TPHA results were associated with older age (1.6 (95% CI: 1.2–2.1)) and higher number of pregnancies (1.3 (95% CI: 1.1–1.5) for all syphilis cases.

Low-level reactivity RPR (titres < 1:2) were present in 100 (79.4%) of the negative RPR_HF to positive RPR_Ref discordances: we did not identify any patient variables associated with RPR_Ref-to-RPR_HF discordant results. Technicians who were unaware of previous results retested a random subsample of RPR_Ref-to-RPR_HF discordant samples, and the RPR_Ref result was corroborated in 41 of 45 (91%) of cases.

Table 3 shows characteristics of diagnostic accuracy of the ICS and RPR by syphilis serological group. The ICS, done both at the health facilities and the reference laboratory out-performed RPR with respect to sensitivity, specificity, and positive and negative predictive values. The differences in sensitivity between the ICS_Ref and ICS_HF tests, and between the ICS_HF and RPR_HF are statistically significant for all syphilis serologic groups shown in table 3 (P<0.001 in pair-wise comparisons).

With respect to specificity, there were significant differences between tests and sites for the all syphilis group (P<0.02). Negative predictive values were also significantly different across tests and testing sites for every syphilis serologic group (P<0.001). Positive predictive values were significantly different in the all syphilis group across tests and testing sites (P<0.006), and in active syphilis for the comparison between ICS_Ref and RPR_HF (P = 0.03). The sensitivity of the RPR_HF was significantly lower in 1:1 reactivity serum than in >1:2 reactivity serum (51% versus 79.2%, P<0.001). We also noted significant differences between the sensitivity of the ICS_HF after stratification by the same serum RPR reactivity levels (77.2% vs 89.2%, P = 0.003), but not for the ICS_Ref (P = 0.4).

During the first month of the study, August 2003, the sensitivity of the ICS_HF was 90% (95% CI: 80–96). However, sensitivity decreased gradually, and by the last month had dropped to 78% (95% CI: 69-86), significantly lower than at the start of the study (c² test of trends P = 0.03). This change correlated with an increase in patient numbers, higher malaria prevalence, and less frequent supervision in the clinic. We noted a similar but non-significant decline for ICS_Ref sensitivity.

Of the 1597 women with known HIV status, 51 (37%) of the 137 active syphilis and 20 (38%) of the 52 old or treated syphilis cases were HIV positive. Of the 4408 women with known malaria status, 53 (14%) of the 376 active syphilis and 19 (13%) of the 150 old or treated syphilis were positive for malaria. The use of different thresholds for definition of "significant" malaria parasite density did not change the test accuracy. There were too few malaria-infected women with fever (n = 8) to allow us to do subgroup analysis.

The ICS_Ref test was the most sensitive of the tests (P<0.01 in pair-wise comparisons), independent of malaria or HIV status. For active syphilis cases the sensitivity of the ICS_Ref was not significantly lower for women with HIV (P = 0.2), but sensitivity was affected by malaria coinfection (P = 0.01). However, we did not note the same pattern in participants with malaria for the ICS_HF results. Furthermore, the number of syphilis-positive participants who also had HIV or malaria was too small to allow us adequate power to assess differences in diagnostic accuracy in coinfected and non-coinfected groups.

The specificity of the ICS_Ref test in the old or treated syphilis group was higher for participants who were HIV negative (92.2%) than for HIV positive (86.4%) patients (P<0.001). The negative predictive value of the ICS_Ref was significantly higher in the all syphilis group for HIV negative (99.2%) than for HIV positive (97.2%) patients (P = 0.006), and in the active syphilis group for patients who did not have malaria (99.6%) compared with those who were malaria-positive (99.0%) (P = 0.01).

There were few invalid and indeterminate TPHA results (n = 18). The proportion of such results was higher in patients with malaria (n = 6, OR 3.3 (95% CI: 1.2–8.9)). Of the 45 RPR biological false-positive cases, 8 (18%) had malaria (1.4 (95% CI: 0.7–3.1)).

Discussion

Our results show that the diagnostic accuracy of the ICS was significantly better than that of the RPR even after controlling for HIV, malaria and level of service (i.e., whether tests were done at a health facility or the reference laboratory). The accuracy of the ICS test compared favourably with that of TPHA carried out at the Beira reference laboratory, but ICS accuracy decreased when done at health facilities. That is, the accuracy of the ICS was greater when used by laboratory staff with higher levels of training and in a setting with better infrastructure and supervision than when carried out by staff at health facilities.

