Fall prevention implementation strategies in use at 60 United States hospitals: a descriptive study

Abstract

Background

To guide fall prevention efforts, United States organisations, such as the Joint Commission and the Agency for Healthcare Research and Quality, have recommended organisational-level implementation strategies: leadership support, interdisciplinary falls committees, electronic health record tools, and staff, family and patient education. It is unclear whether hospitals adhere to such strategies or how these strategies are operationalised.

Objective

To identify and describe the prevalence of specific hospital fall prevention implementation strategies.

Methods

In 2017, we surveyed 80 US hospitals participating in the National Database of Nursing Quality Indicators who volunteered for the study. We conducted descriptive statistics by calculating percentages for categorical variables and the median and IQR for count variables.

Results

A total of 60/80 (75%) of hospitals completed the survey. The majority of hospitals were not-for-profit (98%) and urban (90%); more than half were Magnet (53%), small (53%) and teaching (52%). Hospitals were more likely to use leadership strategies, such as updating fall policies in the past 3 years (98%) but less likely to reward staff (40%). Hospitals commonly used interdisciplinary falls committees (83%) but membership rarely included physicians. Hospitals lacked access to electronic health record tools, such as high-risk medication warnings (27%). Education strategies were commonly used; 100% of hospitals provided fall education at staff orientation, but only 22% educated all employees (not just nursing staff).

Conclusions

Our study is the first to our knowledge to examine which expert-recommended implementation strategies are being used and how they are being operationalised in US hospitals. Future studies are needed to document fall prevention implementation strategies in detail and to test which implementation strategies are most effective at reducing falls. Additionally, research is needed to evaluate the quality of implementation (eg, fidelity) of fall prevention interventions.

INTRODUCTION

Each year, between 700 000 and 1 million falls occur among hospitalised patients in the USA.^1,2 Falls among hospitalised patients can result in injury, longer hospital stays and unnecessary healthcare costs. Studies have estimated that 25%–50% of in-hospital falls result in injuries.^1,3,4 An in-hospital fall resulting in an injury can add 6 days to a hospital stay and an additional $14 000 USD in healthcare spending.^5–7 Researchers have identified patient-level (eg, age and frailty) and organisational-level (eg, lack of leadership and staff training) factors associated with in-hospital falls and falls with injury.^4,8 To address these risk factors, numerous fall prevention interventions have been implemented (eg, risk assessments and enhanced vigilance) but with mixed success.^9–13 One possible explanation is that there is little evidence on how to implement fall prevention strategies effectively.^11–14

To guide fall prevention efforts, numerous US-based organisations, such as the Institute for Healthcare Improvement, the Joint Commission and the Agency for Healthcare Research and Quality have developed implementation guidance.^1,4,8,12 The guidance recommends certain implementation strategies—or specific methods to support the application of an intervention.^15,16 An intervention, such as hourly rounding, is designed to directly influence outcomes (eg, fall rates), whereas implementation strategies are intended to improve the quality with which interventions are implemented (eg, staff education is intended to improve the use of hourly rounding). Similar implementation strategies are recommended by these organisations, including an interdisciplinary falls committee; leadership support; staff, patient and family education; and electronic health record (EHR) tools for fall prevention. While these strategies are recommended, fall prevention intervention studies rarely report on implementation strategies.¹² For example, a systematic review of fall prevention interventions found that only 2 out of 21 studies reported on the use of fall prevention EHR tools.¹² Additionally, these strategies are broad and could be operationalised in numerous ways. Leadership support, for example, could include setting clear expectations for implementation, providing rewards to employees who perform well at implementation or providing additional support for implementation.^17,18 Since the guidelines are not prescriptive, there is likely variation in how organisations carry out fall prevention implementation strategies.

To understand fall prevention implementation at the organisational level, our study aims to identify and describe the prevalence of specific fall prevention implementation strategies used by hospitals at the organisational level. Understanding how hospitals vary in their employment of implementation strategies could provide direction for the development and implementation of future organisational-level strategies.

