Adolescents who use the emergency department are more likely to engage in high-risk sexual activity and are at an increased risk of sexually transmitted infections. We aimed to increase testing for Chlamydia and gonorrhea from 12% to 50% among adolescents presenting to our pediatric emergency department with at-risk chief complaints over 12 months.
Plan-Do-Study-Act cycles were initiated in July 2020. A multidisciplinary team reviewed preexisting data and developed interventions to increase Chlamydia and gonorrhea testing in teens with at-risk complaints, including genitourinary and behavioral health complaints, and females with abdominal pain. Two categories of interventions were implemented: education and electronic medical record optimization. Process measures were the proportion of patients with a documented sexual history and the proportion of patients tested with a documented confidential phone number. Secondary outcome measures included the weekly number of positive test results and the proportion of patients testing positive who were contacted to arrange treatment. Statistical process control charts were used to examine changes in measures over time.
Within 14 months of project initiation, the proportion of at-risk patients tested increased from 12% to 59%. Teen phone number documentation remained unchanged from 23%. Sexual history documentation remained unchanged from 46%. The number of positive test results increased from 1.8 to 3.4 per month, and the proportion of patients testing positive who were contacted to arrange treatment remained unchanged at 83%.
We surpassed our goal and increased the proportion of at-risk patients tested for Chlamydia and gonorrhea to 59%, sustained for 4 months from the last intervention.
In the United States, adolescents and young adults account for approximately one-half of the 20 million new sexually transmitted infections (STIs) each year. An estimated 1 in 4 sexually active adolescents currently have an STI, most commonly Chlamydia, gonorrhea, and human papillomavirus.1 Adolescents are more likely to engage in high-risk sexual behavior, including unprotected sex and having multiple partners.1,2 It is important to promptly identify and treat STIs to prevent acute complications such as epididymitis and pelvic inflammatory disease, as well as long-term complications such as chronic pelvic pain, increased risk of ectopic pregnancy, and infertility.3 Moreover, these infections can be vertically transmitted and impact the newborn population.
The Centers for Disease Control and Prevention and the American Academy of Pediatrics recommend screening all sexually active adolescent females for STIs annually and sexually active males at increased risk (men who have sex with men, high-risk sexual activity, or community prevalence).3 Screening for Chlamydia in sexually active women aged 16 to 24 is a Healthcare Effectiveness Data and Information Set measure, a tool used by the majority of US health plans to measure performance.4 However, many adolescents do not receive this screening because of various barriers to care, including concerns about confidentiality, scheduling, transportation, and cost.5 According to the 2019 National Youth Risk Behavior Survey, only 20% of sexually active high school students reported testing for an STI in the previous year.6 Provider-specific barriers to offering adequate reproductive health screening and counseling include difficulty building rapport during sparse encounters, a lack of education around minor consent laws and confidentiality, language, cultural, or religious differences, discomfort with behavioral counseling, and insufficient training on reproductive health.7–9 There are racial disparities in STI testing, with studies revealing more testing in Black adolescents compared with other groups in primary care clinics and pediatric emergency department (ED) settings.10,11 However, the 2019 National Youth Risk Behavior Survey did not reveal racial or ethnic differences in the proportion of students reporting sexual activity.6
Many adolescents do not have a primary care provider but do seek care in the ED. Nordin et al performed a retrospective analysis of primary care visits and found that one-third of adolescents did not attend a single primary care visit between the ages of 13 and 17 years.12 Access to care has unfortunately declined with the ongoing coronavirus disease 2019 (COVID-19) pandemic, due to closures and limited operational capacity of many community clinics and patients fearing exposure to illness.13 Adolescents who use the ED as their usual source of care are less likely to have regular well visits and more likely to engage in risky behaviors.2 Therefore, the ED is an important clinical setting to address the sexual health of adolescents.
