Falls on inpatient psychiatric units are a major safety concern. The fall rates for inpatient psychiatric units tend to be at least three to four times higher than for the general hospital. Estimated rates for inpatient psychiatric units are 13 to 25 falls per 1,000 inpatient days compared with 3 to 4 falls per 1,000 inpatient days for the general hospital (Blair & Gruman, 2005). Currently, no validated instrument for fall risk has been designed for the inpatient psychiatric patient population, making identification of those at risk for falls extremely challenging.
For these reasons, a fall risk assessment targeting inpatient psychiatric patients is needed. The purpose of this study was to develop and conduct initial validity testing of a fall risk assessment tool in the acute inpatient psychiatric population.
Unique Risk Factors for Psychiatric Patients
Age. Psychiatric patients who fall are younger than medical-surgical patients who fall. Tay et al. (2000) found that the mean age of psychiatric patients who fell was 56.3 (SD = 16.8 years). Researchers studying medical-surgical patients reported mean ages from 65 to 83, with those older than 80 at greatest risk (Oliver, Daly, Martin, & McMurdo, 2004).
Ambulatory Status. Several investigators have identified the ambulatory nature of psychiatric units as a major contributing factor to falls. In contrast to medical-surgical patients who spend much of their day in bed, psychiatric patients are rarely in bed during the day. They are encouraged to walk and participate in activities. Poster, Pelletier, and Kay (1991) reported that 59% of falls in the inpatient psychiatric setting occurred during occupational and recreational therapy. Similarly, Vaughn, Young, Rice, and Stoner (1993) found that 42% of falls occurred in community areas. Furthermore, 68% of the patients who fell did not need assistance with ambulation and had been promoted to ambulatory treatment programming. In both studies, unique risk factors for psychiatric patients include age younger than 65, ambulatory status, experience of anxiety or agitation, and side effects of medications.
Medications. Side effects of various medications affecting a patient during ambulation compound the risk of falling. Several researchers have reported that patients had taken a sedative, antidepressant, or antipsychotic agent within 24 hours prior to their fall. These classes of medications have side effects, including orthostatic hypotension, dizziness, and/or decreased alertness (Tsai, Witte, Radunzel, & Keller, 1998). In a study conducted on a psychiatric unit, 88% of patients who fell demonstrated orthostatic hypotension (Tay et al., 2000).
Psychiatric Illnesses. Illnesses experienced by psychiatric patients may impair their ability to recognize and interpret environmental hazards. Depression, psychosis, and schizophrenia were the most common diagnoses of patients who fell (Tsai et al., 1998). Labile mental status and altered judgment are also risk factors for falls. In another study, more than 80% of patients who fell had symptoms of anxiety and agitation (Vaughn et al., 1993).
Undernutrition and Fluid Intake. Loss of appetite, undernutrition, and poor eating habits are very common in the psychiatric population. Undernutrition, broadly defined as a lack of proper food and fluid intake, is a risk factor for falls. In psychiatric patients, the diagnoses of depression and eating disorders are associated with low food and fluid intake. Fluid and electrolyte imbalances, bradycardia, anemia, and skeletal muscle atrophy in those with eating disorders contribute to falls. Howard, Kirkwood, and Leese (2007) studied patients with schizophrenia who fell and fractured their hips, and attributed the problem to poor nutrition, as well as antipsychotic medications.
Inadequate fluid intake usually accompanies poor food intake. Forsyth et al. (2008) monitored the fluid intake of psychiatric patients admitted to an inpatient unit within a tertiary care medical center. The most common diagnoses were major depression, primary psychotic disorders with depression, and depression related to dementia. Approximately one quarter (25.3%) met the criteria for dehydration. Poor fluid intake is associated with general weakness, which has been linked to falls (Rubenstein, 2006).