METHODS

Study design

We used a cross-sectional study design to identify and describe the prevalence of in-hospital, organisational-level fall prevention implementation strategies in 2017. The unit of analysis was the hospital. We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology checklist for cross-sectional studies to report the study findings.¹⁹

Sample

A subset of US hospitals participating in the National Database of Nursing Quality Indicators (NDNQI) were selected for participation in this study. Eligibility included non-federal hospitals that submitted patient falls data for Q1 or Q2 2017. Press Ganey, the organisation that owns the NDNQI, sent out 800 study invitations to hospitals meeting the above criteria with a goal of recruiting 80 sites. Within 24 hours, 189 hospitals expressed interest in participating. We used this convenience sample of 189 volunteers to randomly select 80 hospitals to participate via stratified random sampling. We sampled to include 20 hospitals in four strata based on teaching status (yes/no) and hospital size (<200 beds and ≥200 beds). Each participating hospital was asked to provide data for one adult medical or medical surgical unit. Of these 80 hospitals, 60 hospitals returned the survey (75% response rate). Hospitals that declined to participate cited reasons such as change in organisational leadership.

Variables

We developed a survey assessing hospital fall prevention implementation strategies, where staff could document whether a particular strategy was currently used in the hospital. The survey was reviewed by 10 fall prevention experts and refined before pilot testing. The survey was piloted at three sites, and the study team made minor revisions to the survey. The final survey contained 63 questions covering fall prevention implementation strategies at the organisational level in the domains of leadership support, interdisciplinary falls committee, EHR tools and staff, patient and family education and training (online supplementary file 1). We selected these domains based on available guidance regarding falls implementation.^1,4,8

In addition, we collected data on hospital characteristics from the NDNQI database: teaching status, hospital size, hospital ownership, Magnet status and urban or rural location. The definitions of these measures are described below.

Leadership support was grouped into three areas: expectations, rewards and support. Expectations included variables measuring whether hospital fall policies have been updated in the past 3 years, whether hospital falls data are included in the hospital’s annual board of trustees report, whether data are reviewed quarterly by hospital leadership and whether data are included in the annual performance review of the hospital’s safety officer or director or unit manager. Rewards included one binary variable assessing whether units that had high-performance on falls rates received rewards (eg, gift cards for staff, pizzas and public notice). Support included three concepts: support for low-performing units, resource availability for all units and staffing availability for adult nursing units. Support for units that had low-performance included: (1) quality management or safety consultation, (2) interdisciplinary consultation, (3) additional equipment or furniture and (4) additional staffing (eg, sitters). Resource availability for all units included bed alarms, other alarm equipment, specialty low beds, safety equipment (eg, hip pads) and sitters. Interdisciplinary staffing availability for adult nursing units included palliative care specialists, psychiatrists, psychatric advanced practice nurses (APNs), geriatric APNs, geriatricians, geropsychiatrists and geriatric pharmacy specialists.

Interdisciplinary committee variables included committee meeting frequency, high-level clinical committee chair and committee membership. Committee membership included the number of staff roles and disciplines represented on the committee, the number of nurses, the number of other allied health staff (eg, pharmacists, dieticians and safety or quality management specialists), the number of therapists (eg, physical, occupational or respiratory), the number of other support staff (eg, informatics, environmental services and central supply) and the number of physicians.

EHR tool variables included four categories: fall risk assessment, fall status alerts, fall strategies list and fall prevention links to resources. Fall risk assessment was measured with two variables: whether the fall assessment tool was embedded within the admission nursing tool and whether fall assessment was embedded in the daily nursing assessment tool. Fall alerts included patients’ fall risk status and high-risk medications. Fall prevention strategies consisted of three binary variables: whether the EHR displayed a list of targeted (based on patient’s risk factors), non-targeted fall prevention strategies (ie, universal precautions) or if the EHR does not provide a list but has an item that states ‘Fall protocol implemented’. Fall prevention links to resources consisted of two variables: whether the EHR was linked to the hospital fall prevention protocol or linked to central supply for ordering fall prevention equipment or furniture.

