Toileting-related falls are estimated to account for approximately 50% of inpatient falls; however, in Pennsylvania, in 2012, only 12.8% of reported falls were toileting-related. Falls event reporting through the Pennsylvania Patient Safety Reporting System (PA-PSRS) requires entry of a single falls event category selected from a list of 13 choices. Pennsylvania Patient Safety Authority analysts noted variations in event type reporting: 22.8% of found-on-floor falls event reports and 11.3% of other/unknown (i.e., unspecified) category event reports were determined to be toileting-related falls. To standardize and improve the reliability and validity of reporting the type of patient fall, a PA-PSRS falls event type decision tree was developed and released in 2012 in collaboration with the PA-PSRS falls reporting program. A two-year analysis of hospital Serious Event falls reports submitted by hospitals enrolled in the PA-PSRS falls reporting program before and after release of the decision tree revealed a 5.7% increase in reports of toileting-related falls. This increase may signal that the specificity of falls event reporting can be improved through the use of the decision tree.
among the most
frequently reported events to the Pennsylvania Patient Safety Reporting System (PA-PSRS).1 In 2012, the Pennsylvania Patient Safety Authority developed and released a new falls reporting program through PA-PSRS to address the need to provide falls rate benchmarking data and process measure reports to hospitals. A key element of this program is the standardization of the definitions of falls and falls with harm and several other data fields, including new data fields (e.g., inpatient status) and existing data fields (e.g., falls event type), to ensure all participating hospitals identify, measure, and report falls in the same manner.
PA-PSRS falls event types were identified with the establishment of PA-PSRS, reflecting event types that range from activity-based descriptions to stationary or location-based descriptions.2 There are 13 different falls event type choices in PA-PSRS, of which healthcare staff can choose only one event type to describe a fall when reporting; see " PA-PSRS Falls Event Types and Definitions" for more information.
Why Focus on Falls Event Types?
In 2012, a question was raised by a Pennsylvania hospital about how to select a PA-PSRS falls event type when there was more than one appropriate option. In the case described, a patient was utilizing a walker and ambulating to the bathroom with a staff member when the patient began to lose balance and was assisted to the floor—should this be classified as an assisted falls or a toileting-related fall?2
Reporting variations can be caused by a lack of a standardized approach among people filling out the falls event reports; incomplete information when reporting; fear of blame, repercussions, or punitive actions; limitations in the transfer of pertinent information from the primary reporter to the designated reporter; insufficient staff education on appropriate methods of filling out the event reports;3 or constraints imposed by the design of the reporting system.
Why Toileting-Related Falls?
Toileting-related falls are estimated to account for approximately 50% of all falls.4-8 Yet in Pennsylvania, in 2012, 12.8% of reported falls were submitted as toileting-related. This low percentage of toileting-related falls led to an examination of report narratives, which revealed a subset of PA-PSRS found-on-floor and other/unknown falls event reports that described falls related to toileting activities.
An analysis of one year of baseline PA-PSRS data (January through December 2012) revealed that 12.8% of falls (n = 4,528 of 35,358) were attributed to toileting-related activities, 33.0% of falls (n = 11,682) were attributed to the found-on-floor falls event category, and 7.7% of falls (n = 2,716) were attributed to the other/unknown falls event category. The identification of toileting-related falls embedded in the found-on-floor and other/unknown event type categories demonstrates variations in reporting.
Following are a few examples of toileting-related narratives reported through PA-PSRS as found-on-floor falls:
The patient needed to go to the bathroom but did not call so as to not bother [the staff]. [The patient] took the water cup and cane into the bathroom, voided in the water cup, and poured the specimen into the toilet to maintain accurate output. Somehow when he was dumping the cup or trying to void, he fell to the floor, hitting his left forehead and left elbow. The patient pulled the call light cord in the bathroom, and no one knew he was in the bathroom.
Patient found on floor in the bathroom. Patient stated, “I was in a hurry to have a BM [bowel movement] and tripped going into the bathroom.”
The patient had been seen by physical therapy. After the session was completed, the patient was placed in a chair. The patient attempted to go to the bathroom independently, became incontinent of urine, slipped, and fell on their buttocks. The chair alarm was not in place.
