Addressing Transportation Insecurity Improves Attendance at Posthospitalization Appointments

The Barbara Bush Children’s Hospital is an urban, academic children’s hospital with 30 inpatient general pediatric beds (excluding newborn nursery, NICU, and PICU) within Maine Medical Center (MMC), a 637-bed hospital located in Portland, Maine. The pediatric hospital medicine (PHM) service is a teaching service with residents and medical students involved in the care of ∼1100 patients annually, which is approximately half of all patients admitted to the inpatient pediatric unit (IPU).

Our institution participates in the Improving Pediatric Patient-Centered Care Transitions (IMPACT) project, a collaborative effort that seeks to improve pediatric patient outcomes through enhancing discharge and follow-up processes. The interprofessional QI team leading project IMPACT at MMC uses plan-do-study-act (PDSA) cycles to implement a 4-part discharge bundle (including a discharge readiness checklist, teach-back education,⁹  timely and complete handoff to the primary care provider [PCP], and a postdischarge phone call).⁸  As part of project IMPACT, an outpatient nurse performs postdischarge follow-up phone calls within 3 days after discharge and asks several questions, including questions regarding transportation insecurity screening and attendance at PHAs, which is then documented in our electronic health record (EHR) in a telephone encounter. All patients discharged from the hospital from the PHM service are included in this QI research project. Patients not discharged from the hospital (eg, transfers to other facilities) are excluded from analysis. Care managers, physicians, and nurses are each responsible for completing sections of the discharge readiness checklist in the EHR during hospitalization. In-person, video, or telephone interpreter services are used to deliver teach-back education at discharge for patients and families with limited English proficiency. A data management assistant reviews charts and inputs data from the above interactions with the patient as well as follow-up attendance into a Research Electronic Data Capture capture tool.¹⁰ 

PHAs are made at the discretion of the inpatient care team when it is felt that a visit is necessary to ensure optimal hospital-to-home handoff. Currently, no consistent approach is used for scheduling PHAs other than being at the discretion of the attending physician at the time of discharge from the hospital. These appointments are included in written discharge education materials and reinforced verbally by using teach-back education.⁹  During our project period, PHAs were felt to be important and scheduled before discharge for 66.4% of the patients.

After noting that several patients missed appointments because of transportation insecurity, we further investigated the subset of patients who missed PHAs between 2014 and 2017. During the baseline period (2014–2017), 3549 patients were discharged from the hospital from the PHM service, 844 (24%) had transportation insecurity screening documented in the discharge checklist (Supplemental Fig 7), and 28 (3.3%) were identified as having transportation insecurity (Supplemental Fig 8). Within this set of patients with documented transportation insecurity, 38% had private insurance, 58% had state Medicaid, and 4% were uninsured or self-pay. Patients with transportation insecurity had a higher frequency of Medicaid insurance or no insurance (22 of 28 [79%]) compared with those not having transportation insecurity (499 of 816 [61%]) (P = .03) (Supplemental Fig 9). Furthermore, when transportation insecurity was noted, the only documented mitigation strategy was provision of a taxi or bus pass or gas card for 1-time use for transportation home from the hospital.

In July 2018, an interprofessional team of nurses, PHM physicians, residents, data management assistants, and care coordinators began meeting monthly using PDSA cycles to trial tests of change, review process control p-charts, and implement practice changes to address the inconsistencies in transportation insecurity screening, employ mitigation strategies when relevant, and monitor for improvement in PHA attendance rates. Our key drivers for this project included lack of staff and family knowledge of transportation resources and benefits, incomplete transportation insecurity screening rates, redundant documentation locations within our EHR, and inconsistent communication back to the PCP offices when transportation insecurity was identified (Fig 1).

Education was provided to residents, PHM physicians, nurses, and care coordinators periodically throughout this project in various formats including in-person education at staff meetings, resident forums, PHM physician division meetings, and nursing skills fairs. Just-in-time education among team members, e-mail reminders, laminated cards with project goal reminders, and updates on our hospital intranet were also employed. Educational topics included the significance of transportation insecurity, suggested scripting to screen for transportation insecurity, EHR documentation instructions, and explanation of resources available to patients with identified transportation insecurity.

The discharge checklist was modified to allow all team members (nurses, care managers, social workers, and physicians) to document transportation insecurity screening at any time before discharge (rather than being only accessible to care coordinators). We also merged data from multiple flow sheets into 1 location within the EHR to reduce redundant charting.

