Effectiveness of a Quality Improvement Intervention to Improve Rates of Routine Chlamydia Trachomatis Screening in Adolescent Females Seeking Primary Preventive Care

Abstract

Study Objectives:

To determine the impact of a multi-component quality improvement (QI) intervention on Chlamydia trachomatis screening for young women in primary care.

Design:

Observational cohort analysis.

Setting:

Urban primary care site providing adolescent primary and confidential sexual health care.

Participants:

Female adolescents aged 15–19 years.

Interventions:

From December 2018 to April 2018, we designed and implemented a multiphase QI intervention. The final intervention, beginning March 2017, consisted of the following at all adolescent well visits: 1) dual registration for well and confidential sexual health encounters; 2) urine collection during the rooming process; and 3) electronic health record-based prompts for chlamydia screening.

Main Outcome Measures:

Annual chlamydia screening rates before and after the intervention, with a goal of achieving a relative increase of 10%.

Results:

There were 1,550 well adolescent encounters from December 2016 to April 2018. The pre-implementation chlamydia screening rate among 15- to 19-year-old female adolescents was 41.2% (95% confidence interval: 20.9–61.5%). Post-intervention, this increased to 50.0% (95% confidence interval: 28.6–71.5%) (p<0.001). The clinic chlamydia test positivity rate remained stable, at 10.7% and 11.1% in the pre and post-intervention periods, respectively. There was no significant change in median visit length in the pre- (79.2 minutes, interquartile range 59.5–103.3) and post-intervention periods (80.4 minutes, IQR 61.7–102.8) (p=0.63).

Conclusions:

This practice-based QI intervention resulted in a statistically significant 21% relative increase in annual Chlamydia trachomatis screening rates among adolescent females, without lengthening median visit time.

INTRODUCTION

Chlamydia trachomatis (CT) is the most common reportable sexually transmitted infection (STI) in the U.S., with approximately 1.7 million cases reported in 2017.¹ It is a leading cause of pelvic inflammatory disease (PID)², which might lead to tubal-factor infertility, ectopic pregnancy, and chronic pelvic pain.³ Untreated chlamydial infection might also lead to increased biologic risk of HIV acquisition by increasing genital tract inflammation and altering the circulating immune milieu.⁴ Adolescent and young adult women have the highest chlamydia prevalence in the U.S.–approximately 1 in 20 among sexually active, young women aged 14–24 years.⁵

Despite significant health risks, chlamydia infection is often asymptomatic, leading to delays in diagnosis in the absence of routine screening. Previous studies have shown that asymptomatic, routine screening might lead to reductions in the prevalence of PID.^6,7 Therefore, the Centers for Disease Control and Prevention, the U.S. Preventative Services Task Force, American Academy of Pediatrics, and American College of Obstetrics and Gynecology have issued guidelines recommending routine, annual screening of all sexually active women younger than 25 years of age.^8,9,10

Unfortunately, routine chlamydia screening rates remain low in pediatric and adolescent primary care.¹¹ Data from the National Survey of Family Growth show that only 40% of sexually active U.S. women aged 15–21 years had been screened for chlamydia in the preceding year.¹² Compared to other providers, pediatricians have the lowest rates of screening. The National Ambulatory Medical Care Survey identified that only 0.9% of pediatrician visits for 15–21 year old women had chlamydia testing performed.¹² In a review of primary care records of one large pediatric hospital system, only 21% of routine adolescent well visits had a documented sexual history and only 2.6% were tested for CT within the preceding year.² Commonly identified barriers to screening are lack of clinician knowledge and self-efficacy, confidentiality concerns, and patient insurance.¹³ These findings demonstrates a critical need for efforts to improve adherence to chlamydia screening guidelines in pediatric and adolescent primary care.

Previous quality improvement (QI) interventions aiming to improve chlamydia screening have demonstrated mixed efficacy for interventions focusing on provider education and increasing access to emergency department or expedited outpatient sexual health services.^14–18 Universal procurement of urine specimens prior to the provider visit has been demonstrated to increase screening rates across several studies.^19–21 However, few QI interventions have been specifically linked to the pediatric primary care context, particularly in high-risk chlamydia prevalence areas. In addition, while there is some evidence that QI interventions may improve chlamydia case detection,²² the impact of improving adherence to screening guidelines on chlamydia prevalence rates remains unknown.

