An Evaluation of Digital Partner Notification Tool Engagement and Impact for Patients Diagnosed With Gonorrhea and Syphilis : Sexually Transmitted Diseases

Secondary Logo

Journal Logo

The Real World of STD Prevention

An Evaluation of Digital Partner Notification Tool Engagement and Impact for Patients Diagnosed With Gonorrhea and Syphilis

Folke, Tomas PhD; Menon-Johansson, Anatole Sebastian BSc, PhD, MB, BChir, MPH, FRCP, MBA

Author Information
doi: 10.1097/OLQ.0000000000001707
  • Free

Before the SARS-CoV-2 pandemic, a steady rise of sexually transmitted infections (STIs) was reported across the world.1 This is of particular concern for gonorrhea because the history of antibiotic resistance development over time and for Syphilis because of the severity of untreated disease.2 The rise of STIs has been associated with an increase in access to and the frequency of testing ‘at risk’ patients but the rise also coincides with a reduction in condom usage because of a decrease in HIV transmission risk. In 2019 in the United Kingdom (UK), gonorrhea diagnoses were the highest in 100 years of measurement and syphilis diagnoses were at levels last seen before penicillin antibiotics became available for clinical use.3

Partner Notification (PN) is a key public health tool to reduce the spread of STIs within a sexual network. The delivery of PN involves informing sexual partners of their infection risk and the need to test. Partner Notification is provided either by the health care provider, the patient, or a combination of the two. The aim of PN is to prevent complications in the partner as well as halt onward transmission of the infection.4–6 Data from the United Kingdom in 2019 show that a diagnosis was made in 41% of partners of patients diagnosed with gonorrhea and 12% of partners of patients diagnosed with syphilis.7

Modeling has estimated that doubling partner notification effectiveness, from 0.4 to 0.8 partners per index patient would result in a reduction in the cost of a chlamydia diagnosis by 10%, increase the equity of screening, and contribute to disease reduction.8 Similarly, for syphilis, partner notification has been shown to reduce the cost of diagnosis and decrease onward transmission.9,10 Despite the economic and public health evidence favoring PN, effective PN delivery remains a challenge, as is illustrated by the fact that repeat testing in patients after a diagnosis continue to remain positive because of reinfection from untreated partners.11 Such retesting practices, created as a backstop for less effective PN, have broadened from young people with chlamydia to men who have sex with men with any bacterial STI.3

A number of digital tools [InSpot, LetThemKnow, TellYourPartner] have been developed to support patients to inform partners of their potential infection risk. These platforms are open access and patients with an STI are encouraged by clinics to use these services to deliver PN themselves; however, it is not possible to measure engagement with—and the impact of—these tools.12

To address these deficiencies, a novel digital PN tool (dPNt) was developed with patients, health care staff and partners to provide a demonstrable end-to-end journey for partner(s) informed from a specific patient. The use of this dPNt can only be initiated by the health care provider, and it enables providers and patients to deliver anonymous PN via a text message or email.

At the time that PN is initiated, some partners will have already been seen and tested, and this is identified from their sexual history; however, the bulk of the PN workload occurs prospectively to identify, inform, and support partners to test.13 The use of dPNt simplifies PN by structuring and complementing the standard workflow. Figure 1 shows the standard workflow and the use of dPNt to deliver PN. Each dPNt step either delivers PN or captures key information to measure the impact of PN (Fig. 2). When a patient chooses not to “opt-in” to use the dPNt, the health care worker can skip step 2 and/or step 3 of the tool, and this is highlighted by an asterisk in the figure. In practice, the clinical teams were encouraged to use dPNt to deliver all PN; however, if the patient opted out of using the software to inform partners, then the PN delivered was equivalent to the standard of care.

F1
Figure 1:
The standard of care to deliver PN and how the digital partner notification tool can deliver PN.
F2
Figure 2:
The steps taken to use the dPNt and the information collected at each step.