We identified that our ED was rarely performing STI testing on adolescents, and we sought to address this issue with a quality improvement (QI) initiative. Our global aim was to improve reproductive health care for adolescents in the ED. Our specific aim with this study was to increase Chlamydia and gonorrhea testing from 12% to 50% in adolescents presenting to our ED with at-risk chief complaints over 12 months. Our secondary aims were to increase sexual history documentation in at-risk patients from 46% to 75% and to increase the documentation of confidential teen phone numbers in patients tested from 23% to 75%.
Methods
Setting
Rady Children’s Hospital San Diego serves as the region’s tertiary children’s hospital. It is an academic medical center affiliated with the University of California, San Diego. The 63-bed free-standing pediatric ED has ∼100 000 annual visits, 15% of which are adolescents. In 2019, the race and ethnicity distribution of patients discharged from our ED was 53% Hispanic, 31% white non-Hispanic (NH), 7% black NH, 5% Asian NH, and 4% other (including Pacific Islander, American Indian/Alaskan Native, and other). In 2019, the insurance payer distribution for patients seen in our ED was 63% government, 27% private, 7% military, and 3% self-pay. The ED is staffed by pediatric emergency medicine (PEM) physicians. Trainees include PEM fellows, residents from pediatrics, emergency medicine, and family medicine programs, and medical students. This project was reviewed by the University of California, San Diego Institutional Review Board and determined not to be human subjects research.
Planning
To identify barriers to STI testing and plan our interventions, we assembled a multidisciplinary team, including 2 PEM physicians, 1 PEM nurse, 2 adolescent medicine physicians, 1 physician informaticist, 1 pediatrics resident, and 2 medical students. The team reviewed 6 months of baseline data from January 2020 to June 2020. We found the following chief complaints most frequently associated with a positive Chlamydia or gonorrhea test result: genitourinary complaints, behavioral health complaints, and abdominal pain in females. When we reviewed adolescents presenting to the ED with these chief complaints, we found a testing rate of only 12% and sought to increase this to 50%. We decided not to focus our efforts on only sexually active patients because of the low rate of sexual history documentation. Our team met and created a Failure Mode Effects Analysis (Fig 1) to analyze our current STI testing process and a key driver diagram (Fig 2) to plan our interventions.
Interventions
There were 2 categories of interventions implemented over the course of 18 months using Plan-Do-Study-Act cycles: education and electronic medical record (EMR) optimization.
Education
Project implementation began in July 2020. First, the project was introduced to our ED Quality Improvement Committee, a small group of PEM physicians and nurses who review our ED QI initiatives and provide feedback. The project was presented at the Division of Emergency Medicine monthly meeting, as well as the ED Nursing Staff monthly meeting. These educational sessions included a literature overview, summaries of current guidelines for STI testing and treatment, baseline data review, and planned interventions. We provided project updates at our division meetings every 3 months. Project reminders were also periodically disseminated by a division e-mail newsletter. We created a poster including recommendations for testing and EMR resources that was posted in the ED physician workrooms.
During these educational sessions, staff had the opportunity to ask questions and provide feedback on interventions. For example, nursing staff raised questions about the feasibility of dirty urine or vaginal specimen testing protocols. This feedback led to the implementation of a single (clean or dirty) urine specimen collection. We used a urine polymerase chain reaction test rather than a vaginal swab to increase our testing rates because urine samples are often collected as part of the medical evaluation. Mangin et al found that the timing of specimen collection is not relevant, and midstream sample sensitivity rivals that of first-catch samples.14 In a study comparing vaginal swab, clean urine, and dirty urine testing, the specificity of clean urine was noninferior in diagnosing Chlamydia and gonorrhea. The sensitivity of clean urine to diagnose Chlamydia and gonorrhea was 86% and 94%, respectively.15 Patients undergoing evaluation for urinary tract infection (ie, those with dysuria, abdominal pain) collected clean urine, whereas behavioral health patients without medical complaints collected dirty urine.
Nurses and physicians discussed time constraints as a barrier to collecting teen contact information, which led us to meet with registration staff to standardize this into their workflow. Registration staff received verbal and written training on obtaining adolescent contact information when gathering parents’ contact information. They were instructed to explain to parents that under California law, certain tests and treatments are confidential for patients 12 and older and that it is hospital policy to collect teens’ contact information.