The primary goal of our intervention was to increase Chlamydia trachomatis screening rates for young women receiving well-adolescent care in a high chlamydia prevalence region. Secondary goals included identifying the impact of the intervention on chlamydia test positivity rates, and identifying the impact of the intervention on visit length.

METHODS

Setting:

The intervention occurred at an urban, pediatric and adolescent primary care clinic located in West Philadelphia. The clinic is affiliated with a nearby academic tertiary care hospital within a robust primary care network. The clinic has approximately 30,000 visits per year, and 50% of well-visit attendees are adolescents. Eighty percent of patients have Medicaid insurance, and the patient mix is overwhelmingly African-American. Clinical services are provided by a multi-disciplinary team consisting of medical assistants, nurses, certified registered nurse practitioners (CRNPs) and physicians. At project initiation, the clinic had twelve attending physicians and two CRNPs.

In addition to general pediatric and adolescent care, the clinic provides confidential adolescent sexual health services through a Title X-grant funded family planning (FP) program. Adolescents can register for general medical services and/or FP services when scheduling or at visit registration. In FP encounters, the electronic health record (EHR) contains additional privacy protections and no bills are sent to the home.

Participants:

The intervention target population was 15- to 19-year-old female adolescents attending annual primary care visits during the study period of January 2016 to April 2018. The pre-intervention period spanned from January 2016 to February 2017. The post-intervention period was from March 2017 to April 2018.

All clinic multi-disciplinary staff, including nurses, registrars and medical assistants were involved in the initial intervention design meetings and implementation process. The core QI implementation team included the clinic Medical Director (AM), one Adolescent Medicine physician from the clinical site (MW), and an Attending Physician (KC) and third-year Fellow (SMW) from the primary institution’s Division of Adolescent Medicine. Consultative support was provided through the institution’s Office of Clinical Quality Improvement. All clinic providers and staff were eligible to participate in the intervention and participating physicians could receive Part 4 Maintenance of Certification (MOC) credit from the American Board of Pediatrics.

In accordance with institutional standards, this study was considered QI rather than research, and was deemed exempt from Institutional Review Board oversight.

Intervention Development:

We utilized an existing institutional improvement framework (Supplementary Figure) derived from Lean Six Sigma methodology²³ including the phases define, diagnose, test, implement and sustain to develop a multicomponent QI intervention to improve chlamydia screening among 15- to 19-year-old female adolescents. A pilot intervention at a separate West Philadelphia clinic had demonstrated short-term gains in screening rates, and served as the model for this intervention.²⁴ Using the SMART criteria for intervention goal-setting²⁵, we developed an intervention objective that was Specific, Measurable, Achievable, Relevant and Time-bound: to achieve a relative 10% increase in the proportion of 15- to 19-year-old women screened for chlamydia infection at or within one year prior to their annual well-visit by 10%. The relative increase of 10% was chosen based on feasibility data from a partner clinic demonstrating a similar relative increase in screening after a pilot QI intervention.²⁴ We used all female adolescents receiving well care as the denominator, rather than those documented history of sexual activity, as sexual histories are frequently incomplete or missing in the pediatric EHR, and strategies for determining sexual activity from additional features of the EHR may still lead to underestimation of sexual activity.²⁶ We used ecologic data on sexual activity from the Youth Risk Behavior Surveillance in the Philadelphia area, demonstrating a 57% rate of sexual activity of African American high-school females²⁷, to identify a ceiling threshold testing rate after which increases in screening in adolescent females may no longer be efficacious.

Phase 1-Baseline Assessment:

In the first phase of intervention development (January 2016), confidential surveys were administered to site clinicians to identify gaps in knowledge and self-efficacy around chlamydia screening. Questions included knowledge of screening guidelines and regional chlamydia prevalence, and assessed self-efficacy in confidential sexual history-tasking and discussing positive results. Answers were provided on a 5-point Likert scale ranging from “strongly agree” to “strongly disagree” (Figure 1).

Figure 1.

Preintervention Q16 provider questionnaire. AAP, American Academy of Pediatrics; STI, sexually transmitted infection.

Phase II-Staff Education:

In the second phase, the team held a multi-disciplinary, staff education session to improve knowledge and cultivate a culture of change among staff, including clinicians, nurses, medical assistants, and patient services representatives. The session content included screening guidelines, local prevalence, complications associated with chlamydia infection, current clinic screening rates, and potential targets for improving screening rates.