Sleep Loss. Sleep loss can impair awareness of the environment, which can contribute to accidental slips and trips. Sleep deprivation is extremely common in psychiatric patients, particularly those with schizophrenia and depression. Sleep disorders are an intrinsic feature of schizophrenia and are consistently present in those with this illness (Chouinard, Poulin, Stip, & Godbout, 2004). In a study of 284 patients in a general hospital setting who had fallen, researchers found that difficulty falling asleep at night, waking often during the night, and waking early in the morning were significantly related to falls (Brassington, King, & Bliwise, 2000).
Hospital Fall Risk Assessment Instruments
Several systematic reviews of fall risk instruments have indicated a lack of validation in psychiatric settings. Oliver et al. (2004) conducted a systematic review of 13 risk assessment tools for use in hospitals that were published from 1966 to 2002. Although patients from many kinds of units were included in these studies, no studies included patients from acute inpatient psychiatry units. A similar set of risk factors emerged as significant: mobility problems, urinary incontinency/frequency or need for assistance with toileting, previous fall history, secondary or specific diagnoses, and prescription of sedative hypnotic agents. The authors concluded that the instruments should be validated in various settings before adoption.
Many hospitals use a single risk assessment instrument throughout the hospital, including psychiatric units. By decreasing variance across units, this policy enables evaluation of hospital fall risk reduction programs (Rutledge et al., 2003). Four commonly used instruments in hospital settings are (Frank-Stromborg & Olsen, 2004; Myers, 2003):
- The Hendrich Risk Assessment Tool.
- The Schmid Fall Risk Assessment Tool.
- The St. Thomas Risk Assessment Tool in Falling Elderly Inpatients (STRATIFY).
- The Morse Fall Scale.
The Hendrich Risk Assessment Tool. The Hendrich Risk Assessment Tool (Hendrich, Nyhuis, Kippenbrock, & Soja, 1995) was originally developed using older adults hospitalized in medical-surgical units. Identified risk factors included (a) confusion/disorientation, (b) depression, (c) altered elimination, (d) recent history of falls, (e) poor mobility/weakness, (f) dizziness/vertigo, (g) primary cancer diagnosis, and (h) other. Psychiatric, obstetrical, rehabilitation, pediatric, and HIV patients were not included in the study (Hendrich et al., 1995). The authors found low construct validity and low internal consistency and did not recommend its use in these settings. No studies of the Hendrich Risk Assessment Tool were conducted on acute inpatient psychiatric units.
The Schmid Fall Risk Assessment Tool. The Schmid Fall Risk Assessment Tool (Schmid, 1990) was developed in a 700-bed, government-owned medical center. Characteristics of 102 patients who fell were matched with a group of 103 patients who did not fall. The tool assesses (a) mobility, (b) mentation, (c) elimination, (d) fall history, and (e) current medications. The original study did not describe the patients in terms of diagnosis or type of unit. The author stated that caution should be exercised in generalizing to the entire hospital population and to other environments (Schmid, 1990).
STRATIFY. The STRATIFY (Smith, Forster, & Young, 2006) was developed using patients from elderly care units at two hospitals. The population included acutely ill medical, stroke, and disabled patients. The instrument has five factors: (a) fall as a presenting complaint, (b) a combined Barthel Index transfer and mobility score of 3 or 4, (c) agitation, (d) frequent toileting, and (e) visual impairment. The STRATIFY was tested with acute stroke patients in a geriatric rehabilitation unit and was found to have a low sensitivity and specificity (Smith et al., 2006).
Morse Fall Scale. The Morse Fall Scale (Morse, Morse, & Tylko, 1989) was developed by comparing 100 patients who fell and 100 randomly selected patients who had not fallen. The instrument assesses (a) history of falling, (b) secondary diagnosis, (c) ambulatory aids, (d) use of intravenous therapy, (e) gait, and (f) mental status. Morse et al. tested the instrument with 2,689 patients in acute care hospital (general surgical, ophthalmology, and medical units), long-term care (psychogeriatric and nursing home settings), and rehabilitation hospital settings (neuromuscular, orthopedic, diabetes, weight control, head injury, and cerebrovascular accident patients). The sample was primarily derived from units occupied by older adults and did not include acute inpatient psychiatry patients.
