Cost-Effectiveness of a Dual Non-Treponemal/Treponemal Syphilis Point-of-Care Test to Prevent Adverse Pregnancy Outcomes in Sub-Saharan Africa

Abstract

Background: 

A dual nontreponemal/treponemal point-of-care test (Dual-POC) that simultaneously detects both nontreponemal and treponemal antibodies has been developed and evaluated. In this study, we compare the health and economic outcomes of the new test with existing syphilis tests/testing algorithms in a high prevalence setting.

Methods: 

We used a cohort decision analysis model to examine 4 testing/screening algorithms; the Dual-POC test, the laboratory-based rapid plasma reagin and Treponema pallidum haemagglutination assay (RPR+TPHA) algorithm, an onsite RPR testing, and point-of-care treponemal immunochromatographic strip (ICS) testing. Outcomes included miscarriage, stillbirth, congenital syphilis, low birth weight, and neonatal death. Disability-adjusted life-years were estimated for all health outcomes. The analytic horizon was the life expectancy for the mother and child.

Results: 

For a cohort of 1000 pregnant women in a historically high syphilis prevalence population (10% infected and 15% previously infected), the model predicted a total of 39 adverse pregnancy outcomes if no serologic screening were performed; 13 for the laboratory-based RPR+TPHA; 11 for the on-site RPR strategy; 5 for the Dual-POC strategy; and 2 for the ICS strategy. On the basis of assumption that the cost of ICS and the Dual-POC tests were the same, the ICS strategy was the most cost saving (saved $30,000) followed by the Dual-POC strategy (saved $27,000).

Conclusions: 

The dual-POC test may help save cost in resource-poor settings where disease prevalence (and loss to follow-up) is high, while substantially reducing overtreatment.

In Brief

The new dual nontreponemal/treponemal point-of-care test was more cost-saving than the onsite rapid plasma reagin and laboratory-based rapid plasma reagin and Treponema pallidum haemagglutination assay tests. However, the onsite immunochromatographic strip test was more cost-saving than the new dual nontreponemal/treponemal point-of-care test.

Adverse pregnancy outcomes attributable to syphilis infection continue to be a serious public health concern in many developing countries.¹ The reported prevalence of syphilis among pregnant women in Sub-Saharan Africa has historically been relatively high, ranging from 3% to 17%.^2–6 The disease may cause fetal or perinatal death in 40% of affected pregnancies⁷ and may also cause complications in surviving newborns, such as central nervous system abnormalities, deafness, multiple skin, bone, and joint deformities, and hematological disorders.⁸

Effective interventions to reduce or prevent adverse pregnancy outcomes should include antenatal serological screening and treatment using a single dose of intramuscular benzathine penicillin for those women whose sera tested positive.⁷ The recommended screening algorithm for syphilis involves 2 types of serologic tests (nontreponemal and treponemal). The use of a nontreponemal test alone, such as the rapid plasma reagin (RPR, Becton Dickinson, Sparks, MD) test, may result in some overtreatment (treatment of uninfected persons) owing to biologic false-positive reactions.⁷ On the other hand, the use of treponemal tests alone, such as the immunochromatographic strip assay (ICS, Abbott Laboratories, Chicago, IL), may also result in overtreatment because they do not differentiate between adequately treated and previous infections, but are relatively more sensitive.⁹ Additionally, treponemal tests may detect other nonsyphilitic treponemes, such as bejel and yaws, which are common in the tropics.¹⁰

Consequently, the 2-step algorithm was recommended.^7,11,12 The 2-step algorithm involves screening with a nontreponemal test followed, if reactive by a confirmatory treponemal test such as the Treponema pallidum haemagglutination assay (TPHA, Fujirebio, Tokyo, Japan). This testing algorithm has been considered as the “gold standard” when assessing other tests and alternative screening algorithms.^13–15 The reversed algorithm (i.e., treponemal followed by nontreponemal) has been shown to be less cost-effective in low and high prevalence settings.¹⁶

