Abstract
Background
Given the most recent report of the nationwide rise of congenital syphilis (CS), with over a 773% increase from 2012 to 2021 per the Center for Disease Control and Prevention, we sought to utilize penicillin administration and ICD-10 billing data as proxies to assess the accuracy of diagnosis of congenital syphilis among major tertiary care pediatric hospitals.
Methods
This retrospective cohort study drawing data from 49 major tertiary children’s hospitals in the United States sought to determine whether administration of penicillin in infants 30 days or younger, excluding other common infectious diseases treated with the medication, correlated with recently reported congenital syphilis epidemiologic data.
Results
2290 infants met inclusion criteria and received penicillin therapy, excluding infants with the most common secondary infections treated with penicillin, with 1123 (49.3%) of those included not having a syphilis diagnosis made. Of the neonates with a coded diagnosis of CS, (1107/1162) 95.3% received more than 1 day of penicillin therapy, and those not coded for CS that received more than 1 day of penicillin therapy was 37.9% (428/1128).
Conclusions
Our findings found a significant number of infants who had penicillin administration and laboratory and procedural workup performed suggestive of congenital syphilis consideration, without a congenital syphilis diagnosis billed. This may reflect an underestimation of what is notably a growing nationwide pandemic. By addressing the screening and treatment needs of patients with CS, we can help to address some of the socioeconomic inequities in pediatric and maternal healthcare, and further accurately characterize the extent of this increasingly prevalent disease process.
Introduction
Congenital syphilis (CS) results from the vertical transplacental transmission, or direct gynecologic mucosal contact in the perinatal period, of Treponema pallidum from mother to fetus. 1 Morbidity and mortality outcomes associated with CS include stillbirth, neonatal death, deafness, neurological impairment, and skeletal deformities.2,3 The American College of Obstetrics and Gynecology (ACOG) recommends universal syphilis screening at the first prenatal visit in the first trimester and rescreening in the third trimester and at delivery, including those who have exposure to a partner infected with syphilis, live in areas with high syphilis morbidity rates, or other risk factors including unprotected sexual encounters.4,5 In pregnant patients testing positive for syphilis, timely treatment with penicillin reduces the risk of CS by 97% and stillbirth by 82%. 4
Recently though, rates of primary and secondary syphilis in women tripled in the United States from 2011 to 2018, and CS cases also increased significantly in recent years.3,4,6 The Centers for Disease Control and Prevention (CDC) reported that CS increased 773% (from 2.1 to 16.3 per 100,000 population) during 2012–2021. 7 Cases of congenital syphilis that meet CDC’s standardized case definition of ‘Probable’ or ‘Confirmed’ are reported to the CDC’s National Notifiable Disease Surveillance System. (NNDSS) 8 They are defined based off either adequate maternal treatment prior to delivery or abnormalities noted on neonatal clinical exam or diagnostic work up with serum tests, long-bone radiograph and/or lumbar puncture.7,8 The most common missed prevention opportunities are lack of timely testing, adequate treatment during pregnancy and by a lack of timely prenatal care.6,7,9 Additional factors include health insurance limitations between different payer types and among those uninsured, and economic disparities.1,9–12 Data from the Western US and Southern US also identified context-dependent factors as gaps in CS prevention9,11 suggesting possible geographic variation in treatment and prevention strategies. Despite the reported rise in cases, there still may be an under-reporting of infants diagnosed and/or treated for CS who met the CDC STI Treatment Guideline definition of Possible or Less Likely CS, and subsequently NNDSS, in the setting of inadequate maternal treatment but were not reported due to negative infant serum rapid plasma reagin (RPR) in the setting of false negative findings or delayed seroconversion.8,13,14
The CDC recommends penicillin as first-line therapy for CS.14–16 There are currently very few uses for the administration of intramuscular (IM) benzathine or intravenous (IV) aqueous penicillin in neonates outside of CS.17–22 This suggests that benzathine or aqueous penicillin administration could be used as a proxy for the diagnosis of CS in administrative datasets. Given the significant rise in cases of CS, 7 it is important to ensure accurate accounting of CS so healthcare facilities and departments of public health may monitor for effective preventative and therapeutic interventions. The purpose of this study was to utilize penicillin administration data from an administrative dataset, representing a cohort of tertiary children’s hospitals, to assess the appropriate presence of a corresponding CS diagnosis by comparing administrative diagnosis codes and examining characteristics of infants who received parenteral penicillin therapy.
