Syphilis is a sexually transmitted infection (STI) caused by the bacterium, Treponema pallidum. In the United States, incidence has increased nationwide; the rate of primary and secondary syphilis in 2016 increased 18% compared with the preceding year and by 74% when compared with 2011.1 The disease burden primarily lies with men who have sex with men (MSM) who accounted for 60% of new cases in 2016.1 The incidence of syphilis in Denver County mirrors the national trend. Between 2012 and 2016, the rate of primary and secondary syphilis in Denver County increased by 16.3% with 78% of these cases occurring in MSM.2
The Denver Public Health (DPH) Outreach Program was established in 2000 to screen for and identify new cases of human immunodeficiency virus (HIV) and STIs, including syphilis, among MSM in nonclinical settings within the community. These include 2 community-based organizations, 2 bathhouses, a mobile van, and a pharmacy; all of which serve lesbian, gay, bisexual, and transgender clientele and have assisted with DPH's prevention activities by allowing outreach workers to provide on-site HIV and STI testing. Historically, syphilis screening in outreach has been conducted with a combination of laboratory-based tests comprised of the rapid plasma reagin (RPR) assay, followed by a reflex confirmatory Treponema pallidum particle agglutination (TPPA) test for positive RPRs (RPR-based algorithm). This testing framework resulted in several barriers to rapid identification and treatment of syphilis, including time for specimen delivery to the laboratory, laboratory processing, time to notification of the patient, and time to arrival in the STI clinic for treatment. In addition, performance of these laboratory-based tests requires skilled phlebotomists in outreach settings for venous blood collection that may be an additional barrier for some programs.
In recent years, rapid syphilis tests (RST) using finger-stick sampling rather than venipuncture have emerged as an alternative to traditional laboratory-based testing methods.3 RST are treponemal antibody tests that have been used successfully for syphilis screening outside the United States among pregnant women and MSM.4,5 In December 2014, the Food and Drug Administration granted a Clinical Laboratory Improvement Amendments waiver allowing an RST, Syphilis Health Check (Trinity Biotech Jamestown, NY), to be used in nonclinical settings. The utility of a treponemal test in settings with high prevalence of past syphilis infection is unclear. To date, there are limited published studies on RST performance in outreach settings within the United States. A few public health departments have attempted adoption of RST with varying success.6,7 Establishing the reliability of this point-of-care (POC) test in nonclinical settings could result in expansion of syphilis testing to other outreach settings and at-risk populations with a resultant increase in identification of syphilis cases. In addition, POC testing reduces time delays associated with specimen handling and processing. Point-of-care tests allow for immediate client notification of positive results and navigation to treatment. Therefore, use of RST in outreach settings could increase access to quality STI treatment services for highly affected populations and may improve syphilis control.
The aims of this pilot program were to compare the performance of an RST-based testing algorithm (RST with reflex RPR and TPPA for positive RST results) versus the traditional RPR-based algorithm (RPR with reflex TPPA only) in the diagnosis of previously untreated syphilis, and to determine whether time to treatment is shortened for new cases of syphilis identified by the RST-based algorithm.
This was an observational study that evaluated an RST-based testing algorithm in 6 outreach settings serving MSM in the Denver Metropolitan Area between April 1, 2016, and January 17, 2018. Four DPH outreach workers rotated to each location and provided screening for STIs and HIV. They collected urine, rectal and pharyngeal specimens for gonorrhea and chlamydia testing using nucleic acid amplification testing methods and performed venipuncture for rapid HIV testing with Uni-Gold (Trinity Biotech) and laboratory-based syphilis testing. The Syphilis Health Check was used for RST. This test requires blood from a finger-stick and delivers results in 10 to 15 minutes.
Clients seen in outreach settings who self-identified as gay, bisexual, or MSM (regardless of HIV status) were included in the study. The RST is a treponemal test that will remain positive in those previously diagnosed and treated for syphilis. Therefore, any person with a self-reported history of syphilis infection was excluded from RST screening. Self-reported prior syphilis infection was later confirmed by review of existing electronic health record (EHR) data or by state public health syphilis records prior to data analysis, but this information was not available at the time of outreach testing.
This study was part of a program evaluation and was reviewed and exempted from institutional review by the Combined Multiple Institutional Review Board of the University of Colorado Denver.