Phase III-Process Mapping:

In the third phase, with support from one of the hospital’s Quality Improvement advisors, the multi-disciplinary staff completed a process mapping activity to detail the clinic’s pre-intervention chlamydia screening practice and identify areas for intervention. The pre-intervention screening process and targeted areas for improvement (in dark grey) are detailed in Figure 2. Prior to the intervention, urine collection for chlamydia nucleic acid amplification testing (NAAT) occurred after confidential sexual history-taking.

Figure 2.

Pre-Intervention Chlamydia Screening Process.

Phase IV-Intervention Model Optimization:

In the final phase, the team developed the model for intervention. A protocol was developed for universal urine collection for adolescent patients at the time of vital signs and prior to provider interview. Although the intervention target was 15–19 year-old females, the urine collection protocol was designed for all adolescents to facilitate fidelity. The final intervention, deployed in March, 2017, consisted of the following components: 1) Dual-registration of all adolescents for well and FP encounters in order to streamline confidential STI test ordering by advanced practice providers and nurses; 2) medical assistants providing labeled specimen cups to all adolescents prior to history and physical exam; 3) clinic room signage informing families that urine would be obtained on all adolescents routinely; 4) EHR-based prompts for ordering of chlamydia testing based on confidential sexual history taking (Figure 3).

Figure 3.

Intervention Chlamydia Screening Process

Study of the Intervention:

To capture intervention effect on a near real-time basis, we developed a clinical application utilizing a commercial business intelligence platform (Qlik, Radnor, PA) which continually captured chlamydia test ordering from the data warehouse that captures our EHR data (Epic Systems, Verona, WI). The application displayed STI test ordering in electronic run charts which were made available to the QI team throughout the process. Data were reviewed weekly by the Project Leader, and at bimonthly meetings of the study team. I these meetings, data were reviewed to make iterative revisions to the intervention using the Plan-Do-Study-Act (PDSA) cycle process. Run charts displaying current screening rates were sent to the entire clinic team on a monthly to bimonthly basis to encourage sustained behavior change.

Measures:

Our primary outcome metric was the proportion of 15- to 19-year-old female adolescents attending annual well-visits who received chlamydia testing at or within one year prior to their well visit, with a goal of achieving a relative increase of 10%. Well visit encounters were identified using International Classification of Disease (ICD) codes corresponding to annual pediatric preventative visits. Receipt of chlamydia testing was assessed via ordered and completed testing within the 365 calendar days prior to all well visits in the observation period. Our baseline measurement period was January, 2016 through February, 2017 (pre-intervention). The post-intervention measurement period was March, 2017 through April, 2018 in to capture the a priori defined 12-month period reach goal and an additional two months of follow up to assess continued fidelity to the intervention.

Secondary outcome measures included the chlamydia test positivity rates post-intervention. As some patients received multiple chlamydia screenings in a year and/or had multiple episodes of chlamydia, this was calculated on the case, rather than the individual level (e.g. cases identified/screening assays sent), with a different denominator than detailed in the screening outcome metric. Balancing measures included staff participation in the QI process; registration time, defined as time from patient check-in to first vital signs; and visit length, defined as the time of patient registration to the time to printing of the visit summary.

Statistical Analysis:

Results of the baseline provider survey and demographic characteristics of clinic patients were summarized via descriptive statistics. The time series analysis of pre and post-intervention screening rates were assessed via the p-chart, a type of statistical process control chart using the binomial distribution, with the following criterion used to determine positive special cause variation due to changes in the process: ≥8 values above the baseline mean.²⁸ Additionally, the equality of the pre and post intervention screening rates were assessed using the Pearson χ² test. As balancing measures, we compared the time from visit registration to visit vital signs, as well as total visit length for the pre- and post-intervention periods using Wilcoxon rank sum testing as the data were non-normally distributed. Statistical significance was set at p<0.05.

RESULTS:

Patient Characteristics:

Between January 2016 and April 2018, there were n= 1,550 well-visit encounters among 1,190 unique female adolescents. The overall descriptive characteristics of the clinic sample are displayed in Table 1. The patient population during the observation period was overwhelmingly African-American, with majority of patients having Medicaid insurance.

