In 2013, 1,401,906 cases of Chlamydia trachomatis (CT) and 333,004 cases of Neisseria gonorrhoeae (NG) were reported to the Center for Disease Control and Prevention (CDC).1 Although both of these numbers represent decreases more than 2012, the case rate of 446.6/100,000 for CT and 106.1/100,000 for NG still represent a significant disease burden.
More than 117 million health care visits are made to emergency departments (EDs) in the United States each year.2 Many visits, especially among adolescents, are related to sexually transmitted infections (STIs)3 and up to 7% of people seeking STI treatment do so in the ED.4 This number may increase as funding for STD clinics decreases. Diagnosis of NG and CT in ED settings is typically made by nucleic acid amplification tests performed on cervical or urethral swabs or urine specimens and have excellent sensitivity and specificity.5 However, providers do not have results at the time they make decisions about whether to treat their patient with antibiotics.
In 2010, the CDC revised guidelines for management of STIs. They recommend presumptive antibiotic treatment for those at increased risk, including those 25 years or younger, with new or multiple sex partners, and engaging in unprotected sex, especially if follow-up cannot be ensured.6
The aim of this study was to determine the proportion of patients treated appropriately, both presumptively in the ED and at follow-up, among patients who ultimately tested positive for NG or CT. Secondary aims were to characterize patients that received appropriate presumptive antibiotic treatment of both NG and CT at ED visit and to compare them to patients with disease who did not receive appropriate presumptive therapy.
MATERIALS AND METHODS
This analysis of 500 NG and CT cases was part of a larger retrospective study. The parent study was an unmatched case-control chart review in which 500 cases (positive for either NG or CT) and 500 unmatched controls (negative for both NG and CT) were randomly selected from 6997 patients who had specimens sent for either NG or CT from the ED between January 2010 and June 2011. This subgroup analysis of cases only was determined before the study commenced to specifically examine ED and follow-up treatment among those with confirmed disease. It was approved through expedited review by the university's institutional review board with a waiver of informed consent.
Study Setting and Population
The study site was an urban, academic ED that receives 130,000 annual visits. The medical record in use in the ED was electronic (Picis IBEX). The ED has a dedicated “callback” nurse who is notified of positive test results and attempts to contact the patient to inform them and encourage treatment, and documents these attempts in the medical record. Cases unable to be reached are referred to the state Department of Public Health (DPH) for further outreach.
All adult and pediatric ED patient visits in the eligible period who had a GenProbe Aptima Combo TMA Assay test result for NG or CT were considered eligible for chart review. A random sample of 500 that were positive for either NG or CT (cases) were selected from the total number of cases and included in data analysis. We included all patients older than 15 years. We excluded from analysis any patient who had a test repeated within 3 weeks of a positive test result as they were considered treatment follow-ups rather than “new” infection, victims of sexual assault, eloped before physician evaluation, and patients whose laboratory specimen was inadequate to interpret results.
To improve accuracy and minimize inconsistencies, previously accepted methodological guidelines were followed up for chart review.7 In brief, before the start, variables were defined and categorized. Three chart abstractors were trained using 10 “practice” medical records and a standardized data collection form. Data coding rules were developed and meetings were held throughout the abstraction period with all investigators and abstractors present to resolve discrepancies in findings. Double-data abstraction was completed on a 10% random sample of charts.
Data obtained from the medical record included test results for NG and CT, demographic data, presence/absence of symptoms and risky behaviors in the patient's history of present illness, presence/absence of physical examination signs documented by ED provider, and antibiotics given during ED visit and prescribed for outpatient use. Charts with a positive result for NG or CT were further analyzed for successful follow-up contact to inform patient of test result. For those not given antibiotics at initial visit, contact by the follow-up nurse, presence of subsequent ED visits with medications administered or prescribed, or phoned/faxed-in prescriptions of appropriate antibiotics were abstracted as well.
Antibiotics were considered appropriate for NG or CT if they were listed as appropriate therapy in the CDC publication, 2010 STD Treatment Guidelines (see Table 1). A patient was considered to have received correct presumptive treatment during their ED visit if they were given antibiotics appropriate for both NG and CT.