Standard practice for syphilis screening prior to this pilot program involved the use of the traditional RPR-based algorithm. Blood samples for laboratory-based syphilis testing were obtained at each outreach site then couriered to the state public health laboratory where syphilis tests were performed using the Macrovue RPR card test (Becton Dickenson, BD Microbiology Systems, Sparks, MD) and Serodia TPPA (Fujirebio America Inc., Fairfield, NJ). Once available, the results were reported to the STI Clinic via the EHR within 7 to 10 days. Positive results were reviewed by a nurse who would also review the EHR, contact the state disease investigation specialists (DIS) and contact the patient to ascertain prior history of syphilis. If there was no prior history and persisting concern for new or previously untreated infection, patients were scheduled for clinical evaluation and treatment. Patients who could not be contacted by the nurse after 3 attempts were referred to DIS for further inquiry.
During the implementation period, clients with no self-reported history of prior syphilis were offered screening with RST. All clients undergoing RST also provided blood for a laboratory-based RPR to ensure that no cases were missed. Anyone with a previous diagnosis of syphilis was deemed ineligible for RST. Clients who declined or were ineligible for the RST were screened with laboratory-based RPR alone. Treponema pallidum particle agglutination confirmation was conducted on those with positive RST or RPR test. Rapid syphilis test–eligible clients first received a written fact sheet on the RST, and verbal consent for testing was obtained. Clients then underwent a fingerstick to obtain a blood specimen for the RST. The test was conducted according to the manufacturer's specifications provided in the package insert.
Those with positive RST results were immediately referred to the STI clinic at DPH where patients are typically seen on a walk-in basis. Specific appointment times were given to each client before departure from the outreach-testing venue. Appointments ranged from same-day to 48 hours later based on the client's schedule. STI clinic providers were notified of the impending visit via secure EHR messaging. Before the patient's arrival, the front desk staff contacted DIS to inquire about past history of syphilis infection. Upon presentation to the STI clinic, patients received a stat RPR test (Arlington Scientific Inc., Springville, UT) and a nurse or nurse practitioner conducted a history and physical examination. All patients waited till stat RPR results were available, typically within 30 minutes to an hour. If a diagnosis of syphilis was confirmed by a positive stat RPR or clinical findings of syphilis, the patient received immediate treatment with 2.4 million units of long-acting benzathine penicillin G administered intramuscularly on site. Subsequent treatment was determined based on stage of syphilis infection according to the Centers for Disease Control Sexually Transmitted Disease treatment guidelines.8
Testing Algorithms and Diagnostic Criteria
The RPR-based algorithm involved sequential testing beginning with the laboratory-based RPR test; if the RPR was positive, confirmatory testing was done with the TPPA. For a diagnosis of syphilis, both tests had to be positive; if the TPPA was negative, the positive RPR was designated a biological false positive. In contrast, the RST-based algorithm began with an RST and if positive, TPPA was conducted at the state laboratory. In addition, as previously mentioned, RPR testing at the state laboratory was conducted for all clients regardless of RST result. In a patient with a positive RST, a conclusive diagnosis of syphilis was made if the state laboratory run RPR and TPPA were both positive, though patients who tested positive for RST were treated if the STI clinic-based rapid RPR was positive even before the results of the state laboratory-based TPPA test were available. If the RPR and TPPA were discordant a fluorescent treponemal antibody test (Zeus Scientific Inc. Branchburg, NJ) was used as a tie breaker. Estimates and confidence intervals for sensitivity, specificity, negative predictive values (NPV) and positive predictive values (PPV) were calculated for clients who received all applicable tests.
All data were recorded in the EHR (Epic Systems Corporation, Verona, WI) used by clinical entities within Denver Health including the STI clinic and outreach program. For each visit detailed medical/sexual history, screening tests completed, and test results were recorded.
Clinical data were extracted from the EHR into a de-identified database and analyzed using SAS 9.4 statistical software (SAS Institute, Cary, NC).9
Clients were separated into 2 groups: “Offered RST” and “Not Offered RST.” The “Offered RST” group comprised clients in the pilot program who presented between January 23, 2017, and January 17, 2018. The “Not Offered RST” group included clients in the historical control group who presented prior to the pilot program (between April 1, 2016, and January 22, 2017) and those seen during the implementation period when RST kits were out of stock.
Descriptive statistics were calculated to summarize the socio-demographic characteristics, sexual practices, and risk behaviors of the clients. The Kruskal-Wallis test was used for bivariate comparison testing within the population.