Table 1:

Characteristics of Patient Encounters (n=1,550)

Characteristic	Number (%)/Mean (SD)
Patient Age	16.6 (1.29)
Patient Race
African American	1,472 (95.0%)
Caucasian	14 (0.9%)
Asian	8 (0.5%)
Other	56 (3.6%)
Hispanic Ethnicity	36 (2.3%)
Provider Type at Visit
Adolescent Medicine Attending	353 (22.8%)
General Pediatrics Attending	564 (36.4%)
General Pediatrics Resident	207 (13.4%)
CRNP	423 (27.3%)
Nurse only	3 (0.2%)
Insurance Coverage
Government	933 (64.1%)
Commercial	457 (29.5%)
None	100 (6.5%)

Provider knowledge and participation:

Prior to the intervention, 12/14 (85%) of providers completed the screening knowledge survey. Overall, median scores fell in the “strongly agree” category for questions regarding local chlamydia prevalence, self-efficacy around confidential sexual history taking, EHR documentation, lab ordering, and delivery of results suggesting a high level of competency in these domains. The only item where the median score fell below the “strongly agree” category was the question assessing knowledge regarding chlamydia screening guidelines for adolescents, where the median score was in the “agree” category. With respect to clinician participation, 9/12 (75%) of attending physicians claimed MOC credit for participation. While the clinic CRNP’s (n=2) could not obtain MOC credit which is available to physicians only, both attended and participated in QI meetings, for a total provider participation rate of 11/14 (78%).

Intervention Evolution:

In QI core team meetings, we reviewed barriers and facilitators to implementation through the Plan-Do-Study-Act process in order to refine the intervention. For example, we identified that specimens were not consistently packaged for laboratory transport after ordering. This was ameliorated by creating a workflow wherein laboratory technologists ensured that nurses had appropriately packaged specimens for all patients for whom urine testing had been ordered. In addition, patients were not consistently being registered for both well and FP encounters. To facilitate rapid dual registration, a system was developed where charts were flagged within the EHR to alert patient service representatives that an FP encounter was needed.

Chlamydia Screening Rates:

For the pre-intervention period, the mean annual chlamydia screening rate among 15- to 19-year-old female adolescents presenting for well visits was 41.2% (95% confidence interval: 20.9–61.5), representing 312 screenings among 757 visits. Post-implementation, there was a statistically significant increase in screening rate to 50.0% (95% confidence interval: 28.6–71.5%), representing 397 screenings among 793 visits (p=0.001) (Figure 4). This represents a relative 21% increase from the baseline screening rate, exceeding the established goal of a 10% increase in screening of the baseline 41.2% rate. While rates overall improved, increased variability in screening rates was noted in thesummer and early fall, both before and after the intervention. In review by the QI leadership team, it was determined that this was likely due to both provider vacations and the influx of new general pediatric residents into the clinic, which may have increased competing time and training demands on the clinic’s attending physicians. Despite this variation, the statistical process control methodology demonstrated 12 values above the baseline mean in the post-intervention period, confirming that the change in chlamydia screening rates before and after the intervention is not due to “normal” variation but due to a deliberate or planned change. ²⁸

Figure 4.

Proportion of 15- to 19-year-old women receiving chlamydia testing. LCL, lower confidence limit; UCL, upper confidence limit.

Chlamydia Test Positivity:

The proportion of positive chlamydia NAATs in the pre-intervention period among 15- to 19-year-old female adolescents receiving chlamydia screening at the clinic was 10.7% (144 cases and 1,350 screenings). The chlamydia positivity rate after the intervention remained relatively stable, at 11.1% (146 cases and 1,320 screenings).

Registration Time and Visit Length:

Registration data was available for 1419/1550 (92%) of the patient visits. The median time from arrival at the registration desk to first vital signs was eight minutes (interquartile range 4–16) both before and after the intervention. Visit length data was available for 1503/1550 (97%) of visits. The pre-intervention period visit length was 79.2 minutes (interquartile range 59.5–103.3). The post-intervention visit length was 80.4 minutes (61.7–102.8). There was no statistically significant different in registration time (p=0.59) or visit length (p=0.63) before and after the intervention.

DISCUSSION

This practice-based QI intervention to improve chlamydia screening in adolescent young women resulted in a statistically significant 21% relative increase in routine chlamydia screening in a high chlamydia prevalence clinic. Our data demonstrate that practice-based QI interventions can be an important tool in improving chlamydia screening in populations at highest risk of chlamydia infection.