For one of our educational interventions, we created a funnel plot to analyze individual physician data and offered targeted feedback to physicians with low STI testing rates and sexual history documentation (Supplemental Figs 6 and 7). The funnel plot was e-mailed to the group with anonymous identifiers so that physicians could see how they compared with others in regard to testing and documentation.
EMR Optimization
To facilitate testing and treatment, we created a new EMR order set that included STI testing, antibiotic treatment of STIs and pelvic inflammatory disease (inpatient and outpatient), and post sexual assault prophylaxis. We added Chlamydia and gonorrhea testing to existing order sets and panels for patients aged >12 years, including testicular ultrasound orders for testicular pain and pelvic ultrasound orders for females with abdominal pain. We also implemented nurse-initiated orders for urine Chlamydia and gonorrhea testing for patients with testicular pain or behavioral health complaints before evaluation by an ED physician. Finally, we preselected the orders within these physician and nursing order sets. For the nursing order sets, the physician had the option to cancel the test if desired or order it if the nurse had not done so.
To improve sexual history documentation, we added “disappearing help text” as a reminder to document the confidential sexual history within the physician note for all patients >12 years of age.16 The disappearing help text was embedded with a hyperlink to the confidential history section of the patient’s chart. To improve teen phone number documentation, we provided training for physicians and nurses on the demographics section of the EMR.
Measures
The primary outcome measure was the proportion of at-risk patients aged 13 to 17 years with urine Chlamydia and gonorrhea testing completed weekly. At-risk patients included patients with behavioral health and genitourinary complaints and female patients with abdominal pain. Patients were included in a structured query language-based report on the basis of age and chief complaint. Process measures included the proportion of at-risk patients with a documented sexual history and the proportion of patients undergoing testing who had a documented confidential phone number. We identified sexual history-taking as an area for improvement, and we used teen phone number documentation as a metric to improve treatment rates. Secondary outcome measures included the weekly number of positive test results and the proportion of patients with positive test results contacted to arrange treatment. Balancing measures included the proportion of at-risk patients with negative test results who were empirically treated for infection and parent complaints about STI testing monitored through our hospital’s patient and family complaint system. We also collected demographic data, including age, sex, race, and ethnicity.
Analysis
We used Microsoft Excel and QI Macros to create statistical process control charts to examine changes in measures over time, and we used established rules for the interpretation of control charts.17 We graphed testing rates weekly because of the large denominator, and we graphed teen phone number documentation rates monthly because of variability over time. The centerline and control limits were revised when an intervention led to a shift, defined as 8 consecutive points greater or less than the mean. Special cause variations in the form of a single data point outside of the control limits were investigated for possible causes to guide improvement. Race and ethnicity distributions by pre- and postintervention periods were compared by using χ square (Stata 16.1, College Station, TX). We included race and ethnicity in our analysis to examine whether our QI interventions would reduce or exacerbate known disparities in testing.
Results
There was a total of 5539 eligible patients aged 13 to 17 years during our 2-year study period (January 2020 to December 2021), with a mean of 52 eligible patients per week. Over the 18 months from the start of the initiative, we achieved 3 statistical process control chart shifts associated with our interventions and increased the testing of at-risk patients from 12% to 59% (Fig 3). Our most significant shift from 18% to 44% occurred after the implementation of preselected order panels for physicians and nurses. There was 1 final shift after this to 59%, which was sustained for the next 4 months. Testing in eligible medical patients (genitourinary complaints and females with abdominal pain) increased from 16% to 46% (Supplemental Fig 8). Testing in eligible behavioral health patients increased from 6% to 70% (Supplemental Fig 9).
Control p chart of proportion of at-risk patients tested for Chlamydia or gonorrhea.
Control p chart of proportion of at-risk patients tested for Chlamydia or gonorrhea.