Toileting-related falls are not the only event type categorized as found-on-floor or other/unknown falls events; medical conditions, ambulating-related falls, and assisted falls are additional examples of event types that have been categorized as found-on-floor or other/unknown falls events.
Examples of other/unknown PA-PSRS falls event narratives that Authority analysts deemed non-toileting-related are as follows:
The nursing assistant found the patient supine on the floor. The patient’s blood sugar was 57. The patient was assisted into bed and denies pain. The patient stated, “I don’t remember what happened.”
An inpatient who was admitted post-op after [major surgery] cried out for help after falling onto his buttocks from the side of the bed when reaching for his eyeglasses.
Defining Toileting-Related Falls
In her research, Tzeng (2010) conceptualized toileting to include both the process of going to the toilet and the intention of going to the toilet. In this study, toileting referred to all activities that were intended to relieve elimination needs, including getting out of bed, moving from the bed to the bathroom, entering the bathroom, using the toilet, stand-to-sit and sit-to-stand movements, moving from the bathroom to the bed, and getting back to bed. Tzeng concluded that “this study should be repeated to investigate any differences in the prevalence of toileting-related falls incidents across different types of hospitals and inpatient care units. This effort would allow nursing staff and managers to allocate scarce healthcare resources to various falls prevention initiatives in a more efficient manner by, for example, prioritizing falls prevention strategies on the basis of the most prevalent toileting-related falls that occur in their work settings.”8
Variations in the reporting of PA-PSRS falls event types and the request for how to select falls event types led to the development of a decision tree.
Utility of Falls Event Data
A falls dashboard (see below) provided by the new PA-PSRS falls reporting program contains two tables that identify the top three falls event types individualized for each reporting facility, combined with
(1) patient characteristics present at the time the patient fell (e.g., altered elimination needs) and (2) falls prevention strategies that were in place at the time the patient fell. If toileting-related falls are one of the top three event types reported through PA-PSRS by a specific facility, the falls event table will display the number of patients with a toileting-related fall who also had altered elimination needs, an altered mental status, or other relevant patient characteristics.
Development of the Falls Event Decision Tree
The Authority's falls event decision tree provides a systematic approach to evaluate the circumstances surrounding falls and to standardize falls reporting. The decision tree contains a series of questions in a yes/no format that guides staff in identifying the falls event type that best reflects the patient’s circumstances.
The reporter starts by dividing patients into categories based on whether a patient is conscious or there is a reliable witness to the fall. The next two questions exclude falls that were intentional or the result of a medical condition (e.g., a seizure). If the answer to both questions is no, then these falls are considered unplanned, unanticipated falls. The reporter next evaluates the patient’s mobility (i.e., whether patients were stationary or ambulatory) when the fall occurred. The answer to the mobility question separates the choice of possible event types in half. The found-on-floor and other/unknown falls event types are category selections for use only when the more descriptive options are not appropriate.
The decision tree was pilot-tested by 13 hospitals participating in the falls program; 76.9% of responding hospitals (n = 10 of 13) found it useful. One year after the release of the PA-PSRS falls decision tree, the Authority’s 2013 annual survey of hospitals asked respondents enrolled in the falls reporting program if they used the falls event decision tree and whether it was useful. Seventy-one percent of responding hospitals (n = 27 of 38) indicated that they found the falls event type decision tree very useful, moderately useful, or somewhat useful.1 In light of positive results from both the pilot and the survey results, an investigation was performed to determine if there were changes in the assignment of falls event types among hospitals participating in the falls reporting program. See "PA-PSRS Falls Reporting Program Enrollment" for information on enrolling in the program.
Analysts queried the PA-PSRS database for hospital-based falls Incidents and Serious Events reported from January 1, 2012, through December 31, 2013. Analysts excluded hospitals that did not report falls events during both calendar years, because the absence precluded a comparison of falls event type reporting before and after release of the decision tree. Also excluded were hospital falls events from outpatient surgical centers. Hospital characteristics (i.e., program participation and PA-PSRS peer-group designation based on bed size and specialty status) were assigned to each falls event report.