Partners informed by dPNt are supported with digital triage and signposting to find a local appropriate service, to access care and to record that they have tested. Partners are signed off automatically by dPNt either by attending an appointment booked within dPNt, self-verification by the partner or the unique partner code is added to the dPNt software or in any browser (http://sxt.health/pn) (Fig. 3).

F3
Figure 3:
The notification pipeline of the dPNt.

During the development of the dPNt both provider and patient delivered PN were made opt-in to support patient autonomy and to fit in with the reality of PN delivery; however, it was not known how these options would work in a real-world setting. We, therefore, analyzed the patient factors associated with engagement with dPNt and the PN outcomes, comparing the latter with national returns to estimate the public health impact.

METHODS

Partner notification data from 2019 were extracted from a cloud database created by the social enterprise SXT Health CIC14 that records the activity of health care providers using dPNt. Each partner informed by SMS or email is allocated a unique partner code that is associated with the index patient who notified them. All analyses were performed using R.15 The data features include:

  1. nonidentifying demographic information
  2. engagement with the dPNt
  3. the number of partners already told and tested
  4. the number of partners self-verifying or signed off by a provider in dPNt or through the unique partner code being added to dPNt webform on any browser
  5. the number of partners told and tested outside of dPNt in the follow-up call.

If a partner using dPNt both attends clinic and self-verifies, the software only records clinic verification and this partner is only counted once.

Data from Public Health England (PHE) were used to estimate the baseline number of partners seen and tested for each STI7 and serve as a benchmark for those clinics using the dPNt. The PHE data are derived from quarterly returns from sexual and reproductive clinics across the country which code for services delivered, diagnoses made and for individuals attending as a partner of an infection.16

This PHE analysis of the partners who test provides the number of partners found to have a diagnosis of that infection. The number of partners diagnosed with an infection was subtracted from the total infections diagnosed in 2019 to give a corrected denominator so that the adjusted aggregate PN ratio for the year can be calculated.

In 2019, PHE reported 63,250 gonorrhea diagnoses and 23,598 gonorrhea contacts in England. Of the latter group 19,771 gonorrhea contacts tested and 8083 (41%) were diagnosed with the infection. The number of gonorrhea diagnoses in 2019 that were not associated with contacts was 55,167 (63,250–8083) and the adjusted aggregate PN performance was 0.43 (23,598/55,167) partners seen per index patient.

Similar in 2019, PHE reported 7931 syphilis diagnoses and 5298 syphilis contacts in England. Of the latter group 3974 syphilis contacts tested and 476 (12%) were diagnosed with the infection. The number of syphilis diagnoses in 2019 that were not associated with contacts was 7455 (7931–476) and the adjusted aggregate PN performance was 0.71 (5298/7455) partners seen per index patient.

To evaluate whether the patients who engaged with the platform were demographically different from those who did not, we ran hierarchical logistic regression models where each patient was an observation (coded 1 if they engaged with dPNt and 0 otherwise). The model included random intercepts depending on which health care worker processed the patient, and fixed effects for the demographic variables. These demographic variables included sexuality, age, gender, and ethnicity. We ran separate models for patients with gonorrhea and patients with syphilis, but some patients featured in both models (if they were diagnosed with both infections).

To test whether the demographic differences in engagement with the dPNt could account for the observed differences in preengagement PN we ran hierarchical Poisson models predicting the number of partners notified per patient, with random intercepts for which health care worker processed the patient and fixed effects based on dPNt engagement, and the significant demographic covariates (gender and age in relation to gonorrhea and age in relation to syphilis).

To establish if the use of the dPNt tool was independent from the standard PN process we examined correlations between the 4 ways that partners can be verified as being seen and tested. A positive correlation between standard PN and dPNt numbers would indicate overlap and potential double counting.

RESULTS

The dPNt tool was used by 23 clinics in 2019 and those providers from London accounted for 89% of the patients diagnosed with gonorrhea and/or syphilis. The demographics of the patients evaluated in this analysis is shown in Figure 4.