Variables with overlapping content (e.g., symptoms and examination findings) and associations in similar directions were collapsed into clinically relevant broad categories. If any one of the variables was recorded in the chart as present, the entire category was considered positive. These categories were as follows: behavioral risk factors, history or reported exposure to STI, unprotected sex, or new/multiple sexual partners; male genital signs/symptoms, urethral discharge, dysuria, testicular or prostate tenderness, lesions/swelling/rash, or genital burning/itching; female genital signs/symptoms, vaginal/cervical discharge or bleeding, dysuria, dyspareunia, adnexal tenderness, cervical motion tenderness, lesions/swelling/rash, or genital burning/itching; systemic signs/symptoms, fever >100.4°F, abdominal pain, or abdominal tenderness; and increased risk, designation based on the CDC definition6: urethritis, cervicitis, and any of the following: age <25 years, new or multiple partners, or unprotected sex.
Patients with multiple visits were included once in the summary of the demographic features. Incomplete charts were excluded from analysis for questions that were missing items but included in the rest of statistical analysis. If particular historical or physical examination features were not documented, they were considered negative.
Primary Data Analysis
The 18-month time frame for this descriptive study was chosen for feasibility and to allow enough participants to identify meaningful trends without being participant to anomalies that may have skewed data over a shorter time frame. A sample size calculation was performed based on the larger study design (unmatched case-control) and the relative proportion of patients the sample who tested positive for NG and/or CT (50% cases, 50% controls) and led to an overall sample size of 1000 (500 cases and 500 controls). The subanalysis described here was primarily descriptive, and the sample size was adequate to describe the population of interest (maximum width of the confidence interval [CI] for the estimate of the proportion treated is 0.09).
Data were analyzed using R v. 2.15.2 (www.r-project.org). Our primary outcome of interest was the proportion of patients treated appropriately, either in the ED or on follow-up, among those patients who tested positive for either NG or CT (i.e., cases).
Secondary Data Analysis
Our secondary aims were to characterize patients who received appropriate presumptive antibiotic treatment of both NG and CT at ED visit and to compare them to patients who did not receive appropriate empiric therapy. Descriptive statistics were generated for demographics, behavioral risk factors, genital and systemic symptoms, and high-risk categorization. Logistic regression models were fit to estimate both crude and adjusted odds ratios to determine factors that were associated with correct presumptive treatment. Generalized estimating equations (GEEs) with a compound symmetry correlation structure were used to adjust for multiple visits by the same patient. We also tested for potential interactions between demographic variables and risk factors.
Interrater Reliability Analysis
From the 10% of charts analyzed by 2 reviewers, 32 data elements considered most likely to be divergent were compared using the κ statistic to calculate interrater reliability.
During the study period, 6997 visits had specimens sent for NG and CT. Eleven patients were younger than 15 years, 3 visits were for sexual assaults, and 6 patients eloped before physician evaluation and were excluded from further analysis. Twelve additional specimens were found to be from visits within 3 weeks of a positive result and were excluded. There were 6905 included specimens sent for CT for which the laboratory reported an interpretable (positive or negative) result. Of these, 631 were positive for an overall CT prevalence of 9.1% during the study period. There were 6880 specimens sent for NG for which the laboratory reported an interpretable result. Of these, 213 were positive, for 3.1% prevalence.
Among 500 randomly selected cases included in further analysis, there were 484 unique patients: 468 with 1 visit and 16 with 2 visits. Demographic and visit characteristics are shown in Table 2. Mean age was 24.8 years (range, 15–62 years). Fifty-one percent were female, most were English speaking (93%), and most were African American (73%). Approximately one-third (81/500; 33.2%) had a chief complaint related to STI, 27.6% had a genital or urinary chief complaint, 23% had abdominal pain as a chief complaint, and 16% had a chief complaint unrelated to genital or abdominal complaints, and it is unclear why STI testing was ordered. No patient was admitted to the hospital.
A total of 268 (54%) of 500 (95% CI, 49%–58%) patients were treated in the ED at initial visit with appropriate antibiotics for the STI for which they eventually tested positive. This included 174 (47%) of 374 appropriately treated for their CT, 69 (94%) of 93 appropriately treated for their NG, and 25 (76%) of 33 appropriately treated for their dual infection with NG and CT. More people with NG were treated appropriately than with CT (P < 0.0001). An additional 157 (31%; 95% CI, 27%–35%) were successfully contacted by the callback nurse and were treated during the recommended follow-up visit at our medical center, for a total of 425 (85%; 95% CI, 82%–88%)) eventually receiving appropriate antibiotics at our institution for their infection.