The sensitivity of the ICS for active syphilis in the reference laboratory decreased significantly in the presence of malaria; HIV infection did not affect the diagnostic accuracy of the ICS significantly. Even though the ICS has lower sensitivity and specificity than the TPHA, ICS could allow syphilis to be diagnosed in health facilities that do not have laboratories and the test also offers hope of improving syphilis diagnosis in facilities with laboratories.

Although the 9.5% RPR overall rate of syphilis in our sample group at first antenatal visits is high, it is lower than the 15.1% rate described for Beira in the previous 5 years from 1998–2003.¹⁶ Reasons for this discrepancy may include a positive secular trend due to increased counselling, diagnosis and treatment especially in antenatal care or to increased condom use prompted by the AIDS epidemic. Other explanations are the incorrect performance of laboratory procedures (use of whole blood, variable time and techniques of rotation, poor conservation of reagents), low levels of supervision, and increased rates of death in the syphilis-susceptible or infected population.³²

Self-reported syphilis history and treatment were not reliable in our study population. The proportion of women with a previous syphilis history who reported having received syphilis treatment and who had negative serologic results for all syphilis tests was higher than expected. Explanations include inaccurate recall, insufficient explanation given to patients and their partners when they receive treatment, previous false-positive RPRs, over-treatment of syphilis as a result of the use of the WHO algorithm for syndromic treatment of sexually transmitted infections, and perhaps reversion of serological tests to non-reactive after treatment.^33–35

We compared our study results with the WHO laboratory-based evaluation of rapid syphilis diagnostics ²³ using a random subsample of 400 sera, and we found no significant differences with the sensitivity (c² test P = 0.09) and specificity (P = 0.5) of our ICS_Ref. Compared with other field studies, our data on the sensitivity of the RPR_HF for active syphilis showed no significant differences to those reported in the Gambia (P>0.3)³⁶ and in Senegal (P>0.5),³⁷ but our sensitivity was lower than that reported in a South African study (P<0.01).³⁸

Antenatal screening for syphilis is usually done with non-treponemal tests, either alone or in combination with treponemal tests.^{6, 9, 25, 27, 28, 30, 38} However, the use of a treponemal test as the only means to diagnose syphilis has important public health implications. About 20% of pregnant women who had a positive result from the ICS^HF test were TPHA-positive/RPR-negative (2.1% of all women tested). However, only 14% of the TPHA-positive/RPR-negative women reported previous syphilis treatment.

In Mozambique there are about one million first antenatal visits every year. Using the ICS_HF as the principal screening test, we would correctly detect 93 800 cases of syphilis every year, assuming the same syphilis prevalence and ICS_HF recorded in our study. From this group of 93 800 women, about 27 000 (15 000 TPHA-positive/RPR-negative and 12 000 TPHA-positive/RPR-positive cases) would not be diagnosed with use of the RPR^HF test. The difference could exceed 20 600 for each of these serological groups if the quality of testing at the health facilities were equal to that of the reference laboratory.

Although with the ICS we would detect more active syphilis cases than are currently detected with RPR, we would also detect more TPHA-positive/RPR-negative cases. Whether these patients are truly past treated cases of syphilis or patients with late untreated syphilis, active syphilis with false-negative RPR, or very early primary syphilis is difficult to know.²⁵ Women with untreated syphilis, particularly if co-infected with HIV, are at risk of having syphilis complications (i.e. neurosyphilis and other tertiary manifestations)^{34, 39, 40} and may pass on the infection to their child or to their sexual partners. In settings of high HIV seroprevalence and poor availability of syphilis diagnosis and treatment, patients who are TPHA-positive/RPR-negative would probably benefit from treatment.

Women who have had a previous syphilis diagnosis and treatment and who are screened with the ICS test in subsequent pregnancies will probably again test positive for the disease. As a result, they and their sexual partners will most likely be offered treatment during each pregnancy. Since syphilis is a sexually-transmitted disease, repeated notification and treatment of sexual partners may place women at risk of domestic violence and family break-ups.

Although this concern about having positive syphilis tests repeatedly in subsequent pregnancies (after receiving appropriate treatment in a previous pregnancy) is more important in relation to the use of treponemal tests, it is not limited to those tests. Especially if time between subsequent pregnancies is short (i.e., there is no time for the non-treponemal test to become negative), and there are neither follow-up quantitative RPRs, nor clinical records.

In view of the small sample sizes in our study, our findings about differences of diagnostic accuracy between patients with malaria or HIV, or both should be interpreted with caution. However, if true, the described differences may have important clinical implications in areas with high rates of HIV and malaria.