Although our institution has used 1-time transportation solutions (eg, taxi or bus passes or gas cards) when transportation insecurity is identified for hospital-to-home transportation, our group focused on sustainable problem-solving to help families not only get home from the hospital but also to successfully attend their PHA and future health care appointments. As part of that effort, our care managers created a list of local transportation resources and benefits. These resources included a nonemergency medical transport service supported by our state Medicaid insurance plan and mileage reimbursement through insurers. Once transportation insecurity was identified, if the patient had a transportation benefit as part of their insurance, care managers or social workers taught the family to use and access that resource to set up transportation to their PHA and any future appointments.

Through the creation of “smart phrases” in our EHR, we were able to add standardized information about local transportation resources to patient discharge instructions. We also created a smart phrase to notify PCPs via the discharge summary when transportation insecurity was identified for their patients. We added International Classification of Diseases, 10th Revision code Z59.8 “other problem related to housing or economic circumstances” (with “42 transportation” submodifier) and changed the display to “Transportation Insecurity” in the patient problem list in our EHR when identified. In our EHR, the problem list will display at any future encounters with providers within our EHR network. Finally, we added clarifying questions to the follow-up phone call template to guide our outpatient nurse to screen for any missed or new transportation barriers (such as lack of a working vehicle, usual ride not being available, no money for gas, inability to afford public transportation, or not being set up with community resources).

Our primary process measure was the percentage of PHM patients with “assessment of transportation insecurity” completed in the discharge readiness checklist in our EHR (Supplemental Fig 6). Our aim was to achieve 75% completion by 18 months. Specifically, the mode of transportation home and to a PHA can be documented from a selectable menu list (Supplemental Fig 7). If there was no identified mode of transportation, a note is entered under “other.” If this section was not completed, we considered this a lack of transportation insecurity screening documentation. Documented screening rates were evaluated and shared monthly at our QI group meetings.

Through collaboration with PCP offices, we collected data on PHA attendance as our primary outcome measure. For patients with PCPs within our health care system, this was determined via encounter audit by using the shared EHR. For those outside of our health care system, a survey was faxed to the PCP office. If the survey was not returned within 2 weeks, it was re-sent.

The percentage of IPU discharges before 2 pm was tracked monthly as a balancing measure to monitor for discharge delays during our intervention.

We used statistical process control p-charts to display the proportion of patients with transportation insecurity screening documented in the EHR, attendance at first scheduled PHA, and discharges before 2 pm; on these charts, we included centerlines to portray overall average proportions and 3-σ control limits. We used Associates in Process Improvement rules for detecting special causes, which come with the program we used, QI Charts, a Shewhart control chart application for Microsoft Excel.¹¹  We used a 2-tailed χ² test with continuity correction to compare the frequency of having Medicaid or no insurance for patients with and without transportation insecurity.

The MMC Institutional Review Board considered the project to be a local QI initiative and thus not human subjects research. Informed consent beyond standard consent for treatment was not required.

During the study period of July 2018 to December 2019, 1731 acute medical patients were discharged from the PHM service and included in analysis.

Our primary process measure, EHR documentation of transportation insecurity screening, improved from 1% to 94% (Fig 2). With more consistent screening, we saw the average percentage of patients and/or families identified as having transportation insecurity increase from 1% to 5% (Fig 3), with a chronological trajectory that mirrors that observed for increases in screening documentation rates (Fig 2).

Combining EHR review data and results from returned fax surveys from PCPs, we were able to ascertain if our patients attended their scheduled PHA 88% of the time (Supplemental Fig 10).

PHA attendance improved from 62% to 81% for all patients (Fig 4A) and improved from 0% to 57% for the transportation-insecure subgroup over the last 12 months of our project (Fig 4B). The predominant inflection point for all discharged patients correlated with care management performing transportation insecurity screening on admission and emphasizing the importance of PHAs (Fig 4A). The predominant inflection point for patients with transportation insecurity correlated with adding transportation insecurity screening to our nursing discharge checklist (Fig 4B). Other interventions were not correlated with inflection points and are detailed in Fig 4 A and B.

The percentage of discharges from the IPU before 2 pm, our balance measure, was consistently 53% during the study period (Fig 5).

To date, there are no specific publications in which authors report improving PHA attendance by implementing routine screening for transportation insecurity. Our initiative revealed that consistent transportation insecurity screening is an interprofessional intervention with the potential for significant impact, as shown by the increase in PHA attendance. Marcondes et al⁵  found that scheduling a PHA was associated with fewer emergency department (ED) visits within 30 days of hospital discharge.⁵  Because the ED is not always the appropriate care location for children,^12,13  it is important to partner with families to devise transportation strategies to increase the successful use of their medical homes and relieve the burden on acute care settings such as urgent care centers or EDs. Inpatient hospitalizations present a unique opportunity to identify transportation insecurity, ensure inclusion of this risk factor into the patient’s EHR, including discharge communication to the PCP, and provide education to families regarding available resources. It is our hope that the impact of these endeavors persists beyond the hospital-to-home transition period.