To summarize the literature, the four general instruments commonly used across hospital settings are similar in that all include risk assessment for mobility, mentation/confusion, and fall history. Three instruments also address elimination. Only one includes medications as a risk factor. None include the unique risk factors for psychiatric patients identified in the literature. The current instruments focus primarily on older patients; however, younger psychiatric patients may also be at risk for falls.
Studies of inpatient psychiatric patients have indicated that the ambulatory setting increases the risk for falls, in conjunction with medications and their side effects. Patients are encouraged to be mobile and to participate in milieu therapy with other psychiatric patients. Additional factors unique to psychiatric patients are anxiety, restlessness and agitation; diagnoses of depression or psychosis; and labile mental status affecting judgment and interpretation of environmental hazards.
Evidence indicates that both undernutrition and sleep deprivation—additional risk factors for falls—are common in the psychiatric population. These risk factors are not adequately captured in any of the four instruments commonly used in the hospital setting. Consequently, we began development of an instrument that assesses fall risk specifically in acutely ill psychiatric patients. Since the Morse Fall Scale was already in use on our unit, we also determined its sensitivity and specificity with our psychiatric population.
Content validity was determined through extensive review of the literature and review of the instrument by an expert panel. The initial domains to determine fall risk were age, mental status, elimination, medications, diagnoses, ambulation/balance, history of falls, nutrition, sleep, and restraint. The uniqueness of the psychiatric population related to the ambulatory nature of the psychiatric milieu was incorporated into the items within these domains.
Although restraint use is a risk factor for falling, patients are not restrained on most psychiatric units, and nurses provide one-to-one supervision for patients who are restrained. Therefore, restraint use was deleted, resulting in an instrument with nine domains to calculate the risk for falls.
Five psychiatric nurse experts reviewed and revised the instrument. Items are designed to capture the increased risk of orthostatic hypotension and the frequent adjustment of psychiatric patients’ medications. For example, one item is “Increase in these medications…in the past 24 hours.” Items also reflect the active/ambulatory nature of inpatient psychiatry patients; for example, “Incontinent but ambulates independently.”
We then applied the instrument retrospectively to the records of 50 patients who had fallen and examined documentation during the 24 hours prior to the fall. Of the 50 patients who fell, 30 were women and 20 were men, and the mean age was 64.6 (age range = 29 to 97). The percentage of those who fell who triggered each of the items in each domain was calculated. For example, for the risk factor of age, 30% of those who fell were younger than 50.
Data were collected retrospectively from the records of all psychiatric in-patients to obtain a sample from 298 patient days. The expected value for each of the items in each domain was defined as the percentage of the inpatient psychiatric population that exhibited the property described in the item. For example, 56% of the total inpatient population was younger than 50. Consequently, the expected value for the item “Age: less than 50” was 56%.
Multidimensional scaling was used to determine the weightings of each item. When the observed value was greater than the expected value (indicating an increased risk of falling), the ratio between the observed value and the expected value was computed and the results rounded to the nearest whole number. For example, for patients 80 and older, the ratio between the observed value of 16% and the expected value of 1% was 16. Consequently, the unadjusted weighting was 16.
However, when the observed value was less than the expected value (indicating a reduced risk of falling), the ratio was inverted and the result multiplied by −1. For example, 30% of patients who fell were younger than 50, whereas 56% of the population was younger than 50. The inverted ratio was 1.87, which was multiplied by −1 and then rounded to the nearest whole number to produce an unadjusted weighting of −2. As a result of this procedure, positive scores indicated an increased risk for falls, and negative scores indicated a reduced risk for falls.
