Patients aged 65 and older were eligible for STEADI unless they had a diagnosis of dementia or “frequent falls” (since this was a screening study), were receiving hospice care, or were nonambulatory. Eligible patients had an office visit with a PCP who was participating in the project during the study time period, and had not previously had a fall screening in the prior calendar year.

A voluntary group of OHSU internal medicine and geriatric PCPs were recruited to participate in the project and took part in a 1-hour training session, which provided information on how to use the STEADI workflow and EHR tools. They were incentivized to participate in the study by being able to receive credit for participation toward Maintenance of Certification through the American Board of Internal Medicine. STEADI intervention leaders—called “STEADI champions” (EE and CMC)—delivered separate trainings to providers and staff to educate them on the STEADI protocol, EHR tools, and workflow. Training for providers focused on how to apply the EHR tools to help guide interventions during the office visit. Staff training focused on the clinic workflow, including how to correctly take orthostatics and perform the Timed Up and Go test. The champions also conducted weekly feedback sessions and two “brown bag” lunch refresher trainings to target areas of concern from PCPs and staff.

Informatics staff built STEADI elements into an EHR (Epic) clinical decision support tool to help the clinical workflow align with the STEADI algorithm (see Supplementary Figure 1). To simplify integration, STEADI tools mirrored EHR technology already being used, including developing an annual fall “health maintenance modifier” and a STEADI “Smartset” containing standardized note templates (“dot-phrases”), data entry tables (“docflowsheets”), checklists for orders and diagnostic codes, and Current Procedural Terminology II (CPT II) codes to report on fall-related national quality measures (Casey et al., 2016). Content from CDC-developed patient educational brochures was embedded into the STEADI “Smartset” to include in patients’ after visit summaries. STEADI champions worked closely with an informatics staff assigned to this project to create, test, and review iterative versions of the STEADI EHR tool before full implementation. All EHR tools have now been published as an Epic Clinical Program, which includes an instruction manual for EHR analysts to build the tools into their own system.

Every eligible patient had a fall “health maintenance modifier” added to their chart at the beginning of the study. Eligible patients’ lists of health maintenance modifiers included “Fall Screening Due.” These modifiers were routinely reviewed by the medical assistants before each day’s appointments to identify any necessary health screenings due (e.g., falls, mammography). If a fall screening was due, the medical assistant would add “Fall Screening” to the patient’s appointment notes so it would be seen by the front office staff. If an eligible patient came in for an office visit or Medicare Wellness Visit with their PCP and their appointment notes indicated they were due for a fall screening, the front office staff gave the patient the 12-question Stay Independent questionnaire at check-in to start the clinic workflow. The patient independently completed the paper questionnaire in the waiting room. When the medical assistant roomed the patient, they reviewed the questionnaire and tallied the positive responses, and entered this score into the EHR’s STEADI “docflowsheet.” A Stay Independent score of four or higher indicated highrisk for falls and a score of three or less indicated low-risk (Rubenstein et al., 2011). Although the STEADI algorithm delineates a moderate risk category based on number of falls or injury related to a fall, for purposes of clinical feasibility, our study used only low- and high-risk categories based solely on the score of the STEADI questionnaire.

If low-risk, the medical assistant entered the score and gave the patient a handout on home safety and other fall prevention strategies at the beginning of the visit. If high-risk, the medical assistant completed a Timed Up and Go walking test and Snellen vision test on the way to the exam room. Once in the exam room, the medical assistant performed orthostatic vital signs as part of the rooming process and entered all data into the EHR (Kalinowski, 2008; Podsiadlo & Richardson, 1991). This front-end risk stratification into high- and low-risk allowed PCPs to have the timed walking test, vision, and orthostatic data early in their visit, eliminating the need for additional testing later.

For patients receiving a full STEADI evaluation because their STEADI score was 4 or more, the PCP would open the STEADI “Smartset” within the EHR as part of the visit. This “Smartset” provided access to pertinent orders, the note template, and all fall-related patient education materials within a single location. The PCP reviewed the results of the Timed Up and Go, vision assessment, and orthostatics. If impairment was present, the PCP recommended interventions such as physical therapy referral or Tai Chi, referral to an ophthalmologist, or adjustment of blood pressure medications and improved hydration, respectively. The PCP also determined whether the patient was on adequate vitamin D based on past laboratory levels (if available) and medication list or patient report of daily vitamin D dose. A footwear assessment included a monofilament exam or review of last monofilament exam if the patient was diabetic; for nondiabetic patients, the PCP evaluated whether the patient generally wore appropriate footwear (e.g., no flip flops, no bare feet at home, no high heels) and made appropriate recommendations. The medication list was initially reviewed by the medical assistant, but the PCP was trained to pay special attention to any high-risk medications (National Guideline Clearinghouse, 2015) and to intervene for a high-risk medication by eliminating, tapering the dose, or substituting the medication with a safer alternative (clinic workflow previously published, see Casey, et al., 2017).