Education variables included staff-level and patient/family-level variables. Staff-level variables included whether fall prevention education was required: (A) at staff orientation or (B) during an annual staff training. Additionally, we measured whether fall prevention education at staff orientation and at annual staff training was mandatory for nursing staff only or for all staff. Patient/family-level variables included whether patients or their family received one-on-one education, printed materials or a fall prevention video on a hospital educational television channel.

Hospital characteristics included five binary variables. Hospital ownership (not for profit nd for profit-investor owned), The American Nurses Credentialing Center Magnet Recognition Program status (Magnet and non-Magnet), hospital size (<200 staffed beds and ≥200 staffed beds), teaching status (non-teaching and teaching/academic medical centre) and urban location (≤50 000 population and >50 000 population).

Data collection procedure

Data were collected between 1 October and 31 December 2017. At each hospital, a person designated by the NDNQI as the site coordinator was responsible for gathering the data. The site coordinator serves as the liaison between the hospital and NDNQI. The individual who is typically selected is a nurse with a leadership role within the hospital. Given the wide range of topics covered in the survey, the NDNQI site coordinator was responsible for identifying which individual(s) within the hospital would be the best person to answer each question (eg, chief information officer and quality coordinator). To maintain site confidentiality, Press Ganey administered the survey through a web-based secure system. Two reminder emails were sent.

DATA ANALYSIS

We conducted descriptive statistics by calculating percentages for categorical variables and the median and IQR for count variables. We used complete case analysis to deal with missing data since the amount of missing data was minimal (<7%). We conducted the analyses using Stata V.13.0.

RESULTS

The majority of study sample hospitals (n=60) were not-for-profit (98%) and urban (90%) (table 1). More than half were Magnet (53%), small (53%) and teaching hospitals (52%). The study sample had a greater percentage of not-for-profit hospitals compared with a nationally representative sample of American Hospital Association (AHA) hospitals (n=4554)²⁰ (98% vs 69%, respectively). The study sample also had a greater percentage of large (>200 beds) (47% vs 29%), urban (90% vs 65%) and Magnet (53% vs 8%) compared with AHA hospitals.

Table 1.

Hospital characteristics: comparison of study hospitals participating in NDNQI and a nationally representative sample of hospitals

Characteristics	NDNQI sample (n=60) n (%)	AHA sample (n=4454)* n (%)
Hospital ownership
Not-for-profit	59 (98.3)	3091 (69.4)
For profit	1 (1.7)	1363 (30.6)
Bed size
Less than 200 beds	32 (53.3)	3185 (71.5)
200 or more beds	28 (46.7)	1269 (28.5)
Urban hospital (>50 000 population)	54 (90.0)	2895 (65.0)
Magnet hospital	32 (53.3)	356 (8.0)

^*We calculated sample characteristics for non-governmental, acute care hospitals in the American Hospital Association (AHA) database, a nationally representative sample.²⁰

NDNQI, National Database of Nursing Quality Indicators.

Leadership support

All hospitals (100%) used at least one implementation strategy for setting expectations. Strategies such as updating hospital fall policies in the past 3 years (98%) and including falls in the annual board of trustees report (95%) were almost universal (table 2). More than half of the hospitals (65%) reported including fall rates in the annual performance review of the hospital’s safety officer or director, and 40% of hospitals reported including fall rates in the performance review of unit managers. Rewards for high-performing units occurred in 40% of all hospitals. Most hospitals (62%) used at least one strategy to provide support to low-performing units. The most commonly used strategy was providing quality management and safety consultation, (50%) and the least commonly used strategy was providing additional staff (10%).

Table 2.