Seventy-five patients did not receive appropriate antibiotics in the ED or in follow-up at our institution. Of those, 14 (19%) reported to the callback nurse that they were treated at another facility, 17 (23%) were unable to be contacted by the callback nurse, and 44 (59%) were successfully contacted by the callback nurse but did not return to our facility for treatment or report treatment elsewhere. Appropriate treatment at ED visit or follow-up visit is displayed in Figure 1.
At initial ED visit, 230 (46%) of patients received presumptive antibiotics appropriate for both NG and CT, in compliance with CDC recommendations. This is in contrast to the control arm of the study, of which 76 (15%) of 500 received presumptive antibiotics. The difference was statistically significant (P < 0.0001). Among cases, patient characteristics associated with correct presumptive treatment in unadjusted analyses included age >25 years, male, and CDC criteria for increased risk (Table 3). Race and language were not associated with presumptive treatment. There was a significant difference in how risk factors/symptoms interacted with sex, so results are reported differentially by sex. Among males, characteristics associated with correct presumptive antibiotics included behavioral risk factors and genital symptoms. Among females, only behavioral risk factors are associated with correct presumptive treatment. Among both males and females, there was a nonstatistically significant trend toward increased presumptive antibiotic treatment in the absence of systemic symptoms.
In the multivariate logistic regression model (Table 2), we report results from a model with a (significant) interaction between sex and behavioral risk factors, genital symptoms, and systemic symptoms. Because of colinearity between CDC-defined increased-risk criteria and other risk factors in the model, CDC criteria were not included in the final multivariate model. In this model, the only factors independently associated with appropriate presumptive antibiotics were behavioral risk factors and, for males, genital symptoms. To assess fit of the model, we fit the model without GEE and found that the model fit the data very well (Hosmer-Lemeshow, P = 0.76) and the resulting estimates and CIs were similar to the GEE results. This is likely due to the paucity of correlated data and gives indication that the model fit was sufficient. Additional regression diagnostics indicated that misspecification of the model was unlikely.
There were 32 data fields abstracted that were considered for interrater reliability. Overall, there were 3296 entries (103 patients with 32 entries). Of these records, 94% were in agreement between 2 abstractors, for a κ value of 0.92. In addition, each of the 32 questions was considered separately and κ was calculated. Of these, 29 (90.6%) of 32 had a κ value of at least 0.8. Those with a κ value of less than 0.8 were fever in females (κ = 0.79), dysuria in females (κ = 0.56), and abdominal tenderness in females (κ = 0.60).
Although there are multiple studies evaluating the prevalence of STIs in the ED (both clinically evident and clinically silent ones diagnosed in screening programs) and practitioner's ability to predict the presence of STIs in particular patients, this study represents one of the largest analysis of treatment of NG and CT diagnosed in the ED setting.
The ED has long been a venue that administers care to patients with STIs. It is hoped that the Accountable Care Act will drive more patients into primary care facilities. However, there will still be portions of the population not covered, and Accountable Care Act may contribute to decreased funding for public STI clinics, leaving EDs as the only safety net.8,9 If patients continue to come to the ED as their choice locale to receive care for their STI, it is incumbent on those working in the ED to ensure that these patients are getting appropriate antibiotics. A significant proportion of ED visits are made by people without other sources of care.10–13 The ED visit may provide the only chance to reach patients who might not otherwise interface with health providers.
Although our providers were somewhat able to predict the presence of disease in that they treated 3 times more cases presumptively than controls, they still treated only 54% of cases of either NG or CT appropriately in the ED. This is in line with other studies, which have consistently found that providers are ineffective in predicting who has STIs. In 1994, in a study looking at women presenting to an ED, Yealy et al.14 found that only 53% of culture-positive females received appropriate antibiotics before discharge. Bachmann et al.15 found that only 42% of women were appropriately treated at the time of ED visit, and in 2011, in an adolescent population, Pattishall et al.16 found that 59% of patients were treated at ED visit.
