Quality Initiative to Increase Delivery of Adolescent Hospital-Based Reproductive Health Care

This study was conducted in a 300-bed Midwestern tertiary academic children’s hospital. In 2018, 32% of all adolescent encounters were through the Pediatric Hospital Medicine (PHM) service.

Our multidisciplinary group of stakeholders, including pediatric residents, PHM providers, and adolescent medicine providers, used knowledge of provider workflow and current evidence to generate a key driver diagram (Fig 1). By using quality improvement (QI) mixed methodology and traditional statistics, we evaluated 2 plan-do-study-act cycles. These cycles were synchronized with the academic year because resident physician behavior is a significant target of this initiative. Therefore, the cycles started July 1, 2018, and ended June 30, 2019.

Pediatric residents and PHM section members, including attending physicians and advanced practice providers, received provider education. In-person education at a PHM and a resident meeting was completed before July 2018 and was repeated at a resident meeting in July 2018. Education included presentation of current rates of SHD and reproductive health services delivery, the importance of addressing reproductive health needs of hospitalized adolescents, and updated guidance on contraception prescription and STI screening and treatment. Initiative e-mail reminders, including current SHD rates calculated from manual abstraction of chart reviews, were sent monthly from March 2018 through February 2019. Along with this education, provider workrooms were stocked with free condoms and reproductive health educational materials for adolescents.

Several PHM history and physical (H&P) templates exist in our electronic health record (EHR), including a general H&P template that is used for the majority of hospitalizations and 4 disease-specific templates, including 1 on asthma. On March 16, 2019, a PHM adolescent H&P template was implemented in the EHR to prompt the admitting provider to ask important SHD questions (Supplemental Information). This H&P template automatically populates when a provider selects the general PHM H&P template and the patient’s age is ≥13 years, aligning with our study inclusion criteria. Of note, critical care providers use a separate H&P template that does not specifically prompt for SHD.

Results from the baseline analysis in hospitalized adolescents discharged between May and July 2017 have been published.³³  All hospitalized patients (inpatient and observation) who were ≥13 years of age and treated in the PHM service, including patients initially treated in the critical care service, were included in this initiative. Patients with any documentation of developmental delays were excluded from this study because of potentially different needs and for consistency with published baseline data. Data were collected through manual chart review and entered into a REDCap database. Abstracted patient demographic data included age, sex, language, insurance, race, and ethnicity. Encounter characteristics included admission service (PHM or critical care) and reason for hospitalization (categorized on the basis of provider documentation of reason for hospitalization in the discharge summary). If a patient had a neurologic symptom from an ingestion, reason for hospitalization was categorized as an ingestion complaint instead of a neurologic complaint.

Upon initiation of the adolescent H&P template, monthly e-mail reminders about the initiative were discontinued. The asthma H&P template, however, did not contain SHD questions. Therefore, the H&P template for adolescents admitted for asthma did not contain SHD questions. Thus, the PHM asthma H&P templates were used as a comparison for the automatic PHM adolescent H&P template because they lacked the SHD documentation reminders within the templates.

Our primary outcome measure was SHD in the EHR at any time during the hospital encounter by any provider, including an emergency department provider or social worker. We used the chart search function to search for any mention of sex in reference to behaviors confirmed or denied by the patient. Secondary outcome measures were STI testing and contraception provision. Consistent with our baseline data, STI testing was determined to be completed if done either during the hospital encounter or during the previous 2 weeks on the basis of outside records available during the current hospital encounter. Contraception provision was determined if a female patient was prescribed some form of contraception and did not include condom distribution, as distribution of condoms was rarely included in a patient’s EHR. Patient immunization status was determined on the basis of the Wisconsin Immunization Registry.³⁶  Balancing measures assessed included length of stay (LOS) because the initiative asked providers to address additional needs and whether a patient relations consultation was placed for each encounter as a way to measure, through chart review, patient or family dissatisfaction.

Proportions and means with SDs of measures are reported. Statistical process control charts (p charts) were used to analyze variation and performance over time for outcome measures. Standard rules were used to identify special cause variation.³⁷  The χ² test was used to compare cycle STI testing proportions, with Fisher’s exact test used as necessary. In the previously published baseline analysis, we associated certain subgroups with lower rates of SHD and STI testing, including boys and patients hospitalized for a reason other than ingestion.³³  Therefore, these subgroups were analyzed for each measure. Encounters were analyzed independently because both patient behavior and provider documentation varied between readmissions.