Methods
Design and data source
A cross-sectional, retrospective cohort study was performed for patients admitted to tertiary referral children’s hospitals participating in the Pediatric Health Information System (PHIS; Children’s Hospital Association, Lenexa, KS). PHIS is a comparative administrative pediatric database including demographic and resource utilization data for inpatient, ambulatory, surgery, emergency department, and observation unit patient encounters for 49 U.S. children’s hospitals, from all major geographic regions of the country. Data available include standard hospital discharge abstract information, including demographics, procedures, and laboratory tests, International Classification of Diseases, 10th Edition (ICD-10) diagnoses 23 (ICD-10 2022), All Patients Refined Diagnosis Related Groups (APR-DRGs), 24 as well as daily financial and resource utilization data.
Patient selection
Patients admitted to a participating children’s hospital who received at least 1 day of either intravenous or intramuscular Penicillin G between January 1, 2017, and December 31, 2022, were considered for inclusion. The available formulations assessed included Benzathine, Aqueous, and Procaine, which are the formulations that may have been clinically available on the market for US pediatric hospitals. Age criteria included all infants between 0 and 30 days old at the time of hospital admission. Infants who received a coded diagnosis of congenital syphilis or other syphilis diagnoses but did not receive parenteral penicillin (n = 111) were excluded as these children may have been diagnosed for CS at birth, and transferred or readmitted to a participating children’s hospital for an unrelated reason, or may have had untreated CS. With the inability to determine exact reasoning and distinguish between these, this was our justification of omission. Additionally, infants who received a diagnosis of Group A Streptococcus, Group B Streptococcus, Streptococcus pyogenes, Streptococcus agalactiae, Listeria monocytogenes, Pneumococcal, or Streptococcus pneumoniae infection were excluded to eliminate confounding data (Figure 1). Consort diagram showing inclusion and exclusion criteria for infants 30 days or younger who received parental penicillin administration from 2017 to 2022.*IM or IV penicillin formulations = IM Penicillin G benzathine, IM Penicillin G procaine, IM Penicillin G benzathine and procaine, IV Penicillin G aqueous. **Non-CS syphilis diagnosis = primary acquired syphilis, secondary acquired syphilis, early latent syphilis, late latent syphilis, neurosyphilis from acquired syphilis, tertiary syphilis. 1Streptococcus pyogenes. 2Streptococcus agalactiae. 3Streptococcus pneumoniae.
Definitions
The cohort was divided into two groups: neonates with and without a syphilis diagnosis. Demographic variables including gestational age at birth, sex, race, ethnicity, birth weight, payor/insurance, hospital geographic region, and source of admission were assessed. The Childhood Opportunity Index (COI) score, a validated metric for estimating multi-dimensional social determinants of health was also assessed. 25 Race and ethnicity were extracted through electronic health record data and were collected and assigned as per the protocol of each hospital system. Race and ethnicity were included in the analysis as previously published data suggest significant differences in outcomes related to syphilis based on race and ethnicity.26,27 Procedure codes and charges were used to identify patients who underwent diagnostic and/or therapeutic procedures associated with a diagnosis of congenital syphilis (Supplemental Table 1).
Statistical analysis
Categorical variables were summarized with frequencies and percentages, which were compared using Chi-square tests. The continuous variables for the two groups were reported as medians with interquartile ranges and compared using Wilcoxon rank-sum tests. Means with standard deviations were also provided to enhance readability. Both of the aforementioned groups were further partitioned into those who received 1 day of penicillin and those who received more than 1 day of penicillin, and the same tests were performed. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), and p < 0.05 was considered statistically significant. This study was considered nonhuman subjects research by the institutional review board (IRB23-1881).