Our project consisted of several overlapping interventions, making it challenging to determine which of these interventions was most responsible for the outcomes observed. Near universal transportation screening helped us to uncover more patients with transportation barriers (Fig 3). We noted that simply emphasizing the importance of attendance at the first PHA alone led to an immediate increase in PHA attendance rates as demonstrated by the improved overall attendance at the first PHA seen in all PHM patients (“All Patients on PHM Service” in Fig 3A). Educating care managers, nurses, and physicians about the significance of transportation insecurity and helping families find resources were other key interventions to improving PHA attendance for those with transportation insecurity. We found that many patients were unaware of their existing insurance transportation benefits, which allowed us to educate them on this sustainable resource for transportation assistance for all patients with state Medicaid and some patients with private insurance. Compared with patients who have private insurance, Medicaid beneficiaries are more often affected by multiple barriers to timely presentation to primary care and have a high association with ED use.⁴  Rideshare programs have been shown to increase show rates for appointments for Medicaid patients.¹⁴  In our state, Medicaid offers rideshare resources to members, and this was a key mitigation strategy.

As with all QI initiatives, local context may affect implementation strategies. Not all patients have a transportation benefit through their insurance plan. Therefore, transportation insecurity may be identified in patients for whom we have minimal solutions to offer. Nonemergency medical transport has been a required benefit for all Medicaid recipients in the past, although some states have obtained waivers to not be required to provide this benefit for some Medicaid recipients in their states.¹⁵  Despite this, we feel that our results indicate that identifying transportation insecurity and emphasizing the importance of the PHA can improve PHA attendance even in the absence of an available mitigation strategy as shown by the improved PHA attendance rate for all patients discharged from the PHM service (with and without identified transportation insecurity).

We were able to determine if follow-up appointments were attended 88% of the time (Supplemental Fig 10). We attribute this high percentage of known follow-up attendance to having access to charts through our shared EHR in our health system. Unknown PHA attendance data were always due to a lack of response via our fax survey method from PCP practices scattered throughout the state outside of our health care system. This was felt to be a reflection on those practices and not to any particular patient characteristics making it less likely that it would affect our outcomes significantly. One of our early PDSA cycles was focused on improving our PCP survey response rate above our initial baseline of 64%. Our research intern started sending a second survey if the first one was not returned within 2 weeks. Subsequently, our response rate improved to a mean of 88.4%. The monthly nonresponse rate was stable over time (all remaining within the confidence limits, except for our last month). We did not focus additional efforts on trying to further decrease the mean nonresponse rate of 12%.

A limitation of this QI project is that we currently have no standardized algorithm guiding scheduling of a PHA. This lack of a consistent approach may lead staff to be unclear with families about the importance of attending their PHA and screening for transportation insecurity. Another limitation of this QI project is that patients and families may not be comfortable discussing transportation barriers with the care team in person; thus, true rates of transportation insecurity may have been underreported. Use of a tablet or anonymous survey may help identify more patients and/or families with transportation insecurity.

When identified, we added transportation insecurity to the problem list in our inpatient EHR, and this problem list remains visible to providers for future encounters within our health care system including admissions, PCP office encounters, subspecialty appointments, and ED visits. However, this may not be feasible in other EHR systems and would require different modes of information sharing if PCP networks do not share the same EHR system as the inpatient facility.

It may not be feasible for other institutions to fund a dedicated position for postdischarge follow-up phone calls. The degree of influence of these calls on first PHA attendance was not directly measured in this study given the overlap of multiple interventions.

Taking a transportation history and identifying patients with transportation insecurity in the workflow on pediatric IPUs before discharge can improve first PHA attendance. For most patients, emphasizing the importance of the PHA was all that was required to lead to this improvement. Familiarizing our care team with available transportation resources allowed care team members to educate families about how to access those resources not only for attendance at PHAs but also future health care appointments. At our institution, the changes we implemented are sustainable as they are now included in an already established workflow (except for our postdischarge follow-up phone call registered nurse being reassigned during the COVID-19 pandemic). To maintain efforts by pediatric residents and new hires, we will need to continue to educate new staff about our workflow and the importance of transportation insecurity screening and mitigation.

We thank Wendy Craig, PhD, for review of our article and statistical support whose work was supported in part by the Northern New England Clinical and Translational Research grant U54GM115516.

ED

emergency department

EHR

electronic health record

IMPACT

Improving Pediatric Patient-Centered Care Transitions

IPU

inpatient pediatric unit

MMC

Maine Medical Center

PCP

primary care provider

PDSA

plan-do-study-act

PHA

posthospitalization appointment

PHM

pediatric hospital medicine

QI

quality improvement