F4
Figure 4:
Gender, sexuality, ethnicity, and age breakdowns of the patients diagnosed with gonorrhea and syphilis evaluated in this article.

For gonorrhea we found that self-reported gender was a clear predictor of engagement. Men were significantly more likely to engage with SXT than women (coefficient = 0.46(0.13), z = 3.38, P > 0.001). In addition, older people were significantly less likely to engage with SXT than younger people (coefficient = −0.11(0.04), z = −2.43, P > 0.05). No other demographic variable significantly predicted engagement with the platform for patients with a gonorrhea diagnosis. For syphilis, only age was significantly associated with engagement in that older people were less likely to engage with the platform than younger people (coefficient = −0.28(0.09), z = −3.00, P > 0.01).

The number of partners seen and tested per index patient can been broken down into: the partners who have already been seen and tested before engagement with dPNt, those partners seen and tested prospectively during the follow-up call, and those partners signed off automatically as being seen and tested by dPNt (Fig. 5).

F5
Figure 5:
How the partners of patients diagnosed with either gonorrhea or syphilis were verified as tested by their engagement with the dPNt. Clinical staff reduced the prospective work to verify prospective partner testing when dPNt was used by 83% (1.04/1.26) and 84% (1.32/1.57) for gonorrhea and syphilis, respectively.

Compared with the adjusted aggregate PN delivered in England in 2019, overall clinics using SXT were able to increase the number of partners seen and tested per index patient from 0.43 to 0.84 for gonorrhea and 0.71 to 0.94 for syphilis. Of the total prospective partners verified as seen and tested, 49% and 52% of the partners of gonorrhea and syphilis, respectively, were signed off automatically by dPNt.

One fifth of patients diagnosed with gonorrhea (21%, 1044/4866) and a quarter diagnosed with syphilis (27%, 226/849) engaged with dPNt using step 2 (provider delivered PN) or step 3 (patient delivered PN) or both to inform partners. Of the index patients diagnosed with gonorrhea who engaged with dPNt, on average 2.63 partners were informed per index patient with 70% alerted by patient delivered PN, 73% of partners opened the dPNt link and 54% were verified as tested by the software. Of the index patients diagnosed with syphilis who engaged with dPNt, on average 3.03 partners were informed per index patient with 63% alerted by patient delivered PN, 75% of partners opened the dPNt link and 58% were verified as tested by the software.

We benchmarked these values for dPNt against the adjusted aggregate PN reported to PHE in 2019. In those index patients where dPNt was completed but no partners were informed using dPNt then 0.67 and 0.68 partners were verified per index patient diagnosed with gonorrhea and syphilis, respectively, which were 55% higher and 4% less when compared with the adjusted aggregate PN. In those index patients where at least one partner was informed by dPNt then 1.40 and 1.69 partners were verified per index patient diagnosed with gonorrhea and syphilis, respectively, which were 225% and 138% higher than the adjusted aggregate PN.

Gonorrhea patients who engaged with the dPNt software reported fewer partners tested before partner notification even when controlling for demographic covariates (coefficient = −0. 94 (0.05), z = −19.69, P < 0.0001). Similarly, syphilis patients who engaged with the platform reported fewer partners tested before partner notification when controlling for demographic covariates (coefficient = −1.07 (0.11), z = −9.77, P < 0.0001). Collectively, these results suggest that the differences in number of reported partners tested before partner notification cannot be explained solely by demographic differences between those who engaged with the platform and those who did not (Fig. 6).

F6
Figure 6:
The prenotification coefficient plots. Patients who engage with the dPNt platform tend to report fewer partners tested before partner notification, both for gonorrhea and syphilis. This holds true when controlling for relevant demographic variables. To create the coefficient plots the fixed effects from the hierarchical Poisson regression were transformed to incidence ratios. Only the significant effects are visualized with error bars showing 95% confidence intervals.