Results
There were 771,182 hospital encounters during the study period that met inclusion criteria. Following the application of exclusion criteria, 2290 encounters (0.3%) had parenteral penicillin administered. Of the 2290 encounters, 1162 (50.7%) were with a CS diagnosis, and 1123 (49.3%) without (Figure 1). In Figure 2, variations in the duration of penicillin administration among children’s hospitals were visually observed for the entire study cohort. Also, similar variations were observed when limited to patients with a CS diagnosis (Figure 3). Hospital variation in the duration of penicillin administration for all infants in the study cohort. Hospital variation in the duration of penicillin administration for infants with a diagnosis of congenital syphilis. The X-axis is the number of patients, and the Y-axis indicates a hospital in our cohort spanning 48 children’s hospitals, which can be followed in this paper.
Demographics of infants who received penicillin partitioned by diagnosis of congenital syphilis.
Values of table are in numeric instances “n” with associated percentages (%), unless otherwise specified. CS = congenital syphilis, IQR = interquartile range, SD = standard deviation. There were 256 patients with missing gestational age. There were 2 patients with missing sex. There were 90 patients with missing birth weight.There were 6 patients with missing COI.
Demographics of infants who received penicillin partitioned by diagnosis of congenital syphilis and the duration of penicillin administration.
Values of table are in numeric instances “n” with associated percentages (%), unless otherwise specified. CS = congenital syphilis, IQR = interquartile range, SD = standard deviation, LFT’s = liver function tests, CBC = complete blood count, RPR = rapid plasma regain, CT = computed tomography, MRI = magnetic resonance imaging, VDRL = Venereal disease research laboratory test, FTA-ABS/FTA = fluorescent treponemal antibody-absorption. *The sample size for this analysis is less than 20, and 2 cells have less than 5 neonates. Thus, the chi-squared test for this row is invalid.
Discussion
In this retrospective cross-sectional study of a large administrative dataset of 48 U.S. children’s hospitals, penicillin administration data was able to identify infants evaluated and treated for CS. Furthermore, the data revealed a cohort of patients who received parenteral penicillin therapy but did not have a corresponding diagnosis code of CS. Specifically, among all infants who received parenteral penicillin but did not have a billing diagnosis of CS, 72–80% of infants treated with some duration of penicillin had testing for CS performed, suggesting at least some clinical concern for CS by the treating clinicians.
The overall trends of infants who received penicillin coincided with nationwide geographic trends of CS, 28 with the highest group of this cohort being from the South (44.3%). However, among those in our cohort who received a diagnosis of CS, the highest was from the West (39.4%), which may reflect that there may be geographic disparities in accurate billing or documentation skewing true nationwide estimates. This also may reflex geographic differences in clinical practice, or regional variation in knowledge and exposure to billing and coding practices,29,30 regional differences in perception and enforcement of medical billing practices, 31 or higher incidences of infants who delivered outside of children’s hospital settings in rural areas which may be found in the West and Southern regions compared to others.32,33 This may suggest that outborn infants included in this analysis may have been transferred to receive therapy, but did not have the diagnosis adequately documented and coded upon arrival to the hospital.
Our study shows a potential association between socioeconomic factors and the prevalence of CS. Approximately 70% of infants in our cohort lived within Low or Very Low COI areas, and nearly 85% of infants who had a diagnosis of CS had insurance through Medicaid, while about 5% had insurance provided by a private vendor. These findings are consistent with recent data revealing associations with higher rates of CS in areas with adverse social determinants of health.6,14,25 Also our study noted that among all infants who received penicillin, regardless of a CS diagnosis or not, were disproportionately affected with poor COI scores which aligns with recent epidemiologic studies.4,34 This suggests an opportunity to address CS as a means to target resources at populations facing significant healthcare inequities.