Analysts categorized the falls events into two groups: (1) hospitals enrolled in the PA-PSRS falls reporting program and (2) hospitals not enrolled in the program. Two sets of analyses were performed to see if there were changes in how falls event types were reported. For the first analysis, chi-square and Cramer’s V statistical tests were conducted to determine any statistically significant difference in event type reporting and the effect size of any changes in the selection of event types related to enrollment in the falls program.
Given that (1) the decision tree had only been available to hospitals since December 2012, (2) hospitals enrolled in the falls program with access to the falls dashboard were more likely to use the decision tree, and (3) new practices take time to spread,9 minimal changes in reporting event types were expected when evaluating the Incident and Serious Event data for 2012 and 2013. This led to the decision to focus the second analysis only on Serious Event reports from hospitals enrolled in the program in 2012 (i.e., before decision tree release) and 2013 (i.e., after decision tree release). Chi-square and Cramer’s V statistical tests were conducted to determine any effect size with the release of the decision tree. A comparison of the percentage change in event reporting between hospitals based on program participation for the two-year time period was also performed.
A third analysis evaluated the found-on-floor and other/unknown event report narratives with IBM SPSS Modeler data mining and text analytics software to identify toileting-related falls reported within these event categories. The SPSS Modeler extracts and groups reports by terms, acronyms, or combinations of terms identified through manual, nonexhaustive analysis of narratives. The following are a sample of terms used for this analysis: “bathroom,” “bathroom assistance,” “bedpan,” “bedside commode,” “bowel movement,” “commode,” “defecate,” “diarrhea,” “elimination,” “feces,” “frequent loose stool,” “incontinence,” “toilet,” “toileting,” “underwear,” “urinal,” “urinate,” and “void.” In cases where a term identified more than 100 reports, an individual review of the first 100 narratives was performed. If 80% or more of these narratives within the sample indicated a toileting-related fall, all of the reports in that term were counted as toileting-related falls.
Program Participation and Event Reporting
Almost half of Pennsylvania hospitals are enrolled in the falls program (44.4%, n = 103 of 232). Figure 1 shows falls program participation by peer group designation. Hospitals reported 68,804 falls events; 51.9% (n = 35,731) were reported by participants in the falls program and 48.1% (n = 33,073) by nonparticipants.
Figure 1. Number of Hospitals by Falls Reporting Program Participants (N = 232)
Note: Peer groups with less than five hospitals were grouped together.
A chi-square test was used to compare the number of falls reports by event type for hospitals enrolled in the falls program to hospitals not enrolled in the falls program. The chi-square test result was 164.6 (p value < 0.001), and the Cramer’s V test result was 0.0489.
Changes in Event Reports Submitted by Hospitals Enrolled in the Falls Reporting Program
A comparison of Serious Event reports between 2012 and 2013 noted a decrease in falls reports in 2013, from 603 to 525. Increases were noted in the change in proportion of falls reported as toileting-related (5.7% increase) and sitting at the side of the bed (2.6% increase). Decreases were noted in the proportion of reports of found-on-floor (2.9% decrease), assisted (2.2% decrease), and other/unknown (2.0% decrease) falls. The chi-square test result was 24.5 (p value = 0.017), and the Cramer’s V test result was 0.1475. Table 1 provides a classification of the actual number of falls per category and the percentage change in proportion. Figure 2 shows the percentage change in proportion of falls with harm by falls program participation.
No. of Falls-Related Serious Events ||
% Change in Proportion of Falls Event Types|
2012 (N = 603)||
2013 (N = 525)|||
|Sitting at the side of the bed||7||20||2.6|
|From the stretcher||6||8||0.5|
|In the exam room/from the exam table||7||8||0.4|
|Hallways of the facility||7||8||0.4|
|Sitting in the chair/wheelchair||38||34||0.2|
|Lying in the bed||29||25||0.0|
|Grounds of the facility||10||8||-0.1|
|Found on floor||209||167||-2.9|
Figure 2. 2012 to 2013 Change in Proportion of Falls with Harm by PA-PSRS Falls Event Type*
* Pennsylvania Patient Safety Reporting System (PA-PSRS) falls event types with fewer than 10 event reports for at least one time period were not included.