The correlation between the 4 ways that partners of patients with gonorrhea or syphilis are verified as seen and tested is shown in Figure 7. The only significantly positive correlation for either diagnosis was between partners verified via dPNt, either in clinic or by self-verification. The ratio of verification in clinic to self-verification of partners informed by dPNt is 1 to 2.35 and 1 to 1.87 for patients diagnosed with gonorrhea and syphilis, respectively. There were no significant correlations between standard PN and dPNt verification, suggesting that partners were not double counted.

F7
Figure 7:
The correlation tables for the 4 ways that partners are verified as seen and tested for patients diagnosed with gonorrhea or syphilis. Gray values are not statistically significant.

DISCUSSION

In this article, we have studied how the dPNt has been used by 23 early adopter clinics in 2019 and evaluated which factors were associated with engagement. The design and flow of the dPNt software (Fig. 2) invites patients to opt-in and inform partners whilst they are in the clinic, additionally consenting patients are sent an SMS and/or email with a link to dPNt so that they can deliver PN using their own mobile phone (Fig. 3). Opting in to engage with the software occurs before the patient is asked about the number of partners who were already told and tested; consequently, index patients are offered dPNt irrespective of the number of partners who have previously been told and tested.

There are a number of potential reasons for nonengagement with dPNt: (1) The patient has only had one or a limited number of partners and those people have already been diagnosed and/or tested for the infection. In this case nonengagement is rational and acceptable response to being offered dPNt. (2) The patient may have concerns about privacy and does not accept the reassurance of the health care provider using dPNt. During the development of dPNt, we decided to only use the minimum of personal information to deliver PN and measure impact, and that all personal information used is encrypted and hidden until the reminders are sent, and then, it is deleted from the software. The dPNt meets all the privacy standards for “technology-assisted contact tracing”17; however, due to the sensitivity of STI diagnoses it may be that some nonengagement relates to privacy concerns. (3) The use of dPNt requires that the patient and the partners have a smartphone. In the United Kingdom, 82% of all adults have a smartphone18; consequently, even though the smartphone ownership is higher in the younger population who are most at risk of an STI, some nonengagement will be related to a small but important number of patients without a smartphone. (4) the dPNt software has been written in English and therefore requires English proficiency for engagement. Across London in 2018, the percentage of adults who speak English at home is as low as 54%19; consequently, language will also be a barrier to engagement for some patients.

A significant fraction of patients diagnosed with gonorrhea (21%) or syphilis (27%) engaged with dPNt and informed at least one partner. The demographic analyses showed that young people are most likely to engage with the platform; however, we are unable from this analysis to determine if age is related to adoption by patients or if it affected the offer by clinical staff. In addition, age may also confound the smartphone ownership discussed earlier, early adoption of new technology or concerns about privacy. The majority of patients in this analysis self-identified as male and men who have sex with men and that is consistent with the demographics of patients who are diagnosed with gonorrhea and syphilis in the United Kingdom.3 The greater adoption of dPNt by men diagnosed with gonorrhea may result from an increase in disease burden within this group, as well as a history of repeat infection and familiarity with PN.

The ability to measure the number of partners that are already told and tested enables the prospective PN service delivery to be estimated more accurately. Standard PN typically combines the number of partners already told and tested at the time of treatment with the prospective PN work. Such prospective work involves either contacting partners on behalf of the patient, reviewing attendance at clinic or contacting the patient at 2 weeks and capturing details from them regarding the number of partners told and tested.

At the time of PN initiation with those index patients who used dPNt, 0.14 and 0.12 partners per index patient diagnosed with gonorrhea or syphilis had already been seen and tested, which equates to 10.0% and 7.1% of total partners verified as tested, respectively. The higher percentage of partners already seen and tested with gonorrhea is likely related to the earlier onset of symptoms with this infection. Looking at the prospective PN work done in patients who engaged with dPNt, we can see that the majority of partners, for gonorrhea 83% and syphilis 84%, were automatically verified as seen and tested by the software.