The diagnosis of CS may be clinically challenging. The 2021 CDC STI Guidelines note that CS can be classified in four distinct categories or scenarios: (1) Confirmed proven or highly probable; (2) Possible; (3) Less Likely; and (4) Unlikely, which affects the treatment course. 14 However, diagnosis of congenital syphilis can be difficult because maternal nontreponemal (RPR) and treponemal immunoglobulin G (IgG) antibodies can be transferred through the placenta to the fetus, complicating the interpretation of reactive serologic tests for syphilis among infants who are less than 30 days of age, which may not necessarily warrant treatment nor diagnosis.35,36 Additionally, nationwide cases of CS may be dependent on reporting practices of infant RPRs to departments of public health and then to the CDC. To add to further complexity to the scenario, the most utilized diagnosis coding system utilized, the International Classification of Diseases (ICD), does not have these CDC diagnostic definitions or reported surveillance as codes, but rather more broad terms such as “Congenital syphilis, unspecified” (ICD-10-CM A50.9) as well as symptom-specific codes such as “Early congenital syphilitic pneumonia” (ICD-10-CM-A50.04s). 23 As the process of creating a CDC-defined clinical diagnosis (Symptomatic/proven, Possible, etc.) 14 is that translated into a discordant ICD-10 billing code by the provider, 23 and then reported to departments of public health who must create a different CDC recommended surveillance definition of probable or confirmed, 8 there are multiple steps in the process where errors and misclassifications may occur. Adding to this complicated process, there are scenarios where the infant may be asymptomatic, but still warrant a diagnosis of CS with penicillin treatment dependent on the adequacy of maternal treatment and/or maternal and nontreponemal serology results, even in the setting of a negative infant RPR. In certain settings, these infants are diagnosed and treated for CS, but not reported to local or national public databases.13,14 These various barriers may lead to significant heterogeneity in diagnosing, managing, and reporting this increasingly prevalent disease, which further reinforces the importance of the findings in our study and the potential for using pharmaceutical administration data to assist in accurately identifying the extent of the current epidemic. This may explain some of the discrepancies that are seen particularly among those infants in our cohort who received penicillin & diagnostic workup for CS but did not receive a coded diagnosis. Additional explanations could include that the children’s hospitals receive referrals and care for infants where diagnosis is uncertain, or there is a complicated prenatal care situation where information is sparse and/or there is no means of obtaining comparative maternal data, so there is a low threshold to evaluate and treat infant for CS empirically regardless of final diagnosis.
As local and national departments of public health continue to reassess case and surveillance definitions related to CS, it may be advantageous to synchronize definitions and terminology that more closely reflect those used in clinical billing practices. Additional studies looking into other primary acquired sexually transmitted infections37,38 and sexual assault cases39,40 have also found discrepancies between ICD-10 coding and surveillance case definitions which may lead to the exclusion of true cases and affect the accuracy of surveillance data.