The ICS test is easy to perform. However, the lowered diagnostic accuracy at health facilities compared with the reference laboratory, and the decrease in accuracy during the course of the study is cause for concern. To obtain consistently accurate results, ongoing training and systematic supervision with regular quality control are necessary. Particular emphasis should be made on recognizing weak-positive ICS results. A good light source is essential for more accurate ICS test results. We did not systematically measure laboratory temperature, so can not describe the association between test performance and ambient temperature.

Widespread use of the ICS in Mozambique and comparable settings would result in a substantial improvement in local capacity to prevent congenital syphilis. Furthermore, use of ICS would allow timely detection and treatment of the disease, thus reducing maternal, perinatal and infant mortality and morbidity in the region.

Our findings are important for the calculation of the cost-effectiveness of the intervention. They also show the need to ascertain the rate of reversion to ICS-negative in syphilis-infected patients who have had appropriate treatment. Furthermore, it will be important to define more clearly the effects of immune-function modifiers on the diagnostic accuracy of the test. Finally, it is critical to design interventions that will promote high levels of accuracy in testing that are sustainable in health facility settings.

Acknowledgements

We acknowledge the assistance of the Ministry of Health of Mozambique, health workers from the clinics involved in the study and from the Beira Central Hospital reference laboratory; the Program for Appropriate Technology in Health (PATH); Health Alliance International staff in Mozambique and in Seattle, Washington; Dr Elena Folgosa, Dr James P Hughes and Dr Christina Marra.

Funding: The Bill and Melinda Gates Foundation funded this study.

Competing interests: none declared.

References

1. McDermott J, Steketee R, Larsen S, Wirima J. Syphilis-associated perinatal and infant mortality in rural Malawi. Bull World Health Organ 1993;71:773-80.        

2. Schulz KF, Schulte J, Berman S. Maternal health and child survival: opportunities to protect both women and children from the adverse consequences of reproductive tract infections. New York: Plenum Press; 1992.        

3. Schulz KF, Cates W, O'Mara PR. Pregnancy loss, infant death, and suffering: legacy of syphilis and gonorrhea in Africa. Genitourin Med 1987;63:320-5.        

4. Fiumara NJ. Syphilis in newborn children. Clin Obstet Gynecol 1975;18:183-9.        

5. Evans HE, Frenkel LD. Congenital syphilis. Clin Perinatol 1994;21:149-62 .        

6. Singh AE, Romanowski B. Syphilis: review with emphasis on clinical, epidemiologic, and some biologic features. Clin Microbiol Rev 1999; 12:187-209.        

7. World Health Organization. Global prevalence and incidence of selected curable sexually transmitted infections overview and estimates. Geneva: WHO; 2001.        

8. Gilson L, Mkanje R, Grosskurth H, Mosha F, Picard J, Gavyole A, et al. Cost-effectiveness of improved treatment services for sexually transmitted diseases in preventing HIV-1 infection in Mwanza Region, Tanzania. Lancet 1997;350:1805-9.        

9. World Health Organization. Report of WHO consultation on maternal and perinatal infections, 28 November–2 December 1988. Geneva: WHO; 1991.        

10. Lumbiganon P, Piaggio G, Villar J, Pinol A, Bakketeig L, Bergsjo P, et al. The epidemiology of syphilis in pregnancy. Int J STD AIDS 2002;13: 486-94.        

11. Hira S, Bhat GJ, Chikamata DM, Nkowane B, Tembo G, Perine PL, et al. Syphilis intervention in pregnancy: Zambian demonstration project. Genitourin Med 1990;66:159-64.        

12. Temmerman M, Gichangi P, Fonck K, Apers L, Claeys P, Van Renterghem L, et al. Effect of a syphilis control programme on pregnancy outcome in Nairobi, Kenya. Sex Transm Infect 2000;76:117-21.        

13. Terris-Prestholt F, Watson-Jones D, Mugeye K, Kumaranayake L, Ndeki L, Weiss H, et al. Is antenatal syphilis screening still cost effective in sub-Saharan Africa? Sex Transm Infect 2003;79:375-81.        

14. Grosskurth H, Mosha F, Todd J, Mwijarubi E, Klokke A, Senkoro K, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: randomised controlled trial. Lancet 1995.346:530-6.        

15. Gloyd S, Chai S, Mercer MA. Antenatal syphilis in sub-Saharan Africa: missed opportunities for mortality reduction. Health Policy Plan 2001;16:29-34.        

16. Gloyd S, Macome C, Floriano F, Chadreque MA, Lafort Y. In: American Association of Public Health, Atlanta, Georgia, October 21-25 2001. Prenatal syphilis screening in Mozambique: Approaches for sustainable low-cost reduction of perinatal mortality. In Portuguese.        