Event Narrative Analysis
The SPSS modeler identified up to 90 different toileting-related terms among the found-on-floor and other/unknown falls event reports for both years. Approximately one-quarter of the found-on-floor falls were toileting-related, and slightly more than 15% of the other/unknown event reports were toileting-related.
The actual number of toileting-related falls from these two event types were moved (reassigned) from the found-on-floor and other/unknown event categories and placed into the toileting-related event category. Moving these event reports shows how toileting-related circumstances contribute to a greater proportion of falls types. A similar shift was noted in falls reported by hospitals not enrolled in the falls program. Table 2 provides the number and percentage of falls as reported in 2013 and with reassignment.
Falls Program Participants ||
Falls Program Non-Participants |
No. (%) of Falls
|Reassignment of Falls,
No. (%) of Falls
No. (%) of Falls
|Reassignment of Falls,|
No. (%) of falLS
|Toileting||2,492 (14.4)||3,980 (23.0)||1,921 (11.6)||3,385 (20.5)|
|Found on floor||5,433 (31.4)||4,068 (23.5)||5,499 (33.3)||4,251 (25.7)|
|Other||1,387 (8.0)||1,264 (7.3)||1,425 (8.6)||1,209 (7.3)|
Total number of falls||
Note: Only 3 of the 13 event categories are represented here. The number of toileting-related falls that were identified in the found-on-floor and other/unknown category were subtracted from their respective category and added to the toileting-related falls totals. The percentages of falls by event type were then calculated based on the reassignment. Percentages reflect the proportion of the total number of falls for 2013. |
The Authority facilitated the development of the PA-PSRS falls reporting program to standardize falls reporting. The falls reporting program includes the aforementioned decision tree, which is designed as a systematic approach to standardize the reporting of a fall event according to the associated circumstances. Based on the results from the pilot testing and the Authority’s 2013 annual survey, it was anticipated that hospitals enrolled in the falls program would report an increased proportion of falls related to toileting. The first analysis compared hospital event reports based on program enrollment. The positive result of the chi-square test is likely due to the large number of falls (n = 68,804); the Cramer’s V test demonstrated a very minimal effect in the assignment of falls event types due to program enrollment.
The second analysis, a chi-square test evaluating the change in proportion of reported Serious Event falls by facilities
enrolled in the falls program before and after the decision tree was released (i.e., 2012 versus 2013) was statistically significant. The Cramer’s V test was indicative of a small effect of the decision tree on event reporting. In this subset (i.e., Serious Event reports from enrolled programs), toileting-related falls reports increased by 5.7% (from 13.9% in 2012 to 19.6% in 2013) and sitting at side of bed falls event increased 2.6% (from 1.2% in 2012 to 3.8% in 2013). While no definitive conclusions can be drawn about the increase in the reporting of toileting-related falls, the positive survey results (from the Authority’s annual survey) about the usefulness of the decision tree may help explain the changes in event reporting.1
An in-depth analysis of the other falls event types is beyond the scope of this article; however, a cursory review of falls identified as assisted falls and ambulating falls uncovered the presence of toileting-related falls being reported as these event types. Assisted falls are identified in the decision tree as occurring when a patient is in physical therapy rather than when the patient is being assisted to the bathroom for elimination purposes. Some falls reported as ambulating falls involved patients who were ambulating to the bathroom when the fall occurred. Even more toileting-related falls are likely to be identified with the use of the decision tree.
The analysis of the event narratives exemplifies the challenge in assigning a falls event type to a single category that best reflects the circumstances surrounding a fall. The reassignment of falls event types highlights the issue that toileting-related falls may occur more often than the current PA-PSRS data shows. Standardizing and clarifying the assignment of falls event types provides a clearer picture of what is actually happening to patients who fall in an inpatient setting, improves the specificity and validity of the data, and better informs falls prevention programs. Table 3 provides an example of how a found-on-floor fall would be reported using the decision tree.
|Example of Current Event Reporting||
Application of the Falls Event Type Decision Tree|
When submitting a patient fall event report through the Pennsylvania Patient Safety Reporting System, an event type is selected based on how the situation was perceived by the reporter. The event type selected is "Found on floor." The event narrative is as follows:
The patient was found on the floor in the bathroom. The patient states that she was going to the bathroom and got urine on the floor and slipped on it. The patient stated she did not fall, she just sat down. The patient stated that she did not hit her head and no other part of her body was injured, except for a skin tear on her left leg. The patient had been walking to the bathroom all day with no problems.