Leadership support strategies

	Total (n=60) n (%)
Setting expectations
Used at least one setting expectations strategy	60 (100.0)
Fall policies have been updated in last 3 years	59 (98.3)
Falls included in annual board of trustees report	57 (95.0)
Falls reviewed quarterly by top leadership	56 (93.3)
Falls included in annual performance review:
Safety officer or director	39 (65.0)
Unit manager	33 (55.0)
Providing rewards (for high-performing units)	24 (40.0)
Providing support
Used at least one support strategy for low performing units	37 (61.7)
Quality management/safety consultation	30 (50.0)
Interdisciplinary consultation	13 (22.4)
Additional equipment or furniture	10 (16.7)
Additional staffing	6 (10.0)
Resource availability for all units
Chair alarms	58 (96.7)
Bed alarms	54 (90.0)
Sitters	53 (88.3)
Low beds	37 (61.7)
Safety equipment	27 (45.0)
Staffing availability for adult nursing units
Palliative care	57 (95.0)
Psychiatrist	55 (91.7)
Psychiatric APN	19 (31.7)
Geriatric APN	15 (25.0)
Geriatrician	14 (23.3)
Geriatric psychiatrist	10 (16.7)
Geriatric pharmacy specialist	5 (8.3)

APN, advanced practice nurse.

In terms of resource availability for all units, most hospitals had chair alarms (97%), bed alarms (90%) and sitters (88%). The majority of hospitals had low beds (62%) and some hospitals had safety equipment (eg, hip pads) (45%).

Adult nursing units frequently had access to palliative care specialists (95%) but had less access to geriatricians (23%), geriatric psychiatrists (17%) and geriatric pharmacy specialists (8%).

Interdisciplinary committee

Most hospitals had an interdisciplinary falls committee (83%). Hospitals with a committee typically met every 1–2 months (86%) and had a high-level clinical leader as the committee chair (66%) (table 3). Hospitals had a median of 8 (IQR: 5–10) staff roles represented on the committee and a median of 4 (IQR: 3–5) disciplines represented on the committee. Nurses were the most represented discipline (median=5, IQR: 4–5) followed by other allied health staff (median=1, IQR: 1–2). Physicians were the least represented discipline (median=0, IQR: 0–0).

Table 3.

Hospital interdisciplinary falls committee characteristics

	Total* (n=50) n (%) or median(IQR)
Committee meeting frequency
Every 1–2 months	43 (86.0)
Quarterly	4 (8.0)
1–2 times annually	3 (6.0)
High-level clinical leader chair	33 (66.0)
Committee membership
Number of staff roles represented	8 (5–10)
Number of disciplines represented	4 (3–5)
Number of nurses	5 (4–5)
Number of other allied health staff	1 (1–2)
Number of therapists	1 (0–2)
Number of other support staff	1 (0–1)
Number of physicians	0 (0–0)

^*The total is less than 60 because nine hospitals did not have a hospital-level falls committee and one hospital had missing data on the falls committee variables.

EHR tools

Most hospitals had a fall assessment tool embedded in the daily nursing assessment tool (98%) or in the admission nursing tool (95%) (table 4). The majority of hospitals (61%) had a fall alert to display high-risk falls status for all departments. Fewer hospitals (27%) had a high fall-risk medication alert. Most hospitals had a fall prevention strategies list including non-targeted (70%) or targeted (61%) intervention lists. Over a third of hospitals had a link available within the EHR to allow staff to view the hospital’s fall prevention protocol in real time (38%) or had a checkbox within the EHR allowing staff to indicate if a fall prevention protocol was used for a specific patient (36%). Very few EHR systems had the ability to order supply tools (9%). Hospitals also reported other EHR tools (38%), such as displaying a patient’s falls risk history or risk for fall injury.

Table 4.