When PCPs felt their schedules were too busy, they could request the MA remove the STEADI “flag” and patients would not be given the Stay Independent questionnaire at check-in, thus deferring the screening until a later date. If a patient screened high-risk, but the PCP did not have time to complete additional STEADI fall risk assessments and interventions, usually because of competing medical priorities, the PCP could “defer” the full evaluation until a later date. PCPs would instruct front desk staff in a patient’s check out note to reschedule the patient for a STEADI follow up appointment and include “STEADI follow up” in the appointment notes. That patient would not need to complete the STEADI questionnaire again at the future appointment.

Chart review was conducted on a subset (405) of the 773 eligible patients who received STEADI from June 9 through December 31, 2014. We reviewed all charts of patients identified as high risk based on either the Stay Independent (170 patients) or three key questions (an additional 111 patients) and used a 1:4 sampling ratio for chart reviews of patients who were low-risk based on both questionnaires (reviewed 124 patient charts of 492 who screened low-risk).

Eighteen providers (of 24, 75%) participated in STEADI and saw 1,495 patients aged 65 and older. No demographic information was collected on providers who chose not to participate in STEADI. We excluded 288 patients (19%) due to a prior diagnosis of frequent falls, dementia, being nonambulatory, or on hospice. Of the remaining 1,207 eligible patients, 773 (64%) completed the Stay Independent questionnaire. Thirty-six percent of eligible patients were not screened with the Stay Independent questionnaire because their provider had felt there was not time at that visit to do the screening. Seventy-three percent of STEADI visits occurred as part of routine office visits, 25% occurred during Medicare Wellness Visits, and 2% occurred during new patient visits. Of the 773 screened patients, 603 (78%) patients screened at low-risk for falls, and 170 (22%) screened at high-risk for falls based on the Stay Independent questionnaire (Table 1).

Of the 170 patients screened as high-risk using the 12 Stay Independent questionnaire, 109 (64%) received additional fall risk assessments and interventions, whereas the remaining 36% had their fall prevention intervention deferred (Figure 1). Providers completed appropriate interventions for 85% of patients with gait impairment, 97% with orthostasis, 82% with vision impairment, 90% with vitamin D deficiency, and 75% with foot or footwear issues. Of the 94% of patients who were on one or more high-risk medications, at least one medication was tapered for 22% of patients, and rationale was provided for not tapering high-risk medications in 56%. Providers referred 60% of high-risk patients without gait impairment for community tai chi or fall prevention classes to help prevent future gait and balance issues (data not shown). For 61 (36%) high-risk patients, the provider “deferred” further assessment to a future office visit, usually due to lack of time. Most “deferred” patients did not have further fall assessment during the study period.

The 12-item Stay Independent questionnaire classified 170 (22%) patients as high-risk based on a score of 4 or more. Of these patients, 161 (95%) would have been identified as high-risk using an affirmative response to any one of the three key questions. An additional 111 patients would have been high-risk using the three key questions (Table 1). Only nine patients who screened high-risk using the Stay Independent questionnaire were categorized as low-risk using only the three key questions (these nine patients were analyzed in the high-risk group for purposes of data analysis).

Several significant differences (p < .05) emerged for patients who scored low-risk using both approaches compared to those who scored high-risk using either approach (Table 2). Low-risk patients were, on average, younger (mean age 71.8 vs 73.5 based on 3-item only vs 76.5 based on 12-item). Low-risk patients had fewer comorbid conditions (1.8 vs 2.3 vs 3.8 for the respective approaches; maximum reported comorbidities for any individual was 7). Fifty percent of patients identified as high-risk using the 12-item Stay Independent questionnaire reported falling in the last year, compared to 39% of those identified as high-risk using the three key questions.

Screened patients may not have been representative of the older adult population since providers came from a volunteer sample and participating providers did not screen all eligible patients or evaluate all high-risk patients. This fact could bias the results toward greater uptake of the intervention. Second, it was difficult to identify whether patients who received some fall-risk reduction recommendations (such as participating in community tai chi classes) carried through on these recommendations. Finally, the data collection period was 6 months, so interventions were still underway for many patients, and we were unable to report on health outcomes, such as fall rates. Anecdotally, providers expressed gratitude for having an evidence-based clinical pathway at their fingertips to offer resources and make recommendations to high-risk patients. Importantly, although not formally studied, patients reported satisfaction with STEADI, and for those who adhered to recommended interventions, a belief that the interventions decreased their fall risk.

This study showed that CDC’s STEADI can be adopted in a busy primary care practice. With the STEADI algorithm embedded into the clinic workflow and EHR, PCPs and their clinical teams could consistently implement recommended interventions. Future work should address whether additional strategies could further streamline the process to improve feasibility and how other team members might contribute to the process (e.g., having a pharmacist do the medication review). More sophisticated tracking and follow up could help ensure that high-risk patients with “deferred” visits receive additional interventions and ensure that recommendations for community fall prevention classes and other interventions are followed. Fall prevention remains one of the biggest public health and medical challenges in caring for older adults. Projects such as ours demonstrate how primary care practices can systematically implement an evidence-based algorithm to address fall risk among older adults, and ultimately reduce falls and fall-related injuries.