17. Cossa HA, Gloyd S, Vaz RG, Folgosa E, Simbine E, Diniz M, et al. Syphilis and HIV infection among displaced pregnant women in rural Mozambique. Int J STD AIDS 1994;5:117-23.        

18. Lafort Y, Gloyd S, Floriano F, Chadreque MA. Evaluation of screening for syphilis and high risk pregnancy in prenatal clinics of Mozambique. In: IX Jornadas de Saude, Maputo, Mozambique, December 13–16 1994. In Portuguese.        

19. Temmerman M, Mohamedali F, Fransen L. Syphilis prevention in pregnancy: an opportunity to improve reproductive and child health in Kenya. Health Policy Plan 1993;8:122-7.        

20. Erbelding EJ, Vlahov D, Nelson KE, Rompalo AM, Cohn S, Sanchez P, et al. Syphilis serology in human immunodeficiency virus infection: evidence for false-negative fluorescent treponemal testing. J Infect Dis 1997;176:1397-400.        

21. Haghighi L, Doust JY, Boroomand K. Biological false positive VDRL test in malaria. Trop Geogr Med 1970;22:482-5.        

22. Zarakolu P, Buchanan I, Tam M, Smith K, Hook EW. Preliminary evaluation of an immunochromatographic strip test for specific Treponema pallidum antibodies. J Clin Microbiol 2002;40:3064-5.        

23. World Health Organization/Special Programme for Research and Training in Tropical Diseases. Laboratory-based evaluation of rapid syphilis diagnostics: results from 8 SDI Sites. Geneva: UNDP/World Bank/WHO; 2003.        

24. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Fam Practice 2004;21:4-10.        

25. Larsen SA, Steiner BM, Rudolph AH. Laboratory diagnosis and interpretation of tests for syphilis. Clin Microbiol Rev 1995;8:1-21.        

26. Goh BT, van Voorst Vader PC. European guideline for the management of syphilis. Int J STD AIDS 2001;12 Suppl 3:14-26.        

27. Genc M, Ledger WJ. Syphilis in pregnancy. Sex Transm Infect 2000;76:73-9.        

28. McElborough DJ. Guidelines for serological testing for syphilis. Sex Transm Infect 2001;77:79.        

29. Young H. Guidelines for serological testing for syphilis. Sex Transm Infect 2000;76:403-5.        

30. Bastos R. Guia para o Tratamento e Controle das DTS [Guideline for Treatment and Control of STD]. Maputo: Ministerio Nacional de Saude, 2001. In Portuguese.        

31. Barreto A. Guia para o Tratamento da Malária nas Unidades Sanitárias. [Guidelines for malaria treatment in Health Facilities]. Maputo: MISAU; 1995. In Portuguese.        

32. Chesson H, Dee TS, Aral SO. AIDS mortality may have contributed to the decline in syphilis rates in the United States in the 1990s. Sex Transm Dis 2003;30:419-24.        

33. Romanowski B. Sutherland R, Fick GH, Mooney D, Love EJ. Serologic response to treatment of infectious syphilis. Annals Intern Med 1991;114:1005-9.        

34. Marra CM. Syphilis and human immunodeficiency virus infection. Semin Neuro. 1992;12:43-50.        

35. Lukehart S. Serologic testing after therapy for syphilis: is there a test for cure? Ann Intern Med 1991;114:1057-8.        

36. West B, Walraven G, Morison L, Brouwers J, Bailey R. Performance of the rapid plasma reagin and the rapid syphilis screening tests in the diagnosis of syphilis in field conditions in rural Africa. Sex Transm Infect 2002;78:282-5.        

37. Van Dyck E, Van de Velden L, Ndoye I, Piot P, Meheus A. Evaluation of the rapid plasma reagin "teardrop" card test for screening of syphilis in field conditions. Sex Transm Dis 1993;20:194-7.        

38. Delport SD. On-site screening for maternal syphilis in an antenatal clinic. S Af Med J 1993;83:723-4.        

39. Hicks CB. Syphilis and HIV infection. Dermatol Clin 1991;9: 493-501.          40. Tikjob G, Russel M, Petersen CS, Gerstoft J, Kobayasi T. Seronegative secondary syphilis in a patient with AIDS: identification of Treponema pallidum in biopsy specimen. J Am Acad Dermatol 1991;24:506-8.        

(Submitted: 28 May 2004 – Final revised version received: 5 August 2005 – Accepted: 18 August 2005)

1 Correspondence to Dr Montoya (email: pablom@teledata.mz).