A central component of our intervention was universal urine collection prior to clinician history and physical. In a meta-analyses by Taylor et al, automatic collection of urine specimens as part of routine visits was identified as a moderately to highly effective and low-cost intervention to improve STI screening rates.²¹ Our data reaffirm this finding, and provide additional support on how to implement this change in routine pediatric care. Other strengths of our intervention, and likely contributors to a higher than anticipated level of success, included the involvement and engagement of all levels of multi-disciplinary clinical staff; the use of process mapping to identify site-specific barriers to facilitators of screening prior to intervention development; and the use of practice feedback to motivate ongoing behavior change. Provider education has been a mainstay of prior QI interventions to improve STI screening.^18,29 However, the data is mixed on the efficacy for interventions using provider education as the sole or central component.^15,21 In addition to provider education, we utilized several strategies to enhance provider engagement, including providing MOC credit and sending screening run charts to the clinic staff in order to motivate ongoing behavior change. The strategy of providing continuing education credit has been demonstrated to improve human papillomavirus vaccination coverage in our network as well, and should be considered in future QI interventons.³⁰

Our results are consistent with a moderate intervention effect per the criteria identified by Taylor et al in their meta-analysis of STI screening interventions²¹, and are similar to the effect sizes identified in other recent screening interventions in the pediatric and adolescent clinical setting. We identified challenges in implementation largely related to role-definition and workflow. No intervention is without potential strategic costs, and the gains in screening achieved in this high-risk population likely offset the challenges in changing workflow. Importantly, we found no statistically significant different in length of visit time in the pre- and post-intervention periods, suggesting that there was no loss in clinically efficiency as a result of the intervention.

Our findings have both strengths and limitations. The intervention site was a combined primary care and FP clinic with Title X funding, which allowed for confidential service delivery. Clinical sites without integrated sexual health services may face more extensive challenges to confidentiality management. However, many of the components of our intervention, including universal urine collection, provider education and EHR optimization can be tailored to general pediatric practices. With respect to confidentiality, as most Medicaid plans do not send out explanation of benefits statements that could inadvertently disclose testing, in many settings confidential service delivery is possible even without Title X funding. The intervention site serves a population that is largely homogenous with respect to race and socioeconomic status. However, clinic patients, who are primarily African-American and low-income, represent the population most disproportionally affected by chlamydia in the U.S. This is highlighted by the fact that in Philadelphia, the prevalence of chlamydia among adolescent girls ages 15–19 is approximately 7%;³¹ our test positivity rate was 11%. Thus our data demonstrate that our intervention model can work in high risk settings, where impact may be greatest. Our study design did not include a control site, due to the fact that similar chlamydia screening QI work had been previously attempted at the network clinic with the closest age, race and payer patient match. Lastly, our data did not allow us to determine whether individual patients were sexually active and instead used an ecologic approach using background rates of sexual activity in the local population. As rates of sexual activity among adolescent females did not increase nationally during the study period, we would not expect an increase in sexually active young women presenting to care at the clinic to lead to our results.²⁷

The lack of standardized tools for identifying sexual activity within electronic health records remains an ongoing challenge in adolescent health services research. A recent study by Tao et al demonstrated substantial discrepancies in estimates of sexual activity in young women between self-report, Medicaid claims data, and Healthcare Effectiveness Data and Information Set (HEDIS) definitions.³² As sexual history information was not included in our data tracking process, we could not determine whether absence of screening was due to provider nonadherence to guidelines, deferral of testing in non-sexually active patients, cancelled tests, screening at another clinical site, or patient refusal. However, our final screening rate of 50% still falls below the likely proportion of sexually active young women seen at the clinical site, given the 57% rate of sexual activity of African American high school-aged female adolescents in Philadelphia,³³ suggesting that lack of sexual activity was not the sole reason for non-screening. In the future, health systems may consider requiring documentation in the EHR for reasons for not performing guideline-based testing to systematically identify reasons for non-screening in key populations.

In conclusion, we utilized practice-based universal urine collection, provider education and feedback and EHR-based utilities to successfully improve chlamydia screening rates within a high chlamydia prevalence pediatric primary care clinic. Our intervention development and implementation processes were designed to be replicable, and may serve as a template for diffusion of the intervention to additional pediatric and adolescent care sites.