Electronic health record tools

	Total* (n=56) n (%)
Fall assessment
Embedded in nursing assessment tool	55 (98.2)
Embedded in admission nursing tool	53 (94.6)
Fall alert
High-risk display for all departments	34 (60.7)
Warnings for high-risk medications	15 (27.0)
Fall prevention strategies list
Use of targeted or non-targeted strategies	45 (80.4)
Non-targeted strategies	39 (69.7)
Targeted strategies (by risk)	34 (60.7)
Othert	21 (37.5)
Fall prevention protocol
Links to hospital fall prevention protocol	21 (37.5)
Checkbox to indicate implementation of protocol	20 (35.7)
Order supply tool	5 (8.9)

^*The total is less than 60 because four hospitals had missing data on the EHR functionality variables.

^†The other category included functionalities such as, displaying falls history, falls-specific surveillance board, patient handoff tool, patient transport communication tool, falls prevention order set, and risk for injury from falls assessment tool.

Education for staff, patients and families

All hospitals provided fall prevention education at staff orientation, which was typically provided to nursing staff (78%) rather than all employees (22%) (table 5). The majority of hospitals provided annual training to employees (69%), which was typically delivered to nursing staff (41%) rather than all employees (28%). Nearly all hospitals (98%) used one-on-one fall prevention education with patients and families and provided printed fall prevention materials to patients and families (85%). Less frequently, hospitals had a fall prevention video on an educational television channel for patients and families (28%).

Table 5.

Fall prevention education for staff, patients and families

	Total* (n=58) n (%)
Staff
Education at staff orientation
Overall	58 (100.0)
Nursing staff only	45 (77.6)
All employees	13 (22.4)
Annual staff training
Overall	40 (69.0)
Nursing staff only	24 (41.4)
All employees1	16 (27.6)
Patient/families
One-on-one education	57 (98.3)
Printed materials	49 (84.5)
Fall prevention video on TV	16 (27.6)

^*The total is less than 60 because two hospitals had missing data on the education and training variables.

DISCUSSION

The purpose of this study was to identify and describe the prevalence of hospital fall prevention implementation strategies at the organisational level. Overall, we found that hospitals varied in how leadership support was operationalised; for example, hospitals were more likely to use strategies for setting expectations for falls (eg, including falls in the annual board of trustees report) than support strategies (eg, providing quality improvement consultation for units with poor performance in falls). We also found that while most hospitals had an interdisciplinary falls committee, the committees were mainly composed of nursing staff. Hospitals also had limited access to EHR tools for fall prevention, with the exception of fall risk assessment tools. Education strategies for staff, patient and families were the most commonly used implementation strategy across hospitals; however, hospitals were more likely to educate nursing staff compared with other hospital staff.

Our study found that hospitals frequently use leadership strategies to set high expectations for fall prevention, such as having leadership review fall policies or hospital fall rates. This finding contrasts with a previous survey of Belgium hospitals (n=113) that suggested only 1/3 of hospitals had a formal falls prevention policy.²¹ The Belgium study, however, was conducted in 2006 so it is possible that more hospitals have implemented formal policies since then. Less commonly, hospitals rewarded units for high-performance or provided additional support to units with low performance, such as interdisciplinary consultation. It is possible that this imbalance—high expectations but lower rewards and support—could undermine hospitals’ climate for fall prevention.^17,18 Failing to create a strong implementation climate—that is, the extent to which an innovation is expected, rewarded and supported by an organisation—may undermine employees’ motivation, which negatively affects how consistently employees use prevention practices.^17,18 Past studies suggest that lack of consistency (ie, implementation fidelity) may explain null findings of fall prevention interventions.^4,22–25

Most hospitals had an interdisciplinary falls committee; however, the leadership and membership varied. Most hospital fall committees had the greatest representation from nurses and the lowest representation from physicians. Researchers have recommended an interdisciplinary approach to falls prevention due to the complex array of fall risk factors.¹⁰ Past fall prevention interventions have cited lack of physician engagement as a barrier to implementation, and researchers have called for increased involvement of physicians in falls prevention.^26–28 Physicians typically initiate clinical guidelines (eg, delirium and ambulation orders) for fall prevention and can provide leadership to engage staff and patients in fall prevention efforts. Increased representation, however, is only part of the solution. Interdisciplinary teams need to be high functioning to be effective. Past studies on interdisciplinary teams suggest that having shared goals and clear roles may enhance team functioning.^29,30 However, little is known about what makes interdisciplinary falls committees effective.