Measuring and Comparing Time to Treatment
Time to treatment was defined as the interval between initial blood draw or finger-stick and the time the client received syphilis treatment. The Wilcoxon signed-rank test was used to compare time to treatment for clients evaluated with the RST-based algorithm and those evaluated with the RPR-based algorithm.
2,963 clients presented for outreach testing from April 1, 2016, to January 17, 2018; 233 were excluded for previous history of syphilis. Another 434 were excluded as they declined all syphilis screening tests. The final analysis included 2296 clients; 690 clients were tested with RST, 391 declined or were ineligible for RST but accepted the RPR-based algorithm testing, and 1215 were not offered RST (1095 were seen before the RST program was initiated and 120 presented for testing during times when RST were out of stock).
Client demographics and sexual risk factors are summarized in Table 1; there was a higher proportion of self-reported chlamydia, gonorrhea and sex with anonymous partners in the “declined RST” group compared to the other 2 groups. There was no difference in the proportion of persons living with HIV in all groups and number of sexual partners reported in the last 3 months.
Identification of New Syphilis Cases and Time to Treatment Using RPR- or RST-based Algorithms
Among the 690 clients accepting RST, 671 tests were negative and 19 (2.8%) were positive. Of these, 9 were confirmed by positive RPR and TPPA while the remaining 10 had positive RST but negative RPR and TPPA thus the RST result was deemed false positive. The overall false positive rate was 10 of 671 (1.5%). The median RPR titer in persons with positive RST and positive RPR was 1:32 with a range of 1:1 to 1:64.
Negative RST was confirmed by negative RPR and TPPA in 670 of 671 cases. A single case had negative RST with positive RPR (titer 1:1) and TPPA. When comparing RST to the traditional RPR-based algorithm for the diagnosis of syphilis, RST performance was as follows: sensitivity, 90%; specificity, 98.5%; PPV, 47.4%; and NPV, 99.9% (Table 2).
All 9 cases identified by RST-based algorithm were brought to treatment in the STI clinic and determined to be new cases of untreated syphilis based on positive stat RPR. Some individuals also had clinical findings consistent with syphilis. The stage of syphilis and time to treatment for each individual are shown in Table 3. All were treated according to the Centers for Disease Control treatment guidelines with a median time to start of treatment of 1 day (range, 0–6 days).8 Of note, 2 of these cases were diagnosed with neurosyphilis and hospitalized for intravenous penicillin treatment within 24 hours of positive RST.
Ten cases with positive RST but negative RPR and TPPA results, presented to the STI clinic and were not treated for syphilis when the RPR was found to be negative.
New cases of syphilis were identified in 3 (0.77%) of 391 clients who declined RST but opted for the RPR-based algorithm; the median time to treatment was 9 days (range, 7–13 days). In the group not offered RST, 25 (2.1%) had a new syphilis diagnosis with a median time to treatment of 9 days (range, 6–21 days). One case was lost to follow-up in this group.
When comparing the median time to treatment in both groups, we found time to treatment was significantly less for those screened with the RST-based algorithm than for those screened with the traditional RPR-based algorithm (P < 0.0001).
When comparing the RST-based algorithm to the traditional RPR-based algorithm, we found the RST-based algorithm identified all new cases of syphilis among MSM in outreach settings, missing only 1 case which was likely a late latent infection. Furthermore, use of RST significantly decreased time to treatment among the 9 persons diagnosed with syphilis, including the 2 persons diagnosed with neurosyphilis who were both evaluated in the STI clinic and were admitted to the hospital for intravenous penicillin treatment within a day of testing in outreach. Rapid identification and treatment of syphilis may have limited further disease progression and possible complications related to syphilis infection for these 2 individuals. At the population level, the benefit of earlier treatment for syphilis to curb transmission may have significant public health impact. This could be an important tool to decrease the rising incidence rates seen nationally, particularly among MSM.
Although there were 10 cases in which the RST was falsely positive, this number was considered relatively small compared to the 690 individuals who were tested with RST, though it did result in unnecessary travel to the STI clinic for those patients. The high NPV in this study indicated that when the RST was negative, the diagnosis of syphilis was highly unlikely.