In recognition of the difficulties of predicting STI positivity, the CDC recommends empiric treatment for symptomatic patients who are at increased risk for infection or are “unlikely to return for a follow-up evaluation.” Our providers were somewhat effective in predicting disease, in that 3 times more patients were treated presumptively compared with controls. They were more likely to treat patients with behavioral risk factors presumptively; however, many patients meeting CDC-increased risk criteria were not given antibiotic treatment at ED visit. Furthermore, our data indicate that although those with positive STI test results are likely to have behavioral risk factors, even if physicians treat patients with increased risk presumptively, they would fail to treat half of patients who eventually test positive. Other criteria for detecting high-risk patients have been developed, but none has shown sufficient sensitivity, specificity, and consistency across different patient populations to be recommended for decision making.17,18
Providers were 15 times more likely to treat males presumptively for STIs, compared with females. This has been suggested in one prior study as well.16 This is likely due to both the increased likelihood of symptoms in males and the lack of other differential diagnoses for males with potential STIs compared with females. It is, however, concerning because most long-term complications from untreated STIs occur in females. Providers were more likely to appropriately presumptively treat patients found ultimately to be infected with NG than they were to treat patients found to be infected with CT, likely secondary to the prominent symptoms those having NG manifest compared with those with CT.
Our results support use of a callback system, as the follow-up nurse identified and coordinated treatment for an important proportion of patients who might otherwise have not been treated. However, the difficulties of poor follow-up that resulted in 15% of patients ultimately not receiving appropriate treatment of their STI at our institution is consistent with reports from other EDs. Al-Tayyib et al.,19 partnering with a local public health department to improve follow-up, reported that 20% did not get contacted or treated after 20 telephone contact attempts and certified mail. Pattishall et al.16 found that 43% did not receive appropriate antibiotics in follow-up. Bachmann et al.,15 working collaboratively with their local Department of Health Disease Intervention Specialists, were able to limit lack of treatment to only 8% of those initially untreated.
Our data add to the body of evidence that even with guidelines advocating presumptive treatment of STIs and dedicated programmatic attempts to identify cases of NG or CT after ED visit, a large number of identified patients are still not appropriately treated in a timely fashion. One solution would be presumptive treatment for all patients for whom NG and CT tests are sent. That would result in a large number of noninfected patients being unnecessarily treated, causing unnecessary adverse effects and the emergence of drug resistance. Although a proportion of unnecessary treatments may be acceptable to ensure that true positives are treated, there is not agreement about what that proportion should be. The benefits of capturing all true positives must be weighed against drawbacks of overtreating those without disease.
Another solution is increased use of point-of-care testing for NG and CT.8 There are a number of tests available and Food and Drug Administration approved for detecting chlamydia in the United States, and there has not been widespread adoption of their use. These tests are considered clinical laboratory improvement amendments moderate complexity, which is a barrier to use, and are generally less sensitive than traditional methods of testing. Although point-of-care tests are in development for gonorrhea in the United States, none are currently Food and Drug Administration approved.20 Mathematical models suggest that given difficulty with follow-up, even this lower detection rate may lead to increased treatment because results are in real time.21,22
In our institution, when we are unable to contact patients with STIs, their information is sent to the DPH who does further outreach. We do not have information about the number of patients who were treated this way and may have underestimated the number who do eventually receive treatment. It is less than ideal, even if a portion of patients are ultimately treated via DPH outreach. There is a significant lag time before treatment, potentially leading to increase disease spread and complication risk. These results are from a single institution and may not be generalizable to other EDs. This was a retrospective chart review and subject to the difficulties associated with this method. We tried to improve accuracy and limit inconsistencies inherent in chart reviews by following standard guidelines for this method. It is possible that certain information was recorded in the chart inaccurately. Many charts did not have specific historical or physical examination findings recorded and were considered negative, but this may not be an accurate assumption. We attempted to minimize bias resulting from missing data by combining associated variables into broader categories for analysis, but potential for misclassification remains.
Another important limitation to note is that this study addresses only patients who had an STI test sent by their provider as part of clinical workup. Previous studies screening patients presenting to the ED setting have found many unsuspected, clinically silent cases of NG and CT,23 and recently published work by Jenkins et al.18 found that there was no association between complaint and infection. Unsuspected cases were not captured by study methods and not included in this analysis.
In this ED with a dedicated callback nurse, eventual treatment of STIs at our institution occurred in 85% of patients who tested positive for NG or CT. Only 50% of cases of NG or CT met CDC-defined increased-risk criteria, so even presumptive treatment in this group of patients at time of ED visit may not have improved treatment success. Additional interventions are required to improve the accuracy of ED treatment of STIs.
1. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance 2013. Atlanta, GA: Department of Health and Human Services; 2014.
2. Niska R, Bhuyia F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Rep 2010; 26: 1–31.
3. Curtis KM, Hillis SD, Keike BA, et al. Visits to emergency departments for gynecologic disorders in the United States, 1992–1994. Obstet Gynecol 1998; 91: 1007–1012.
4. Brackbill RM, Sternberg MR, Fishbein M. Where do people go for treatment of sexually transmitted diseases? Fam Plann Perspect 1999; 31: 10–15.
5. Johnson R, et al. Screening tests to detect Chlamydia trachomatis
and Neisseria gonorrhoeae
infections. MMWR 2002; 51: 1–27.
7. Gilbert EH, Lowenstein SR, Koziol-McLain J, et al. Chart reviews in emergency medicine research: Where are the methods? Ann Emerg Med 1996; 27: 305–308.
8. Merchant RC. Commentary. Ann Emerg Med 2011; 58: 68.
9. Leider JP, Castrucci BC, Russo P, et al. Perceived impacts of health care reform on large urban health departments. J Public Health Manag Pract 2015; 21: S66–S75.
10. Petersen LA, Burstin HR, O'Neil AC, et al. Nonurgent emergency department visits: The effect of having a regular doctor. Med Care 1998; 36: 1249–1255.
11. Walls CA, Rhodes KV, Kennedy JJ. The emergency department as usual source of medical care: estimates from the 1998 National Health Interview Survey. Acad Emerg Med 2002; 9: 1140–1145.
12. Schanzer BM, Morgan JA. Indigent men's use of emergency departments over primary care settings. Am J Public Health 2004; 94: 906–907 author reply 907–8.
13. Schanzer B, Dominguez B, Shrout PE, et al. Homelessness, health status, and health care use. Am J Public Health 2007; 97: 464–469.
14. Yealy DM, Greene TJ, Hobbs GD. Underrecognition of cervical Neisseria gonorrhoeae
and Chlamydia trachomatis
infections in the emergency department. Acad Emerg Med 1997; 4: 962–967.
15. Bachmann LH, Pigott D, Desmond R, et al. Prevalence and factors associated with gonorrhea and chlamydial infection in at-risk females presenting to an urban emergency department. Sex Transm Dis 2003; 30: 335–339.
16. Pattishall AE, Rahman SY, Jain S, et al. Empiric treatment of sexually transmitted infections in a pediatric emergency department: Are we making the right decisions? Am J Emerg Med. doi:10.1016/j.ajem.2011.09.028.
17. Marrazzo JM, Celum CL, Hillis SD, et al. Performance and cost-effectiveness of selective screening criteria for Chlamydia trachomatis
infection in women: Implications for a national chlamydia control strategy. Sex Transm Dis 1997; 24: 131–141.
18. Jenkins WD, Kovach R, Wold BJ, et al. Using patient-provided information to refine sexually transmitted infection screening criteria among women presenting in the emergency department. Sex Transm Dis 2012; 39: 965–967.
19. Al-Tayyib AA, Miller WC, Rogers SM, et al. Health care access and follow-up of chlamydial and gonococcal infections identified in an emergency department. Sex Transm Dis 2008; 35: 583–587.
20. Huppert J, Hesse EB, Gaydos CAM. What is the point? How point-of-care sexually transmitted infection tests can impact infected patients. J Near Patient Test 2010; 9: 36–46.
21. Gift TL, Pate MSM, Hook EWI 3rd, et al. The rapid test paradox: When fewer cases detected lead to more cases treated: A decision analysis of tests for Chlamydia trachomatis
. Sex Transm Dis 1999; 26: 232–240.
22. Vickerman P, Watts C, Alary M, et al. Sensitivity requirements for the point of care diagnosis of Chlamydia trachomatis
and Neisseria gonorrhoeae
in women. Sex Transm Infect 2003; 79: 363–367.
23. Mehta SD, Rothman RE, Kelen GD, et al. Unsuspected gonorrhea and chlamydia in patients of an urban adult emergency department: A critical population for STD control intervention. Sex Transm Dis 2001; 28: 33–39.