Although the RPR+TPHA algorithm is more specific than the use of the RPR test alone, it may result in significantly lower rates of treatment because these tests are usually conducted at laboratories remote from the clinical site and a substantial number of individuals fail to return for their test results and appropriate treatment.¹⁷ To overcome problems associated with failure to return for results, some researchers have advocated the use of on-site RPR testing performed by primary health care workers,¹⁸ whereas others have promoted the use of treponemal ICS.²

The New Dual Nontreponemal/Treponemal Point-of-Care Test

To combine the advantages offered by both serologic tests (nontreponemal and treponemal) and to reduce the time it takes for results to become available, an innovative dual treponemal/nontreponemal point-of-care immunochromatographic test (Dual-POC, Chembio Diagnostics Systems Inc, Medford, NY) for the simultaneous detection of reagin and treponemal antibodies has been developed and evaluated at the United States Centers for Disease Control and Prevention (CDC).¹⁹ The Dual-POC test is based on the principle of a dual path platform that has 2 antigens; 1 nontreponemal, 1 treponemal, and a control line striped onto the surface of a nitrocellulose membrane within the device. The Dual-POC test has the capacity to qualitatively detect nontreponemal antibodies (reagin) and confirm the specificity of reactive tests within 15 minutes. The test requires no expertise in reading test results.¹⁹

The main objective of this study was to estimate and compare the health and economic outcomes of 5 alternative prenatal serological screening protocol alternatives in Sub-Saharan Africa: onsite Dual-POC; onsite ICS; onsite RPR testing without confirmation; the conventional laboratory-based RPR+TPHA testing algorithm; and No-Program (which assumed no screening would take place). We estimated the adverse pregnancy outcomes prevented, the overtreatment rates, and the expected total costs from societal and health clinics' perspective. The results of this study may help provide important health and economic outcome information about the new testing technology when compared with existing strategies.

METHODS

Performance of the Dual-POC

The results of a laboratory evaluation of the Dual-POC test conducted at the CDC indicated that the sensitivity and specificity of the nontreponemal line were 88.6% and 98.6%, respectively, when compared with the RPR test. The sensitivity and specificity of the treponemal test line of the Dual-POC test were found to be 96.5% and 95.5%, respectively, when compared with a T. pallidum passive particle agglutination assay (TPPA) which is equivalent to a TPHA test. The overall ability of the Dual-POC to detect both nontreponemal and treponemal antibodies in sera that were dually RPR and TPPA positive, (i.e., sensitivity) was 88.6%, while the ability of the test to exclude those sera that were not dually reactive (i.e., either RPR and/or TPPA negative), namely specificity, was 98.7%.¹⁹ Finally, the specificity of the Dual-POC test for previously treated infections was estimated to be 95.7%, following the CDC evaluation of the test.¹⁹

Study Model and Assumptions

Because the laboratory evaluation did not differentiate between the various stages of the disease, we assumed primary and secondary syphilis for those infected. We assumed that the screening program was undertaken at the first trimester of pregnancy for a cohort of 1000 pregnant women presenting in a Sub-Saharan African country with a historically high prevalence of syphilis (10% currently infected, and 15% previously infected and successfully treated). Thus, there were 1000 pregnancy outcomes for each alternative testing strategy. The analytic horizon assessed pregnancy outcomes attributable to untreated maternal syphilis infection (i.e., miscarriage in the second trimester, stillbirth, neonatal death, low birth weight, and congenital syphilis) and the natural history of undetected and therefore untreated primary and secondary syphilis infection for the mother. Cost included program costs and cost of pregnancy outcomes obtained from the literature. We assumed that the on-site tests (i.e., RPR, Dual POC, and ICS) were the basis for immediate subsequent treatment while the laboratory-based test (i.e., RPR+TPHA) assumed loss to follow-up before treatment. On the basis of published studies, we assumed that a single injection of 2.4 MU benzathine penicillin before the third trimester, which is used extensively in resource-poor settings, was 98% effective in treating early syphilis and therefore also effective in preventing vertical transmission of infection.^11,20 Details of parameter values including sources are presented in Table 1. For the base-case analysis, we assumed that the cost of the Dual-POC test was the same as the ICS test ($3.73, 2008 US dollars).