There were a very high number of partners already told and tested according to the patients diagnosed with gonorrhea who did not engage with dPNt. In fact, for gonorrhea this number was equivalent to the total number of partners seen per index patients based on PHE adjusted aggregate PN data. In addition to nonengagement with dPNt being rational, when all the partners have been told and tested, it is possible that the higher number of reported partners may represent a confirmation bias by the patient to support their nonengagement with dPNt and the PN discussion. In addition, this behavior may be compounded by an acceptability bias on the part of the health care provider using dPNt, during nonengagement, to avoid exploring the details of the information provided by the patient. Sending a dPNt link to the patient in step 3 enables more partners to be told than were declared in the clinic; consequently, framing step 3 by the clinician as an “opt-out” step could be used by the clinic to support more partners being told.

Health care workers do follow-up calls with patients to ascertain if they tolerated treatment, were abstinent in the posttreatment period, and to ask if any other partner were told and tested outside of dPNt. It is possible that some partners reported during these calls were also reported through the dPNt software. To investigate this potential double-counting, we computed the correlations between all the ways that partners are reported as seen and tested (results are shown in Fig. 7). The absence of correlation between the standard PN and dPNt verification steps implies that there was no significant double-counting and suggests that the 4 ways that partners are verified as seen and tested represent distinct partners. The only significant correlations between the 4 separate ways that partners were verified as seen and tested was when they were either signed off by a health care worker or the partner self-verified via dPNt. When partners are informed by dPNt a correlation is expected between the two separate verification routes. It is important to note that nearly twice as many partners self-verified than were signed off in clinic. The higher number of partner self-verifications is likely to reflect the use of online services by partners, as well as the lack of health care workers sign off when the partner attends the clinic.

There are a number of caveats that have been identified in the analysis presented in this paper. The first is that we were not able to determine the number of patients diagnosed with an STI where dPNt was not used. Even though the software simplifies PN delivery and provides real-time outcome data to the clinic, we cannot be sure that all patients are given the opportunity to use the software.

The second caveat is that this was an observational analysis of dPNt being used in the real-world setting and so does not have the structure and oversight that would be afforded by a research trial. This analysis has been done using the data from clinics that have opted in to use dPNt to support the delivery of their standard PN. This self-selection bias might limit the generalizability of the results to the rest of London and other cities in England.

The third caveat is that we have assumed that the patient was naïve to the use of dPNt and therefore the number of partners that were already seen and tested was captured after nonengagement. However, there will have been patients diagnosed more than once who will have used the tool previously. In addition, we cannot rule out the impact of the health care provider on how the dPNt is described to—and used with—the patient.

The fourth caveat is that this study chose a single year of analysis and compared the results to a national benchmark as opposed to a before and after the adoption of dPNt in each of the individual clinics. The latter approach would have enabled us to adjust for the baseline performance of each clinic; however, this was not possible because clinic adoption of dPNt was staggered between 2016 and 2018, when changes were being made to the software, such as the introduction of an integrated booking system, that would have affected performance.

The final caveat is that the adjusted aggregate PHE data include some of the clinics using dPNt and this is going to increase national PN performance and reduce the effect size of dPNt. In 2019, the 23 providers using dPNt for gonorrhea and syphilis managed 7.7% and 10.7%, respectively, of the total diagnoses made that year in England.

Evaluation during the development of dPNt showed that the services were acceptable to patients, partners, and health care providers. We now have a behavioral confirmation in the real world setting that a significant fraction of patients in 2019 did engage with dPNt and that automatic reporting by the software contributes significantly to the number of prospective partners reported as being seen and tested.

Multiple behavioral interventions are being planned and we are keen to support independently evaluated randomized controlled trials to understand the impact of future changes in dPNt use and software features. The value of dPNt to patients, partners and health care providers is likely to increase disproportionately as more testing services adopt the platform and stakeholders benefit from the network effect. These early results support the potential to scale up dPNt to deliver more effective PN.