This project was reviewed by the institutional review board and determined exempt.

Descriptive characteristics of the study sample are outlined in Table 1. In each cycle, most of the encounters were with female patients, and most spoke English. The mean age in each cycle was 15 years. The top 4 reasons for hospitalization were the same for cycles 1 and 2.

Special cause variation was identified with regard to our primary outcome of SHD (Fig 2A). Over the course of the initiative, SHD increased from a baseline median of 60% to 82% in cycle 1, a 37% increase from baseline. During cycle 2, 84% of encounters (n = 104) had SHD in the H&P. For the remaining encounters, SHD was located in the progress note (6%), social history tab within the EHR (4.0%), discharge summary (2%), PHM acceptance note (2%), emergency department note (1%), and adolescent medicine consultation note (1%). SHD increased from 37% to 73% in the male subgroup and from 57% to 80% in the noningestion hospitalization subgroup (Figs 2B and 2C).

SHD percentages during baseline and cycles 1 and 2 for each of the top reasons for hospitalization are outlined in Fig 3. Patients hospitalized secondary to asthma were the only group to have increased SHD from baseline to cycle 1 but decreased SHD from cycle 1 to cycle 2. SHD associated with the other three reasons for hospitalization (neurologic complaint, infection, and ingestion) increased during each cycle of the initiative.

For the secondary outcome measure of STI testing, the p chart did not reveal special cause variation over the course of the initiative, including for patients overall, male patients, and patients with noningestion hospitalization (Fig 4). STI testing in some months was above the upper control limit, indicating special cause variation for those months; however, none of the p charts for STI testing revealed a shift or trend toward sustained change or special cause variation overall. STI testing per cycle increased from 12% (18 of 150) of encounters at baseline to 21% (81 of 385 and 30 of 143) during cycles 1 and 2. This increase was statistically significant between baseline and cycle 1 (P = .016) and baseline and cycle 2 (P = .038) but not cycle 1 and cycle 2 (P = .988). STI testing per cycle was also significantly increased in noningestion hospitalizations, with an increase from 11% (11 of 101) of encounters at baseline to 21% (62 of 290 and 23 of 111) during cycles 1 and 2. This increase was statistically significant between baseline and cycle 1 (P = .020) but not baseline and cycle 2 (P = .051) or cycle 1 and cycle 2 (P = .885). STI testing per cycle also increased in boys from 5% (3 of 56) of encounters at baseline to 13% (17 of 130) during cycle 1 and 14% (6 of 42) during cycle 2 but was not statistically significant.

Contraception initiation during the hospital encounter remained low. At baseline, 2.1% (9 of 94) of female patients were started on contraception compared with 2.4% (6 of 255) in cycle 1 (P = .999 compared with baseline) and 1.0% (1 of 101) during cycle 2 (P = .612 compared with baseline and P = .678 compared with cycle 1). Of note, of the 6 patients started on contraception in cycle 1, 4 were found to have an STI on testing and another had disclosed a recent sexual assault.

Potential negative consequences of this initiative, such as prolonged LOS or increased patient relations consultations, were not observed. Average LOS was 3.9 days during baseline, 2.5 days during cycle 1, and 2.6 days during cycle 2. Patient relations consultations remained low at 2.7% of encounters in the baseline period, 1.3% in cycle 1, and 1.4% in cycle 2.

Through provider education, monthly reminders, and use of an adolescent H&P template in the EHR, we successfully increased SHD within our PHM service. The increases occurred across previously identified subgroups of hospitalized adolescents, including boys and patients with medical concerns unrelated to ingestion. Increases occurred upon instituting and reinforcing provider education on missed opportunities to address reproductive health in the hospital setting. These changes were maintained through an adolescent H&P template with social history prompts within the EHR, and monthly reminder e-mails were discontinued. STI testing per cycle also increased, although special cause variation was not found. The initiative did not change rates of contraception provision. Interventions were not associated with prolonged LOS or increased patient relations consultations.