The decision tree has a series of yes/no questions that assist the reporter in identifying a falls even type that best describes the underlying circumstances.Using the example narrative, here are the series of questions and the results using the decision tree.
Was the patient awake? Yes
Did the patient have an unplanned (unintentional) fall? Yes
Did the patient have a seizure or other contributory medical conditionat the time of the fall? No
Was the patient standing or moving when the fall occurred? Yes
Was the patient standing or ambulating for purposes of elimination? Yes
Select event type #7: Toileting fall.
Improving Falls Reporting
Standardizing the assignment of falls event types using a systematic approach of yes/no questions (i.e., the falls decision tree) increases the consistency of the data reported back to the falls team and provides falls team members with more reliable information about the circumstances and patient’s activities at the time of the fall. The following are ways to engage staff in standardizing reporting:
- Ensure leadership recognizes the value of and supports standardization of adverse event reporting.
- Educate staff about the PA-PSRS falls event type decision tree (what it is and how to apply it in everyday use), and provide the decision tree in convenient locations and formats (e.g., printouts at the point of care, electronic links to the decision tree).
- Show staff the PA-PSRS falls dashboard, which is where the falls event type data is reported.
- Teach staff how applying the decision tree can change the data in the falls dashboard, how to read the tables within the dashboard, and how the information can be used.
- Provide simulations or vignettes for staff to help them learn how to use the decision tree.
- Ensure ongoing educational programs that focus on reporting involve all healthcare staff, including experienced staff and new staff.
The use of two years of falls data provided a sizeable database to test the differences between hospitals enrolled versus hospitals not enrolled in the falls program; however, when using such a large database, a statistically significant test result is almost certain. Not knowing which of the hospitals enrolled in the falls program actually used the decision tree prevented a comparison of falls event type reporting based on implementation of the tool. Conversely, personnel in hospitals that were not enrolled in the falls program also had access to the decision tree through PA-PSRS and may have used it when submitting their falls reports; however, not having access to the falls dashboard, this group may have had less incentive to use the decision tree. In addition, the analysts were unaware of any data mining and text analytic software features that would have allowed assignment of other event types within the found-on-floor reports to mutually exclusive categories; hence, this analysis focused solely on toileting-related event types.
Standardized reporting is not a panacea for preventing falls; it is a method to obtain specific data about inpatient falls that teams can use to tailor their falls programs to meet the needs of their patients. A falls event type decision tree was developed to address variability and questions related to categorizing and reporting falls event types. Analysis of reports submitted in the year following release of the decision tree demonstrated a shift in categorization patterns, with a small effect size likely attributable to the decision tree. Education of healthcare staff on the use of the decision tree provides staff with the ability to improve the specificity and validity of reporting falls event types. Continued monitoring of falls event reporting is required to determine if the classification of falls changes over time, resulting in a more precise picture of the circumstances affecting inpatient falls in Pennsylvania.
Pennsylvania Patient Safety Authority. Pennsylvania Patient Safety Authority 2013 annual report [online]. 2014 Apr 30 [cited 2014 Oct 28].
Pennsylvania Patient Safety Authority. Falls event type decision tree for hospital users [memorandum]. Program memorandum no. 2012-06. 2012 Dec 20.
Mason KP, Green SM, Piacevoli Q. Adverse event reporting tool to standardize the reporting and tracking of adverse events during procedural sedation: a consensus document from the World SIVA International Sedation Task Force.
Br J Anaesth 2012 Jan;108(1):13-20.
Perell KL, Nelson A, Goldman RL, et al. Fall risk assessment measures: an analytic review.
J Gerontol A Biol Sci Med Sci 2001 Dec;56(12):M761-6.
Hendrich A, Nyhuis A, Kippenbrock T, et al. Hospital falls: development of a predictive model for clinical practice.