During the first 18 months after the start of the initiative, there was significant variation in teen phone number documentation for tested patients. There was no shift from the baseline at 23%; however, the last 7 data points are all above the center line, approaching a shift (Fig 4). There were 2 time periods with special cause variation: in October 2020, after physician and nursing education, and September 2021 to December 2021, after the training of registration staff. Sexual history documentation did not improve during the study period and remained at a baseline of 46%. Of those with sexual history documentation, 871 of 2574 (33.8%) reported being sexually active. Sexual history documentation was associated with higher rates of testing (P < .001).
Positive test results per month increased over the course of the initiative from 1.8 to 3.4 (Fig 5). There were 65 of 1462 (4.4%) test results that were positive during the study period, with 56 Chlamydia, 19 gonorrhea, and 10 coinfections. The most common chief complaints associated with a positive test result included suicidal ideation, abdominal pain or vomiting, and dysuria. Interestingly, of the patients testing positive, 5 (7.7%) denied sexual activity. Approximately one-third (23/65) of patients testing positive had behavioral chief complaints. The majority (58/65, 89%) of patients testing positive were contacted to arrange treatment. Only 34 of 1395 (2.4%) patients with a negative test result were empirically treated during their ED visit. There were no parent complaints about testing or billing during the study period. These were monitored by our ED physician representative, who provides input on all parent complaints.
The demographics of the study population are described in Tables 1 and 2, with comparisons of the tested patients and all eligible patients during the baseline and intervention periods. During the baseline period, tested patients were significantly older than the eligible population. During the intervention period, there was a slight increase in age, a decrease in the proportion of males, and racial and ethnic differences in tested patients compared with the eligible population. During the intervention period, there was an increase in testing for white, Black, and Asian NH patients and a decrease in testing for Hispanic and Other patients compared with the eligible population.
Demographics of Eligible and Tested Patients: Baseline Period
| Patient Characteristics | Baseline Period (January 1, 2020–June 30, 2020) | P | |
|---|---|---|---|
| Eligible Patients | Tested Patients | ||
| (n = 1223) | (n = 137) | ||
| Age (mean +/− SD) | 15.3 +/− 1.6 | 15.9 +/− 1.5 | <.0001 |
| Sex (% male) | 22.8 | 18.2 | .16 |
| Race/ethnicity (%) | .32 | ||
| Hispanic | 47.9 | 54.0 | |
| White NH | 32.3 | 24.1 | |
| Black NH | 6.7 | 7.3 | |
| Asian NH | 4.5 | 3.6 | |
| Other | 8.6 | 11.0 | |
| Patient Characteristics | Baseline Period (January 1, 2020–June 30, 2020) | P | |
|---|---|---|---|
| Eligible Patients | Tested Patients | ||
| (n = 1223) | (n = 137) | ||
| Age (mean +/− SD) | 15.3 +/− 1.6 | 15.9 +/− 1.5 | <.0001 |
| Sex (% male) | 22.8 | 18.2 | .16 |
| Race/ethnicity (%) | .32 | ||
| Hispanic | 47.9 | 54.0 | |
| White NH | 32.3 | 24.1 | |
| Black NH | 6.7 | 7.3 | |
| Asian NH | 4.5 | 3.6 | |
| Other | 8.6 | 11.0 | |
Demographics of Eligible and Tested Patients: Intervention Period
| Patient Characteristics | Intervention Period (July 1, 2020–December 31, 2021) | P | |
|---|---|---|---|
| Eligible Patients | Tested Patients | ||
| (n = 4306) | (n = 1321) | ||
| Age (mean +/− SD) | 15.3 +/− 1.6 | 15.4 +/− 1.6 | .047 |
| Sex (% male) | 17.1 | 13.9 | .007 |
| Race/ethnicity (%) | .001 | ||
| Hispanic | 46.5 | 45.2 | |
| White NH | 33.0 | 36.5 | |
| Black NH | 5.6 | 7.0 | |
| Asian NH | 3.3 | 5.5 | |
| Other | 11.6 | 5.8 | |
| Patient Characteristics | Intervention Period (July 1, 2020–December 31, 2021) | P | |
|---|---|---|---|
| Eligible Patients | Tested Patients | ||
| (n = 4306) | (n = 1321) | ||
| Age (mean +/− SD) | 15.3 +/− 1.6 | 15.4 +/− 1.6 | .047 |
| Sex (% male) | 17.1 | 13.9 | .007 |
| Race/ethnicity (%) | .001 | ||
| Hispanic | 46.5 | 45.2 | |
| White NH | 33.0 | 36.5 | |
| Black NH | 5.6 | 7.0 | |
| Asian NH | 3.3 | 5.5 | |
| Other | 11.6 | 5.8 | |
Discussion
Through our QI initiative, we successfully increased Chlamydia and gonorrhea testing in at-risk patients presenting to the ED. Our most effective intervention was the use of preselected orders in the EMR. We initially started our interventions with physician and nursing education. However, education alone did not lead to increased testing rates. Our study team met to discuss additional barriers to testing and explore how to make testing as easy as possible. We found that the provider-perceived burden of remembering to order testing or taking the time to do so was a potential contributor. We created nurse-initiated orders for testing in behavioral health patients and added testing orders to existing order panels for testicular and pelvic ultrasounds. This did not lead to a significant improvement until we preselected the orders, which shifted the testing rate from 18% to 44%, then 59%.