Provider education and reminders led to a significant increase in SHD in cycle 1, which continued in cycle 2 after we implemented the adolescent H&P template and discontinued reminders. We do not know whether the H&P template alone would have resulted in such an increase, but sustaining this increase upon discontinuing time-intensive educational reinforcements is promising. Furthermore, the asthma H&P template served as an interesting comparison with the adolescent H&P template. The H&P template for patients admitted for asthma did not include the social history prompts present in the adolescent H&P template. Among adolescent patients hospitalized for a secondary diagnosis of asthma, SHD increased during cycle 1 but decreased during cycle 2. This divergence in SHD among asthma hospitalizations in cycle 2 suggests that the EHR modifications contributed to maintaining the increased rates seen in cycle 2, but this interpretation is limited by the small number of asthma encounters.

Medical findings that signify risk behaviors were associated with higher rates of reproductive health care delivery. The majority of adolescents hospitalized for ingestion had SHD at baseline, which increased to almost all patients with ingestion hospitalizations by cycle 2. Furthermore, although contraception provision was low throughout the study, most of the patients who were prescribed contraception had an STI diagnosed during hospitalization. Perhaps STI diagnoses highlighted the presence of high-risk behaviors, thus prompting providers to offer contraception to those patients. Additional EHR modifications could help to ensure that these interventions are delivered when STIs are diagnosed, as well as allow bundling of additional reproductive health services when STI testing is performed. Although it is essential for providers to give special attention to reproductive health in such cases, the absence of behaviorally influenced medical findings, such as STIs or ingestions, does not mean that high-risk behaviors are absent. We find encouraging that these interventions significantly increased the rates of SHD for patients hospitalized with complaints other than ingestions. These findings are promising steps toward the goal of addressing reproductive health for all hospitalized adolescents.

We did not find a sustained special cause variation in STI testing with our initiative. A possible reason for this lack of special cause variation is that although all patients in this study should have SHD before discharge, not all present with a social or clinical need for STI testing or agree to it if offered. This variation in need and acceptance likely contributed to the lack of a consistent increase in STI testing. It is encouraging that STI testing proportionally increased for each cycle compared with baseline. Additional work should assess barriers specific to STI testing.

Although pediatric hospitalists endorse the importance of providing contraception during hospitalizations,³⁸  we were unable to have an impact on practice through our interventions. Hospitalists report barriers, including insufficient knowledge about contraception options and sociocultural differences between providers and patients.³⁸  Similar factors could have had an impact on the results of our study. Additional interventions are necessary and could include providing hospitalists and residents with specific scripting about contraception and EHR prompts within templates. Previously identified effective methods for adolescent contraception counseling should be considered when implementing future interventions.³⁹  Another barrier to contraception for adolescents is concerns about confidentiality.⁴⁰  Our institution has since adapted the use of a confidential note type to protect teenagers’ privacy, but its use is in its infancy. Future research could explore the use of disappearing text in the H&P template to prompt providers to appropriately document SHD and place sensitive information in a confidential location outside the H&P, which is especially important given the 21st Century Cures Act. Additional related outcome measures to consider in future studies include offering of STI testing, STI treatment delivery, and referrals to specialists such as adolescent medicine or obstetrics/gynecology.

Our study is limited by several factors. Providers may have discussed sexual history with a patient but not documented that discussion, and we did not assess interventions on sexual risk-reduction counseling. Importantly, our analysis did not include patients with developmental delays, whose needs often go unaddressed by providers.⁴¹  Comparisons between our intervention cycles are limited because of having a longer time frame, and thus more patients, in cycle 1 than in cycle 2. This discrepancy was based on the timing of EHR template production and our methodical decision to align cycles with the academic year. Finally, our study contains limitations inherent in QI methodology; other factors could have contributed to our results beyond the interventions we applied.

We demonstrated increased SHD for hospitalized adolescents during our QI initiative. No special cause variation was noted for STI testing and contraception provision. Increases in SHD occurred across subgroups of patients noted at baseline to receive less attention to reproductive health, including boys and patients hospitalized for reasons other than ingestion. Improvement was maintained when provider education reinforcement stopped, suggesting that the EHR intervention was effective in prompting providers to discuss sexual history. This finding is especially exciting because the EHR intervention is automatic; does not require a significant amount of time, labor, or finances; and can be replicated across other institutions. Future research could assess the feasibility and impact of clinical decision support tools within the EHR for delivery of services. Although future interventions are needed to increase comprehensive delivery of reproductive health care to adolescent patients, including those with developmental delays, this study revealed successful interventions to begin to meet needs of adolescents in nontraditional clinical settings.^{14–19,33,34}