Appl Nurs Res 1995 Aug;8(3):129-39.
Evans D, Hodgkinson B, Lambert L, et al. Falls risk factors in the hospital setting: a systematic review.
Int J Nurs Pract 2001 Feb;7(1):38-45.
Hitcho EB, Krauss MJ, Birge S, et al. Characteristics and circumstances of falls in a hospital setting: a prospective analysis.
J Gen Intern Med 2004 Jul;19(7):732-9.
Tzeng HM. Understanding the prevalence of inpatient falls associated with toileting in adult acute care settings.
J Nurs Care Qual 2010 Jan-Mar;25(1):22-30.
Diffusion of innovations. 4th ed. New York: the Free Press; 1995: 11.
Falls Event Types and Definitions
The following are the Pennsylvania Patient Safety Reporting System (PA-PSRS) falls event types and their definitions:
Ambulating: Patient falls while walking anywhere within the facility (not associated with toileting activities).
Assisted: A caregiver sees a patient about to fall and intervenes, lowering them to a bed or floor. This includes therapeutic falls.
Found on floor: Patients are found on the floor and the reason for the patient being out of the bed cannot be identified.
From stretcher: Patient falls off a stretcher, even if the patient is being transferred.
Grounds of facility: Patient falls in a non-care-area unit, such as the cafeteria or admissions office.
Hallways of facility: Patient falls in a unit hallway or elevator. This event type can be used for environmental falls, such as when a fall occurs due to the floor being wet.
In the exam room/from the exam table: Patient falls while receiving services that require the patient to use an exam table or room, such as a radiology room, operating room, or a physician’s office.
Lying in bed: Patient falls out of bed. For example, this situation can occur when one or more siderails are not up or when a patient is reaching for an object while in bed.
Other/unknown (e.g., unspecified, intentional): The fall is not clearly identifiable from the other 12 choices or is an unanticipated physiologic fall (e.g., seizure or other medical condition at the time of the fall).
Sitting at side of bed: Patient falls after sitting at the side of the bed when getting ready to stand, eat a meal, move to a chair, or stand to determine stamina.
Sitting in chair/wheelchair: Patient falls out of a chair or wheelchair while sitting, getting ready to stand, moving back to bed, or lowering into a chair or wheelchair.
Toileting: Patient falls while leaving the bed and going to or coming from the bathroom or when standing up to use the bedside commode, bathroom commode, or urinal. This event type includes patients who are found on the floor while in the process of going to or returning from the bathroom or bedside commode for elimination purposes.
Transferring: Patient falls while transferring from stretcher to bed or wheelchair to bed or while being transferred off the unit to a different unit or department, such as the physical therapy department.
Source: Pennsylvania Patient Safety Authority. Falls event type decision tree for hospital users [memorandum]. Program memorandum no. 2012-06. 2012 Dec 20.
Falls Reporting Program Dashboard
The falls prevention strategies figure identifies the top three falls event types stratified by prevention strategies that were in place for patients who fell.
|Top 3 falls
|Hourly (or more
frequent) Comfort or
|Found on floor||8||1||2||7||4||1||6||0||0||12|
* Individual patients could have multiple prevention strategies in place.
** Patient total represents all patients for each event type who had a prevention strategy in place.
Table interpretation: There were 12 patients who fell who were found on the floor that had at least 1 prevention strategy in place. Of those 12 patients, 8 had patient identifiers in place, 1 had patient and family education, 2 had hourly rounding, 7 had a nurse call system, 4 had appropriate footwear, 1 had an alarm present, and 6 had some form of equipment in use (e.g., high/low bed, bedrails up). The total number of patients represents only those patients with a prevention strategy in place.
PA-PSRS Falls Reporting Program Enrollment
The Pennsylvania Patient Safety Reporting System (PA-PSRS) falls reporting program is a computer-based program for hospitals to receive falls analytic reports (e.g., falls dashboard, falls rates with benchmarking data) based on their own falls data submitted through PA-PSRS. The program requires hospitals to standardize the definition of falls and falls with harm and enter utilization data (e.g., patient-days) into PA-PSRS. Enrollment is available to Pennsylvania hospitals through PA-PSRS.