REFERENCES

1. Unemo M, Bradshaw CS, Hocking JS, et al. Sexually transmitted infections: Challenges ahead. Lancet Infect Dis 2017; 17:e235–e279.
2. Holmes KK, Sparling PF, Stamm WE, et al. Sexually Transmitted Diseases, Fourth ed. New York, NY: McGraw-Hill Companies, 2008. ISBN: 978-0-07-159311-3.
3. Public Health England. Sexually transmitted infections and screening for chlamydia in England, 2019. Published September 2020, PHE publications, gateway number: GW-1518. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/914184/STI_NCSP_report_2019.pdf. Accessed September 3, 2022.
4. Ferreira A, Young T, Mathews C, et al. Strategies for partner notification for sexually transmitted infections, including HIV. Cochrane Database Syst Rev 2013; (Issue 10):Art. No.: CD002843.
5. Cowan FM, French R, Johnson AM. The role and effectiveness of partner notification in STD control: A review. Genitourin Med 1996; 72:247–252.
6. Ward H, Bell G. Partner notification. Medicine 2014; 42:314–317.
7. Public Health England. Sexually transmitted infections (STIs): annual data tables. Table 6: STI diagnoses and partner notification, 2016 to 2020. Available at: https://www.gov.uk/government/statistics/sexually-transmitted-infections-stis-annual-data-tables. Accessed September 3, 2022.
8. Turner K, Adams E, Grant A, et al. Costs and cost effectiveness of different strategies for chlamydia screening and partner notification: an economic and mathematical modelling study. BMJ 2011; 342:c7250.
9. Oxman GL, Doyle L. A comparison of the case-finding effectiveness and average costs of screening and partner notification. Sex Transm Dis 1996; 23:51–57.
10. Reynolds SL, Kapadia AS, Leonard L, et al. Examining the direct costs and effectiveness of syphilis detection by selective screening and partner notification. J Public Health Med 2001; 23:339–345.
11. Public Health England. NCSP: re-testing of positive chlamydia cases report. Published August 1, 2013. Available at: https://www.gov.uk/government/publications/ncsp-re-testing-of-positive-chlamydia-cases-report. Accessed September 3, 2022.
12. Htaik K, Fairley CK, Bilardi JE, et al. Evaluation of the online partner messaging service for Sexually Transmitted Infections Let Them Know. Sex Transm Dis 2022; 49:12–14.
13. Murphy RD, Wohl AR, Ma Y, et al. Adaptation of the HIV care continuum as a method for evaluating syphilis and gonorrhea disease control activities in Los Angeles County. Sex Transm Dis 2015; 42:686–690.
14. SXT Health Community Interest company, Registered UK: Company Number: 7310310. Available at: http://sxt.health. Accessed September 3, 2022.
15. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2020. Available at: https://www.R-project.org/. Accessed September 3, 2022.
16. GUMCAD STI Surveillance System. Available at: https://www.gov.uk/guidance/gumcad-sti-surveillance-system. Accessed September 3, 2022.
17. American Civil Liberties Union (ACLU)–ACLU White paper ‘Principles for technology-assisted contact-tracing. Available at: https://www.aclu.org/report/aclu-white-paper-principles-technology-assisted-contact-tracing. Accessed September 3, 2022.
18. Online Nation 2020 report OfCom. Available at: https://www.ofcom.org.uk/__data/assets/pdf_file/0027/196407/online-nation-2020-report.pdf. Accessed September 3, 2022.
19. First language spoken at home, Borough. Labour force study, Office of National Statistics, 2018. Available at: https://data.london.gov.uk/dataset/first-language-spoken-home-borough. Accessed September 3, 2022.
Copyright © 2022 American Sexually Transmitted Diseases Association. All rights reserved.