A study comparing the same RST to an RPR-based algorithm among patients seen in the STD clinic in Escambia County, FL had a higher proportion of false positive RST than this study. Of the 202 samples tested, 26 had reactive RST, of which 16 (61.5%) were not confirmed by either enzyme immunoassay (EIA) or reactive RPR resulting in a PPV of 38.5%. The difference in test performance may be accounted for by the smaller sample size and differing populations and settings.7
Nakku-Joloba et al10 compared 2 rapid tests including the RST, syphilis health check used in our study to a traditional algorithm in antenatal clinics of a major tertiary referral hospital in Uganda. In comparison to reference tests, RPR with reflex treponemal hemagglutination test, RST showed sensitivity of 89.8% and PPV of 97.6%. Higher RST performance is unsurprising given that our study population has significantly lower rates of syphilis infection. In 2016, the overall rate of syphilis infections in Uganda was 372.8 per 100,000 compared with 27.3 per 100,000 in the United States.1,11 When prevalence is low as in our study population, there are fewer true positives and in turn the PPV is lower.
Other studies comparing reverse sequence algorithms beginning with treponemal tests with reflex nontreponemal tests, similar to the RST-based algorithm here, found that these algorithms have an increased rate of false-positive test results.12,13 In a study using serum from 5 US laboratories, the reverse sequence algorithm (starting with EIA then reflex RPR) was compared to the RPR-based algorithm discussed here, and in it, 0.6% of over 4800 samples were false positives (i.e. first treponemal test was reactive but the confirmatory RPR and the second treponemal test were not). Binnicker et al compared another reverse screening algorithm to the RPR-based algorithm and found that of 1000 samples tested from a low prevalence population, the overall false-reactive rate was 0.6% compared to none with the reverse algorithm.
Our results support these findings although most of these studies evaluated algorithms that did not include RST, and instead, the treponemal tests used were treponemal EIA and chemiluminescence immunoassays (CIA) which are automatable laboratory-based tests. Nevertheless, as with EIA/CIA based algorithms, the RST-based algorithm despite having an increased rate of false positives, correctly identified all early syphilis (primary, secondary, and early latent) cases in this study.
Although RST is not recommended as a stand-alone test for the diagnosis of syphilis, incorporating it into a screening algorithm with confirmatory RPR before treatment is a viable option and may be particularly useful in certain settings, such as in outreach, where ability to perform phlebotomy may be limited, and there is a higher risk of loss to follow-up.6 Rapid syphilis test may reduce loss to follow-up because there is less time between diagnosis, notification of results, and treatment.
This study was designed to compare the performance of an RST with the traditional RPR-based algorithm and not to compare the RST to a gold standard treponemal test, such as the TPPA. As the use of treponemal testing by EIA is becoming increasingly common as a first test for syphilis screening (the so-called reverse algorithm), such a comparison would be relevant. In fact, we conducted a substudy among 409 subjects in which both RST and TPPA were conducted on all individuals. The results of this substudy fall outside the scope of this article but are presented in an online supplement (http://links.lww.com/OLQ/A312).
For future implementation of the RST-based algorithm, the ease with which RST can be done in outreach, the excellent NPV, and the benefit of shortening time from diagnosis to treatment should be weighed against the low PPV of the RST and the time lost going to the STI clinic for clients with false positive results. Another key factor to consider with use of RST is the relative ease and speed with which the STI clinic can evaluate and treat those who present after a positive test in outreach. Reliance on patients to report previously treated syphilis may also prove to be a challenge for health departments seeking to implement this program.
A key limitation of this study is the small number of cases identified. Determination of the sensitivity of the RST-based algorithm compared to the RPR-based algorithm in outreach was limited by the low incidence of syphilis in our study population resulting in wide confidence intervals and low PPV. Furthermore, while time to treatment was significantly shorter in the RST arm of this study, this comparison was between small numbers of patients: 9 identified by RST and 28 by the RPR-based algorithm.
The use of a treponemal test in a population with high likelihood of past syphilis infection is an inherent challenge for the appropriate use of this test in these settings. It is not an option for many clients and may falsely identify those with prior history as new cases. Obtaining an accurate history for individuals who tests RST positive but did not disclose past syphilis history was crucial to successful interpretation of test results prior to treatment; this time-consuming process was completed as previously described; however, this may not be possible in other health departments with limited staff or DIS.
Although further study is needed to better define the sensitivity of RST-based algorithms in outreach settings, incorporating RST into a screening algorithm could significantly improve time to treatment for individuals diagnosed with syphilis in outreach settings. By shortening the time from diagnosis to treatment by more than a week, this approach could meaningfully augment public health efforts to curb the increasing rates of syphilis observed nationwide.