TABLE 1: Description, Point Values, Range and Sources for All Variables Used in the Model to Compare the Dual-POC Test With the Traditional Laboratory-Based Screening Algorithm, On-Site RPR and ICS Tests for Screening Pregnant Women in a Developing County Setting

TABLE 1: (Continued)

We constructed a cohort decision analysis model to estimate and compare the total costs, number of women treated, and total number of adverse pregnancy outcomes prevented. Following previous studies,^16,37 we computed overtreatment rates as the ratio of uninfected to infected women treated. Following recommended cost-effectiveness analyses guidelines,³⁸ we computed cost as the net cost (total cost minus cost of No-Program). Negative net costs were reported as cost savings. We conducted several one-way sensitivity analyses as well as a multiway (Monte Carlo simulation) sensitivity analysis using the ranges provided in Table 1 to assess the relative health and economic outcomes. However, for the purpose of this study, we focused largely on the variables whose ranges potentially affected the relative costs and effectiveness of the Dual-POC strategy. In the Monte Carlo simulation, we focused on the Dual-POC and ICS results. We determined the disability-adjusted life years (DALYs) for both mother and child using Global Disease Burden measurement methods.³⁹ To do this, we used separate Markov progression models based on Life Table values for South Africa and syphilis-related disability weights as provided in Table 1. An illustration of the Markov progression model showing the health states used to determine the DALYs for the mother is shown in Figure 1. We also repeated the analysis using global Life Table values.

Figure 1.: A schematic of the Markov progression model used to estimate the disability-adjusted life-years (DALYs) for the mothers. Due to lack of DALY weights for all stages of syphilis disease, we used 4 states: early syphilis (i.e., primary and secondary syphilis), tertiary/late syphilis (neurosyphilis), recover/healthy, and death. We assumed that the pregnant women were in the early syphilis state at screening. Those who were not identified and treated (or whose treatment failed) may recover (become healthy) or progress to the tertiary state where they may stay or die. Those who recovered may relapse (become infectious) into early syphilis or stay healthy. State transition probabilities were obtained from the literature (Table 1).

We used the medical care component of the consumer price index for all urban consumers and adjusted all costs to 2008 United States dollars based on data from Sub-Saharan Africa.⁴⁰ We used DATA Professional version 4.0 (TreeAge Software, Williamstown, MA) to construct the decision tree and conduct the comprehensive sensitivity analyses. Microsoft Excel, version 2007 (Microsoft Corporation, Redmond, WA), was used for summary analyses and presentation of results.

RESULTS

Health and Cost Outcomes

For a cohort of 1000 pregnancies, our model predicted a total of 39 adverse pregnancy outcomes for the No-Program strategy (total cost, $106,000); 13 for the lab-based RPR+TPHA (total cost, $86,000); 11 for the on-site RPR strategy (total cost, $84,000); 5 for the Dual-POC strategy (total cost, 79,000); and 2 for the ICS strategy (total cost, $76,000, see Table 2). Thus compared with the No-Program strategy, the ICS strategy prevented the most cases of adverse pregnancy outcomes, averted the most DALYs, and saved the highest costs, followed by the Dual-POC strategy (Tables 2, 3). The estimated total societal costs were slightly (<3%) higher than the total health clinic costs for all the screening strategies.

TABLE 2: Summary Results From the Cohort Decision Model Comparing Treatment Rates and the Expected Effects (Pregnancy Outcomes) for All 4 Prenatal Syphilis Tests/Testing Algorithms and a No-Program Strategy for 1000 Pregnancies in a Developing Country Setting

TABLE 3: Summary Results From the Cohort Decision Model Comparing the Expected Effects (Disability-Adjusted Life-Years (DALYs)) and Costs for All 4 Prenatal Syphilis Tests/Testing Algorithms and a No-Program Strategy in a Developing County Setting

Overtreatment Rates

The results of our analyses are summarized in Table 2. Based on our assumptions and the base-case values used, no treatment was provided to uninfected women for the No-Program and laboratory-based RPR+TPHA strategies. Overtreatment rates were 0.32 (i.e., 23/71) for RPR, 0.18 (i.e., 16/89) for the Dual-POC, and 1.42 (i.e., 141/98) for ICS.