Our study revealed that preselected orders can be a more effective QI intervention than provider education to change decision-making. Preselected ordering is an example of a clinical nudge, “an action that changes the choice architecture of a decision and influences provider behavior without restricting options.”18 Studies reveal that preselected orders are an effective way to improve desired outcomes; for example, preselected orders have been used effectively to decrease opioid prescribing.19 Our providers were receptive to the preselected orders and found that they improved their workflow. Many were appreciative that they were able to diagnose an STI in patients they otherwise may not have tested or had the time to obtain their sexual history. To our knowledge, this is the first study to implement preselected orders to increase STI testing in the ED.
Patients were typically notified of testing by their physician or nurse and had the option to opt out. In the 2021 STI Treatment Guidelines, the Centers for Disease Control and Prevention proposed that providers consider opt-out screening for Chlamydia and gonorrhea in adolescent females, regardless of reported sexual activity.20 Because our study revealed low rates of sexual history documentation and several patients with STIs denying sexual activity, relying on only screening patients known to be sexually active for STIs may lead to missed cases. Cost-effectiveness analyses suggest that opt-out screening among adolescent females may be cost-saving and identify infections among those who do not disclose sexual activity.21
Sexual history documentation did not improve over the duration of the study and remained at 46%, despite EMR reminders and education. After reviewing individual physician data, we noted that several physicians consistently obtained sexual history, whereas others did not. This is likely multifactorial and related to the physician’s training, comfort level, EMR and work efficiency, and, perhaps, cultural factors. After a discussion with the physicians, we also found that many perceived that there was inadequate time to conduct sexual history, especially when the ED was busy. We considered more active EMR interventions for sexual history and teen phone number documentation, such as required documentation to sign the orders, but did not pursue this because of challenges in streamlining the EMR workflow for providers. Individual physician feedback on performance did not lead to improvements in testing or documentation, so we did not continue this intervention. Previous studies have also revealed poor documentation of sexual history in pediatric EDs.7,22 Because we found that sexual history documentation was associated with increased STI testing (P < .001), further work is necessary to explore ways to improve this measure.
Confidential teen phone number documentation occurred in 23% of tested patients before the study interventions and improved after the education of providers and registration staff. Physicians and nurses were given reminders to document the teen’s phone number but reported time and task recall as barriers. When we implemented teen phone number documentation into the registration workflow, initially, some registration staff did not feel comfortable asking teens for their contact information. After continued training, there was improvement and special cause variation for 4 consecutive months. Previous studies have revealed high rates of loss to follow-up of 20% to 60% for patients with STIs.21,23,24 We had a much lower rate of loss to follow-up of 9%, likely due to the improvement of teen phone number documentation, as well as some of the patients testing positive being admitted to inpatient psychiatric facilities that could be contacted to arrange treatment.