There were multiple limitations in our study. The administrative database contains which tests and procedures were performed, but not the results of the tests. Additionally, we did not have access to maternal medical records or birth records of outborn infants. This limits our ability to confirm a diagnosis of CS that was billed by the providers or stratify by CDC Case Definition. 14 While we were able to see if various formulations of penicillin were administered, we were unable to view specific doses of administration which also could more accurately reflect if it were given specifically for CS indications. There may be rare situations where a patient requires penicillin in situations such as congenital asplenia or heterotaxy that was not captured in our initial analysis. 41 Given that 28.3% of neonates without a CS diagnosis were born at hospitals outside of the cohort of hospital studies, there may even be a greater number of infants who had a CS diagnosis considered through these laboratory tests and secondary procedures obtained at their birth hospital that were not captured in our analysis. Among those cases with a diagnosis of CS, but no penicillin administration, they may have been born at hospitals outside the cohort and already received their effective penicillin management but may have required escalated care for other secondary reasons at the tertiary care children’s hospital. It may also be due to a diagnosis coding error on the part of the provider either at the outside hospital or within the included children’s hospital. Both of these two limitations regarding those born outside the hospital system may introduce the risk of misclassification bias. Another limitation to our interpretation of the findings is that infants could simply be receiving penicillin as a part of routine or targeted empiric septic states or specific bacterial infections. We sought to account for this by eliminating cases that had diagnoses for the most likely pathogens to cause sepsis in a neonate that is treated with penicillin.42,43 To expand on this, patients with lower COIs have been noted to have worse critical care outcomes in previous studies, and given the high overlap of lower COI and penicillin use in our cohort, it’s possible providers may deviate from sepsis guidelines by utilizing penicillin for empiric sepsis coverage.44,45 Among infants who did not receive a diagnosis of CS, penicillin may have been utilized as the drug of choice in these hospitals for empiric sepsis management. Although possible, this would be a clinical practice contrary to published evidence-based recommendations19–22 and accentuating the heterogeneity in the approach of neonatal sepsis management.
Conclusion
In this retrospective cohort analysis of neonates admitted to tertiary referral children’s hospitals, we identified a discrepancy between patients treated for CS and those receiving a diagnosis of CS. Additionally, we observed a higher prevalence of lower COI scores among those infants who received penicillin and received a diagnosis of CS. Addressing the screening and treatment needs of patients with CS can help to address some of the socioeconomic inequities in pediatric and maternal healthcare, accurately characterize the extent of this increasingly prevalent disease, and optimize targeted and efficient public health interventions.
Supplemental Material
Supplemental Material - Assessing diagnostic accuracy of congenital syphilis using penicillin administration data through the pediatric health information system (PHIS) database
Supplemental Material for Assessing diagnostic accuracy of congenital syphilis using penicillin administration data through the pediatric health information system (PHIS) database by John Flores, Natalie Grills, Jason Kane, Lilly Cheng Immergluck, Nikki Kasal, Madan Kumar and Allison Bartlett in International Journal of STD & AIDS.
Footnotes
Acknowledgements
We would like to graciously acknowledge Dr Matthew Hall from the Children’s Hospital Association, and Dr Yeo-won Anh and Dr Julia Rosebush from University of Chicago Comer Children’s Hospital for their contributions to the conceptualizing and design of the study. Additionally, the lead author John Flores is supported by the University of Chicago Primary Care Investigators Training in Chronic Disease & Health Disparities (PITCH) Fellowship (Health Resources and Services Administration T32 HP42019) and we would like to acknowledge their support.
Author contributions
Dr John Flores conceptualized and designed the study, collected data, carried out the initial analyses, drafted the initial manuscript, and critically reviewed and revised the manuscript. Natalie Grills designed the data collection instruments, collected data, carried out the initial analyses, and critically reviewed and revised the manuscript. Nikki Kasal conceptualized and designed the study, assisted in drafting the initial manuscript, and critically reviewed and revised the manuscript. Drs Allison Bartlett, Jason Kane, and Madan Kumar conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This was a cross-sectional, retrospective cohort study of patients admitted to tertiary referral children’s hospitals participating in the Pediatric Health Information System (PHIS) through the Children’s Hospital Association (Lenexa, KS). The PHIS is a comparative administrative pediatric database including demographic and resource utilization data for inpatient, ambulatory surgery, emergency department, and observation unit patient encounters for more than 54 U.S. children’s hospitals. This is funded by contracts with the participating children’s hospitals, including Comer Children’s Hospital through the University of Chicago. John Flores is supported by the University of Chicago Primary Care Investigators Training in Chronic Disease & Health Disparities (PITCH) Fellowship (Health Resources and Services Administration T32 HP42019). No additional funding was secured for this study. The content is solely the authors’ responsibility and does not necessarily represent the official views of the funding organizations.
Supplemental Material
Supplemental material for this article is available online.
Appendix
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