The EPFRAT was then examined to determine nurses’ perceptions of usability. Twenty-two RNs, 3 licensed practical nurses, and 10 nursing students working on the adult psychiatric inpatient units used the instrument and reported on its ease of use. Because these users reported that working with both positive and negative numbers was needlessly complex and prone to error, the ratios were all converted to positive numbers by adding 10 to every weight. For example, the unadjusted weight for “Age: less than 50” was −2; the adjusted weight became 8 (−2 + 10). Similarly, the adjusted weight for the item “Age: 80 and over” was 16; the adjusted weight became 26 (16 + 10). Using these transformed weights, a total score greater than 90 indicated an increased risk for falls. The weighting development is shown in the Table, and the instrument is displayed in the Figure.
Table: Development of Weighting of Instrument
Figure. Edmonson Psychiatric Fall Risk Assessment Tool. ©2009, Deborah Edmonson.
We educated all staff, including nurses, nursing assistants, technicians, and clerks, on all shifts about the EPFRAT. We used a combination of poster presentations, education in small groups during shift report, and individual instruction. The EPFRAT was then implemented on the unit as a quality improvement measure. At the time of the study, all units of the hospital including the psychiatry unit used the Morse Fall Scale. Thus, data were available from the Morse Fall Scale for comparison. For the next month, both the EPFRAT and the Morse Fall Scale were used. A total of 138 patient records were examined retrospectively to determine whether the patient fell and whether the Morse Fall Scale and the EPFRAT predicted the fall. Positive predictive values were calculated for both tools.
During the data collection period, 43 patients fell. The Morse Fall Scale identified 21 of those at risk of falling; the EPFRAT identified 27. Sensitivity of the EPFRAT was 0.63, compared with 0.49 for the Morse Fall Scale, indicating that the Morse Fall Scale missed more patients who were actually at risk for falls. Specificity of the EPFRAT was 0.86, compared with 0.85 for the Morse Fall Scale, indicating that neither tool overpredicted falls. The positive predictive value of the EPFRAT was 0.68, and the positive predictive value of the Morse Fall Scale was 0.60, indicating the EPFRAT has a higher probability of correctly identifying fall risk patients in this population.
This article describes development and initial testing of the EPFRAT. Similar to current hospital fall risk instruments, the EPFRAT includes domains for history of falls (past 3 months), use of ambulatory aids and problems with gait and transferring, and mental status. However, different from medical-surgical instruments, the weighting process revealed some important criteria that are more predictive of falls in psychiatric patients.
Experiencing an increase in medication in the past 24 hours proved to be a heavily weighted criterion. Physicians frequently adjust medications to relieve symptoms. Some patients are admitted because of psychiatric symptoms caused by medications. Others are admitted for observation during medication adjustments. Still others may need to have an initial increase or change in medications to help them through the acute phase of their illnesses. A common side effect of many of these medications is orthostatic hypotension. The milieu of the acute psychiatric unit encourages all patients to be as mobile as possible; thus, they are more prone to fall as a result of a medication change. It is crucial that nurses remain vigilant for side effects of medications, especially in patients who have had a recent increase in medication.
Sleep disturbance was another heavily weighted criterion. Sleep disturbance is defined in the EPFRAT as report by the patient, family, or caregiver of 4 hours or less of consecutive sleep. Patients admitted to psychiatric inpatient units often have altered sleep and insomnia. In this population, difficulty sleeping can be a result of many factors including psychiatric illness, metabolic illnesses secondary to mental illness, substance abuse, and social situations such as homelessness.
In the weighting of items, little food or fluid intake resulted in a fairly high rating. Nutritional status is defined in the EPFRAT as the patient’s or caregiver’s report of the patient not eating/drinking, and/or documentation of less than 50% consumption, and/or nurse’s documentation of poor intake for more than one meal per day. Alterations in food/fluid intake can occur due to symptoms related to depression, mania, psychosis, dementia, and delirium. In our sample, 24 patients who fell were reported to have very little or no food/fluids in the past 24 hours.