One-Way Sensitivity Analyses

In the one-way sensitivity analyses, we varied select variables and examined the total expected cost keeping all other variables constant. As expected, the relative test performance, test costs, and costs associated with adverse pregnancy outcomes were all influential. However, in almost all scenarios the ranges we used for the 1- and 2-way sensitivity analyses showed the screening programs to be cost-saving. When we increased the sensitivity of RPR from 60% to 98%, our results indicated that the RPR strategy was more cost-saving than the Dual-POC strategy when its sensitivity was at ≥85% and the ICS strategy when its sensitivity was at ≥93% (not shown). Figure 2 shows one-way sensitivity analyses results for syphilis prevalence, sensitivity of the Dual-POC test, cost of the Dual-POC test, and the sensitivity of the ICS test. The total cost for all the alternatives increased with syphilis prevalence (Fig. 2A). The Dual-POC strategy was less cost-saving than the onsite RPR strategy when the sensitivity of the Dual-POC test was less than 75%. However, the Dual-POC was the most cost-saving strategy when its sensitivity was ≥97% (Fig. 2B). Over the range of the cost of the Dual-POC test used (i.e., $0.5–$5), it remained more cost-saving than the onsite RPR and lab-based RPR+TPHA strategies (Fig. 2C). Finally, the sensitivity of the ICS test would have to be less than 90% for it to be less cost-saving than the Dual-POC strategy (Fig. 2D).

Figure 2.: One-way sensitivity analyses for select variables from the decision model used to compare the total societal costs for all 4 tests/testing algorithms (including a No-Program strategy) to prevent adverse pregnancy outcomes for 1000 pregnant women in a developing county setting. A, One-way sensitivity analysis on the prevalence of syphilis. B, One-way sensitivity analysis on the sensitivity of the Dual-POC test. C, One-way sensitivity analysis on the cost of the Dual-POC test. D, One-way sensitivity analysis on the sensitivity of the ICS test. Vertical lines indicate the base case value for the variable on the horizontal axis; Dual-POC, new dual treponemal/nontreponemal point-of-care test; and ICS, immunochromatographic strip.

Multiway Sensitivity Analyses (Monte Carlo Simulation)

Using the ranges provided in Table 1 and assuming triangular distributions, we conducted a Monte Carlo simulation in which all variables (i.e., those with ranges) were varied at the same time for 1000 samples. The linear cost-saving trend line for ICS was higher than that for the Dual-POC ($13,000 vs. $9000).

DISCUSSION

We evaluated the health and economic outcomes of a new Dual-POC test compared with other diagnostic approaches, by examining expected pregnancy outcomes for 1000 women using conservative assumptions based on the performance characteristics of the test determined during an evaluation at the US CDC, serological profiles, and patient characteristics among antenatal clinic attendees previously reported in studies conducted in southern Africa.^13,15,41 The laboratory based RPR+TPHA algorithm prevented 26 adverse pregnancy outcomes; the onsite RPR prevented 28; the Dual-POC prevented 34; and the ICS prevented 37 out of a total of 39 expected adverse pregnancy outcomes when there was no screening program. In terms of costs, the Dual-POC saved more cost than the onsite RPR and RPR+TPHA strategies, while the ICS strategy saved the most cost.