Our study findings suggest that hospital EHRs lack tools for fall prevention, such as fall risk alerts and targeted fall prevention strategies. Past studies suggest that lack of EHR tools leads to poor provider-to-provider communication about patients’ fall risk.^26,31–34 Conversely, access to EHR tools, such as risk assessments combined with targeted fall prevention strategies, has shown promise in reducing fall rates; however, not all hospitals can implement such tools due to lack of EHR sophistication.^34,35 Prior research suggests that EHR sophistication varies by hospital characteristics, such as hospital size.^36,37

Our study suggests that fall prevention education for staff, patients and families is commonly used by hospitals; however, hospitals vary in how and to whom education is delivered, which may impact effectiveness. Hospitals were more likely to provide education to nursing staff rather than delivering education to all employees, which is consistent with prior studies.^25,38,39 Lack of interdisciplinary education may negatively affect fall prevention implementation; one study reported that a lack of interdisciplinary fall prevention education led to lack of non-nursing staff engagement (eg, pharmacists).³⁸ Our findings suggest that one-on-one patient education strategies are common; however, education via other modalities (print materials and television) were less common. Prior studies suggest that single component and multi-component patient education programmes are effective in reducing falls, but it is unclear which strategies are most effective (eg, delivery methods).^34,40–43

Overall, our study identified hospital-level variation in organisational fall prevention implementation strategies; however, further research is needed to understand how this variation affects fall rates. Many of the recommended implementation strategies for fall prevention, such as leadership support, have demonstrated a positive impact on falls but have not been rigorously tested through randomised trials.²² Additionally, few fall prevention studies report on implementation strategies making it difficult to replicate past interventions, although more recent studies have started reporting on implementation.^44–46 Future research is needed to better document fall prevention implementation using standardised templates,¹⁶ rigorously evaluate fall prevention implementation approaches and examine whether fall prevention implementation approaches are associated with fall rates.

Limitations

This study has several limitations. First, the study included a small number (n=60) of hospitals participating in NDNQI,⁴⁷ and the characteristics of the study sample differed from AHA hospitals, a nationally representative sample. For example, our study included a higher percentage of Magnet hospitals, which may have higher quality fall prevention implementation than non-Magnet hospitals. Second, hospitals self-selected to participate in the survey; therefore, hospitals participating in the survey may be different than hospitals that did not participate. Third, the comprehensiveness of data collection might vary across hospitals given that one individual at the hospital was tasked with collecting data from multiple sources. To minimise this risk, we relied on the NDNQI site coordinator in each hospital to assist with data collection. Site coordinators frequently participate in data collection requests and have experience gathering data from multiple sources within the hospital. It is also likely that fall implementation strategies vary across units within a given hospital. This study collected data at the hospital level and did not allow for comparisons of fall prevention implementation across units within the same hospital. There are also some recommended implementation strategies, such as involving staff in the planning of fall prevention interventions, which were not addressed in our survey.¹² Furthermore, information about the quality of implementation was not captured through the survey (eg, whether educational interventions were passive vs active). Our survey also did not capture risk factors for falls, such as environmental factors or hospital practices regarding bed moves.^48–50 It is also unclear whether the low quality of evidence for many fall prevention interventions^11–14 partially explains hospital variation in fall prevention strategies. The study authors tried to address this by gathering information on implementation strategies that are endorsed by national quality organisations.^1,4,8,12

CONCLUSION

Our study is the first to our knowledge to examine which expert-recommended hospital fall prevention implementation strategies^1,4,8,12 are being used and how they are being operationalised in the USA. Future studies are needed to document fall prevention implementation strategies in detail and to test which implementation strategies are most effective to guide intervention development.