One-third of the patients diagnosed with Chlamydia or gonorrhea during the study period were behavioral health patients. Other studies have revealed that individuals with psychiatric disorders are at an increased risk of STIs.25 These patients have less access to primary care and health screening and increased rates of high-risk sexual practices, impulsivity, cognitive immaturity, substance abuse, and self-harm.26 Some are at an increased risk of STIs because of sequelae of sexual abuse.25 We had a large increase (>25%) in our ED volume of behavioral health patients after the onset of the COVID-19 pandemic, when we started this project, which led to a significant increase in sexual health screening in this high-risk population.
Our highest testing rates were in Hispanic, followed by white NH patients, consistent with the demographics of our patient population (Tables 1 and 2). Although there were statistically significant differences in the demographics of the patients tested compared with the eligible population during the intervention period, these statistical differences are likely reflective of the large patient volumes in the statistical analysis. Our data revealed small increases in testing for white, Black, and Asian NH populations, a small decrease in testing in Hispanic patients, and a decrease in testing in other patients. Previous studies have revealed disparities in testing with higher testing in Black and Hispanic populations. Overall, the race and ethnicity distribution of our tested patients better mirrored the distribution of the eligible population during the intervention period when compared with the baseline period. However, it is important to emphasize that QI methodology may not overcome the well-documented disparities in numerous quality metrics.27 We did not design our QI interventions to specifically target disparities in testing, and further work should be done in this area in the future.
There were several limitations to our study. Although the baseline period occurred from January 2020 to June 2020, an additional baseline variation in health care utilization by season or stage of training of providers may not have been captured. The baseline period coincided with the onset of the COVID-19 pandemic when the ED volume was low; this may not have captured the typical ED adolescent population. We did not examine the insurance status of included patients, which may have impacted physician decision-making regarding testing. A small number of patients (1% preinterventions, and 1.5% after the start of the interventions) did not provide their race and ethnicity, and some listed “other” under these categories. We did not measure testing by race and ethnicity over time.
We did not screen all adolescents for STIs and only targeted patients with specific complaints because we felt this would be more feasible to implement, more acceptable to staff, and more cost-effective than universal screening. Moreover, screening only patients with documented sexual activity would not be feasible because of the low rates of sexual history documentation and occasional unreliable history. There were instances in which the physician or nurse ordered the test, but it was never collected, or the laboratory did not run the test correctly (25/1458, 1.7% of tests ordered). Although sexual history documentation did not improve, some physicians may have asked about sexual history but did not document it if it was negative or not relevant to the visit. Behavioral health patients diagnosed with STIs may have had genitourinary or other complaints, but they were not offered, elicited, or documented in the history during the ED visit.
Conclusions
We significantly increased Chlamydia and gonorrhea testing in adolescents presenting to the ED with at-risk complaints from 12% to 59% within 14 months. We found that the optimization of the EMR was more effective than education and feedback in enacting change. Further work is needed to determine how best to improve sexual history and teen phone number documentation and target disparities in STI testing. We hope to sustain these improvements in our ED and spread our interventions to other areas of the hospital and community.
Acknowledgments
Thank you to the members of the RCHSD ED QAPI committee for their valuable recommendations and support with this initiative. Thank you to Dr Kanegaye for assisting with data analysis. Thank you to Drs Seema Shah and Laurel Moyer for their mentorship and teaching in QI methodology.
Dr Vayngortin conceptualized the study, designed QI interventions, reviewed and analyzed data, supervised data collection, and drafted and critically revised the manuscript; Dr Gracia designed QI interventions, reviewed and analyzed data, drafted the initial manuscript, and reviewed the manuscript; Ms Clark and Ms Vayngortin reviewed and analyzed data and reviewed the manuscript; Dr Bialostozky designed data collection and analysis and critically revised the manuscript; Ms Heitzman designed QI interventions; Dr Hollenbach designed and supervised data collection and analysis and critically revised the manuscript; Drs Titchen, Kumar, and Bryl conceptualized the study, designed QI interventions, and critically revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no potential conflicts of interest relevant to this article to disclose.
Comments