We checked the consistency of the results from our model by comparing our results with 2 recent studies that compared RPR+TPHA, RPR, and ICS for a similar setting.^13,15 We found that the relative costs and effectiveness were consistent with their results based on the test performance values used. Additionally, the relative adverse pregnancy outcomes (miscarriage, congenital syphilis, low birth weight, stillbirth, and neonatal death) were consistent with the results reported in previous studies in similar settings.^2,13,15

The strengths of this study are that we accounted for previously treated infections in the population that was studied.^16,37 Second, we provided detail information on the expected overtreatment rate, which has been ignored (or has not been reported) in previously published studies. Finally, we improved the robustness of the results by conducting a comprehensive sensitivity analyses to account for inconsistencies in the data we used.

Our study has all the limitations associated with models in general: models are simplifications of reality and do not capture all the important characteristics of the events/phenomena being studied. In addition, the values used in our study were estimates obtained from other studies that were not consistent. There was no substantial difference between the total societal costs and total health clinic costs due to lack of reliable societal cost estimates. Another major limitation of our study is the inability to independently assess the Dual-POC. This was because the performance of the Dual-POC was assessed using RPR and TPPA (equivalent to TPHA) tests under ideal laboratory conditions. Also, we ignored ongoing transmission in the population. Nonetheless, to the extent that the screening strategy adopted does not affect the transmission dynamics, we believe that the relative costs and outcomes would not change for the onsite strategies (RPR, ICS, and Dual-POC). We did not account for costs associated with follow-ups, which may include assessment of response to therapy using serial quantitative titers.

The effect of transmission on the RPR+TPHA and No-Program strategies on our final results is difficult to assess because it would require more data and assumptions about mixing patterns and the prevalence among men. Additionally, pregnant women may not be an important source of ongoing transmission, which may limit the importance of this omission. Our assumption of untreated early syphilis (i.e., the exclusion of early and late latent forms of syphilis) in the hypothetical cohort of pregnant women implied that the estimated overall adverse health outcome was relatively higher for each strategy. However, it is difficult to determine how this omission affected the relative costs and effects of the testing strategies assessed in this study. Finally, due to lack of data, we did not include the cost of advanced stages of syphilis for the mothers.

With regard to comparing point-of-care tests to the 2-step algorithm, our results were consistent with previous studies in similar settings.^2,13,15 Consistent with results reported by Rydzak and Goldie,¹⁵ our results indicated that the screening strategies examined (onsite RPR, ICS, lab-based RPR+TPHA, and onsite Dual-POC) were cost-saving when compared with the No-Program strategy. The cost-savings found in Rydzak and Goldie¹⁵ were substantially higher than we found because they accounted for lifetime pregnancies (6 per woman). Although we used a similar time frame and similar cost values as Blandford et al.,¹³ we included 2 other substantially more expensive adverse pregnancy outcomes (neonatal death and low birth weight) that may result from untreated maternal syphilis which resulted in higher savings for the screening strategies compared with the No-Program strategy.

Although the use of the RPR test may result in a small number of false positives,⁵ studies have demonstrated that nontreponemal test reactivity has a higher correlation with disease activity than detection of treponemal antibody.⁴² In addition, nontreponemal tests are more likely to be nonreactive following successful treatment.⁹ On the other hand, the use of a treponemal ICS test alone will inevitably result in high rates of overtreatment because all treponemal tests detect antibodies that may be present for a lifetime even following provision of adequate treatment.⁴³ As demonstrated by our analyses, even when the specificity of the Dual-POC was as low as 85% (or 30% for previous infections), it resulted in a substantially lower overtreatment rate when compared with the ICS strategy.

Given these limitations of treponemal and nontreponemal tests (when used individually), the Dual-POC can help to detect syphilis cases in resource-poor settings while substantially reducing the rate of overtreatment. Additionally, the Dual-POC test should prove to be a useful screening/confirmatory test for the prevention of congenital disease because further laboratory evaluation indicated that its sensitivity was substantially higher (99.7%) for sera with RPR antibody titers ≥1:8 (during pregnancy, transplacental transmission of syphilis rarely occurs at confirmed maternal RPR titers <1:8^5,19). Further independent assessment of the Dual-POC should be undertaken to enable wider and more objective comparability, including guidelines on interpretation of results.